Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 1 | ---------------------- |
Willy Tarreau | 8317b28 | 2014-04-23 01:49:41 +0200 | [diff] [blame] | 2 | HAProxy |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 3 | Configuration Manual |
| 4 | ---------------------- |
Willy Tarreau | 991b478 | 2015-10-13 21:48:10 +0200 | [diff] [blame] | 5 | version 1.7 |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 6 | willy tarreau |
Willy Tarreau | 7d1b48f | 2016-05-10 15:36:58 +0200 | [diff] [blame] | 7 | 2016/05/10 |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 8 | |
| 9 | |
| 10 | This document covers the configuration language as implemented in the version |
| 11 | specified above. It does not provide any hint, example or advice. For such |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 12 | documentation, please refer to the Reference Manual or the Architecture Manual. |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 13 | The summary below is meant to help you search sections by name and navigate |
| 14 | through the document. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 15 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 16 | Note to documentation contributors : |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 17 | This document is formatted with 80 columns per line, with even number of |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 18 | spaces for indentation and without tabs. Please follow these rules strictly |
| 19 | so that it remains easily printable everywhere. If a line needs to be |
| 20 | printed verbatim and does not fit, please end each line with a backslash |
Willy Tarreau | 62a36c4 | 2010-08-17 15:53:10 +0200 | [diff] [blame] | 21 | ('\') and continue on next line, indented by two characters. It is also |
| 22 | sometimes useful to prefix all output lines (logs, console outs) with 3 |
| 23 | closing angle brackets ('>>>') in order to help get the difference between |
| 24 | inputs and outputs when it can become ambiguous. If you add sections, |
| 25 | please update the summary below for easier searching. |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 26 | |
| 27 | |
| 28 | Summary |
| 29 | ------- |
| 30 | |
| 31 | 1. Quick reminder about HTTP |
| 32 | 1.1. The HTTP transaction model |
| 33 | 1.2. HTTP request |
| 34 | 1.2.1. The Request line |
| 35 | 1.2.2. The request headers |
| 36 | 1.3. HTTP response |
| 37 | 1.3.1. The Response line |
| 38 | 1.3.2. The response headers |
| 39 | |
| 40 | 2. Configuring HAProxy |
| 41 | 2.1. Configuration file format |
William Lallemand | f9873ba | 2015-05-05 17:37:14 +0200 | [diff] [blame] | 42 | 2.2. Quoting and escaping |
William Lallemand | b2f0745 | 2015-05-12 14:27:13 +0200 | [diff] [blame] | 43 | 2.3. Environment variables |
| 44 | 2.4. Time format |
| 45 | 2.5. Examples |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 46 | |
| 47 | 3. Global parameters |
| 48 | 3.1. Process management and security |
| 49 | 3.2. Performance tuning |
| 50 | 3.3. Debugging |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 51 | 3.4. Userlists |
Cyril Bonté | dc4d903 | 2012-04-08 21:57:39 +0200 | [diff] [blame] | 52 | 3.5. Peers |
Cyril Bonté | 307ee1e | 2015-09-28 23:16:06 +0200 | [diff] [blame] | 53 | 3.6. Mailers |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 54 | |
| 55 | 4. Proxies |
| 56 | 4.1. Proxy keywords matrix |
| 57 | 4.2. Alphabetically sorted keywords reference |
| 58 | |
Willy Tarreau | 086fbf5 | 2012-09-24 20:34:51 +0200 | [diff] [blame] | 59 | 5. Bind and Server options |
| 60 | 5.1. Bind options |
| 61 | 5.2. Server and default-server options |
Baptiste Assmann | 1fa6666 | 2015-04-14 00:28:47 +0200 | [diff] [blame] | 62 | 5.3. Server DNS resolution |
| 63 | 5.3.1. Global overview |
| 64 | 5.3.2. The resolvers section |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 65 | |
| 66 | 6. HTTP header manipulation |
| 67 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 68 | 7. Using ACLs and fetching samples |
| 69 | 7.1. ACL basics |
| 70 | 7.1.1. Matching booleans |
| 71 | 7.1.2. Matching integers |
| 72 | 7.1.3. Matching strings |
| 73 | 7.1.4. Matching regular expressions (regexes) |
| 74 | 7.1.5. Matching arbitrary data blocks |
| 75 | 7.1.6. Matching IPv4 and IPv6 addresses |
| 76 | 7.2. Using ACLs to form conditions |
| 77 | 7.3. Fetching samples |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 78 | 7.3.1. Converters |
| 79 | 7.3.2. Fetching samples from internal states |
| 80 | 7.3.3. Fetching samples at Layer 4 |
| 81 | 7.3.4. Fetching samples at Layer 5 |
| 82 | 7.3.5. Fetching samples from buffer contents (Layer 6) |
| 83 | 7.3.6. Fetching HTTP samples (Layer 7) |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 84 | 7.4. Pre-defined ACLs |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 85 | |
| 86 | 8. Logging |
| 87 | 8.1. Log levels |
| 88 | 8.2. Log formats |
| 89 | 8.2.1. Default log format |
| 90 | 8.2.2. TCP log format |
| 91 | 8.2.3. HTTP log format |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 92 | 8.2.4. Custom log format |
Willy Tarreau | 5f51e1a | 2012-12-03 18:40:10 +0100 | [diff] [blame] | 93 | 8.2.5. Error log format |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 94 | 8.3. Advanced logging options |
| 95 | 8.3.1. Disabling logging of external tests |
| 96 | 8.3.2. Logging before waiting for the session to terminate |
| 97 | 8.3.3. Raising log level upon errors |
| 98 | 8.3.4. Disabling logging of successful connections |
| 99 | 8.4. Timing events |
| 100 | 8.5. Session state at disconnection |
| 101 | 8.6. Non-printable characters |
| 102 | 8.7. Capturing HTTP cookies |
| 103 | 8.8. Capturing HTTP headers |
| 104 | 8.9. Examples of logs |
| 105 | |
Christopher Faulet | c3fe533 | 2016-04-07 15:30:10 +0200 | [diff] [blame] | 106 | 9. Supported filters |
| 107 | 9.1. Trace |
| 108 | 9.2. HTTP compression |
| 109 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 110 | |
| 111 | 1. Quick reminder about HTTP |
| 112 | ---------------------------- |
| 113 | |
| 114 | When haproxy is running in HTTP mode, both the request and the response are |
| 115 | fully analyzed and indexed, thus it becomes possible to build matching criteria |
| 116 | on almost anything found in the contents. |
| 117 | |
| 118 | However, it is important to understand how HTTP requests and responses are |
| 119 | formed, and how HAProxy decomposes them. It will then become easier to write |
| 120 | correct rules and to debug existing configurations. |
| 121 | |
| 122 | |
| 123 | 1.1. The HTTP transaction model |
| 124 | ------------------------------- |
| 125 | |
| 126 | The HTTP protocol is transaction-driven. This means that each request will lead |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 127 | to one and only one response. Traditionally, a TCP connection is established |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 128 | from the client to the server, a request is sent by the client on the |
| 129 | connection, the server responds and the connection is closed. A new request |
| 130 | will involve a new connection : |
| 131 | |
| 132 | [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ... |
| 133 | |
| 134 | In this mode, called the "HTTP close" mode, there are as many connection |
| 135 | establishments as there are HTTP transactions. Since the connection is closed |
| 136 | by the server after the response, the client does not need to know the content |
| 137 | length. |
| 138 | |
| 139 | Due to the transactional nature of the protocol, it was possible to improve it |
| 140 | to avoid closing a connection between two subsequent transactions. In this mode |
| 141 | however, it is mandatory that the server indicates the content length for each |
| 142 | response so that the client does not wait indefinitely. For this, a special |
| 143 | header is used: "Content-length". This mode is called the "keep-alive" mode : |
| 144 | |
| 145 | [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ... |
| 146 | |
| 147 | Its advantages are a reduced latency between transactions, and less processing |
| 148 | power required on the server side. It is generally better than the close mode, |
| 149 | but not always because the clients often limit their concurrent connections to |
Patrick Mezard | 9ec2ec4 | 2010-06-12 17:02:45 +0200 | [diff] [blame] | 150 | a smaller value. |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 151 | |
| 152 | A last improvement in the communications is the pipelining mode. It still uses |
| 153 | keep-alive, but the client does not wait for the first response to send the |
| 154 | second request. This is useful for fetching large number of images composing a |
| 155 | page : |
| 156 | |
| 157 | [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ... |
| 158 | |
| 159 | This can obviously have a tremendous benefit on performance because the network |
| 160 | latency is eliminated between subsequent requests. Many HTTP agents do not |
| 161 | correctly support pipelining since there is no way to associate a response with |
| 162 | the corresponding request in HTTP. For this reason, it is mandatory for the |
Cyril Bonté | 78caf84 | 2010-03-10 22:41:43 +0100 | [diff] [blame] | 163 | server to reply in the exact same order as the requests were received. |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 164 | |
Willy Tarreau | 70dffda | 2014-01-30 03:07:23 +0100 | [diff] [blame] | 165 | By default HAProxy operates in keep-alive mode with regards to persistent |
| 166 | connections: for each connection it processes each request and response, and |
| 167 | leaves the connection idle on both sides between the end of a response and the |
| 168 | start of a new request. |
Patrick Mezard | 9ec2ec4 | 2010-06-12 17:02:45 +0200 | [diff] [blame] | 169 | |
Willy Tarreau | 70dffda | 2014-01-30 03:07:23 +0100 | [diff] [blame] | 170 | HAProxy supports 5 connection modes : |
| 171 | - keep alive : all requests and responses are processed (default) |
| 172 | - tunnel : only the first request and response are processed, |
| 173 | everything else is forwarded with no analysis. |
| 174 | - passive close : tunnel with "Connection: close" added in both directions. |
| 175 | - server close : the server-facing connection is closed after the response. |
| 176 | - forced close : the connection is actively closed after end of response. |
| 177 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 178 | |
| 179 | 1.2. HTTP request |
| 180 | ----------------- |
| 181 | |
| 182 | First, let's consider this HTTP request : |
| 183 | |
| 184 | Line Contents |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 185 | number |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 186 | 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1 |
| 187 | 2 Host: www.mydomain.com |
| 188 | 3 User-agent: my small browser |
| 189 | 4 Accept: image/jpeg, image/gif |
| 190 | 5 Accept: image/png |
| 191 | |
| 192 | |
| 193 | 1.2.1. The Request line |
| 194 | ----------------------- |
| 195 | |
| 196 | Line 1 is the "request line". It is always composed of 3 fields : |
| 197 | |
| 198 | - a METHOD : GET |
| 199 | - a URI : /serv/login.php?lang=en&profile=2 |
| 200 | - a version tag : HTTP/1.1 |
| 201 | |
| 202 | All of them are delimited by what the standard calls LWS (linear white spaces), |
| 203 | which are commonly spaces, but can also be tabs or line feeds/carriage returns |
| 204 | followed by spaces/tabs. The method itself cannot contain any colon (':') and |
| 205 | is limited to alphabetic letters. All those various combinations make it |
| 206 | desirable that HAProxy performs the splitting itself rather than leaving it to |
| 207 | the user to write a complex or inaccurate regular expression. |
| 208 | |
| 209 | The URI itself can have several forms : |
| 210 | |
| 211 | - A "relative URI" : |
| 212 | |
| 213 | /serv/login.php?lang=en&profile=2 |
| 214 | |
| 215 | It is a complete URL without the host part. This is generally what is |
| 216 | received by servers, reverse proxies and transparent proxies. |
| 217 | |
| 218 | - An "absolute URI", also called a "URL" : |
| 219 | |
| 220 | http://192.168.0.12:8080/serv/login.php?lang=en&profile=2 |
| 221 | |
| 222 | It is composed of a "scheme" (the protocol name followed by '://'), a host |
| 223 | name or address, optionally a colon (':') followed by a port number, then |
| 224 | a relative URI beginning at the first slash ('/') after the address part. |
| 225 | This is generally what proxies receive, but a server supporting HTTP/1.1 |
| 226 | must accept this form too. |
| 227 | |
| 228 | - a star ('*') : this form is only accepted in association with the OPTIONS |
| 229 | method and is not relayable. It is used to inquiry a next hop's |
| 230 | capabilities. |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 231 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 232 | - an address:port combination : 192.168.0.12:80 |
| 233 | This is used with the CONNECT method, which is used to establish TCP |
| 234 | tunnels through HTTP proxies, generally for HTTPS, but sometimes for |
| 235 | other protocols too. |
| 236 | |
| 237 | In a relative URI, two sub-parts are identified. The part before the question |
| 238 | mark is called the "path". It is typically the relative path to static objects |
| 239 | on the server. The part after the question mark is called the "query string". |
| 240 | It is mostly used with GET requests sent to dynamic scripts and is very |
| 241 | specific to the language, framework or application in use. |
| 242 | |
| 243 | |
| 244 | 1.2.2. The request headers |
| 245 | -------------------------- |
| 246 | |
| 247 | The headers start at the second line. They are composed of a name at the |
| 248 | beginning of the line, immediately followed by a colon (':'). Traditionally, |
| 249 | an LWS is added after the colon but that's not required. Then come the values. |
| 250 | Multiple identical headers may be folded into one single line, delimiting the |
| 251 | values with commas, provided that their order is respected. This is commonly |
| 252 | encountered in the "Cookie:" field. A header may span over multiple lines if |
| 253 | the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5 |
| 254 | define a total of 3 values for the "Accept:" header. |
| 255 | |
| 256 | Contrary to a common mis-conception, header names are not case-sensitive, and |
| 257 | their values are not either if they refer to other header names (such as the |
| 258 | "Connection:" header). |
| 259 | |
| 260 | The end of the headers is indicated by the first empty line. People often say |
| 261 | that it's a double line feed, which is not exact, even if a double line feed |
| 262 | is one valid form of empty line. |
| 263 | |
| 264 | Fortunately, HAProxy takes care of all these complex combinations when indexing |
| 265 | headers, checking values and counting them, so there is no reason to worry |
| 266 | about the way they could be written, but it is important not to accuse an |
| 267 | application of being buggy if it does unusual, valid things. |
| 268 | |
| 269 | Important note: |
| 270 | As suggested by RFC2616, HAProxy normalizes headers by replacing line breaks |
| 271 | in the middle of headers by LWS in order to join multi-line headers. This |
| 272 | is necessary for proper analysis and helps less capable HTTP parsers to work |
| 273 | correctly and not to be fooled by such complex constructs. |
| 274 | |
| 275 | |
| 276 | 1.3. HTTP response |
| 277 | ------------------ |
| 278 | |
| 279 | An HTTP response looks very much like an HTTP request. Both are called HTTP |
| 280 | messages. Let's consider this HTTP response : |
| 281 | |
| 282 | Line Contents |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 283 | number |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 284 | 1 HTTP/1.1 200 OK |
| 285 | 2 Content-length: 350 |
| 286 | 3 Content-Type: text/html |
| 287 | |
Willy Tarreau | 816b979 | 2009-09-15 21:25:21 +0200 | [diff] [blame] | 288 | As a special case, HTTP supports so called "Informational responses" as status |
| 289 | codes 1xx. These messages are special in that they don't convey any part of the |
| 290 | response, they're just used as sort of a signaling message to ask a client to |
Willy Tarreau | 5843d1a | 2010-02-01 15:13:32 +0100 | [diff] [blame] | 291 | continue to post its request for instance. In the case of a status 100 response |
| 292 | the requested information will be carried by the next non-100 response message |
| 293 | following the informational one. This implies that multiple responses may be |
| 294 | sent to a single request, and that this only works when keep-alive is enabled |
| 295 | (1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to |
| 296 | correctly forward and skip them, and only process the next non-100 response. As |
| 297 | such, these messages are neither logged nor transformed, unless explicitly |
| 298 | state otherwise. Status 101 messages indicate that the protocol is changing |
| 299 | over the same connection and that haproxy must switch to tunnel mode, just as |
| 300 | if a CONNECT had occurred. Then the Upgrade header would contain additional |
| 301 | information about the type of protocol the connection is switching to. |
Willy Tarreau | 816b979 | 2009-09-15 21:25:21 +0200 | [diff] [blame] | 302 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 303 | |
| 304 | 1.3.1. The Response line |
| 305 | ------------------------ |
| 306 | |
| 307 | Line 1 is the "response line". It is always composed of 3 fields : |
| 308 | |
| 309 | - a version tag : HTTP/1.1 |
| 310 | - a status code : 200 |
| 311 | - a reason : OK |
| 312 | |
| 313 | The status code is always 3-digit. The first digit indicates a general status : |
Willy Tarreau | 816b979 | 2009-09-15 21:25:21 +0200 | [diff] [blame] | 314 | - 1xx = informational message to be skipped (eg: 100, 101) |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 315 | - 2xx = OK, content is following (eg: 200, 206) |
| 316 | - 3xx = OK, no content following (eg: 302, 304) |
| 317 | - 4xx = error caused by the client (eg: 401, 403, 404) |
| 318 | - 5xx = error caused by the server (eg: 500, 502, 503) |
| 319 | |
| 320 | Please refer to RFC2616 for the detailed meaning of all such codes. The |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 321 | "reason" field is just a hint, but is not parsed by clients. Anything can be |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 322 | found there, but it's a common practice to respect the well-established |
| 323 | messages. It can be composed of one or multiple words, such as "OK", "Found", |
| 324 | or "Authentication Required". |
| 325 | |
| 326 | Haproxy may emit the following status codes by itself : |
| 327 | |
| 328 | Code When / reason |
| 329 | 200 access to stats page, and when replying to monitoring requests |
| 330 | 301 when performing a redirection, depending on the configured code |
| 331 | 302 when performing a redirection, depending on the configured code |
| 332 | 303 when performing a redirection, depending on the configured code |
Willy Tarreau | b67fdc4 | 2013-03-29 19:28:11 +0100 | [diff] [blame] | 333 | 307 when performing a redirection, depending on the configured code |
| 334 | 308 when performing a redirection, depending on the configured code |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 335 | 400 for an invalid or too large request |
| 336 | 401 when an authentication is required to perform the action (when |
| 337 | accessing the stats page) |
| 338 | 403 when a request is forbidden by a "block" ACL or "reqdeny" filter |
| 339 | 408 when the request timeout strikes before the request is complete |
| 340 | 500 when haproxy encounters an unrecoverable internal error, such as a |
| 341 | memory allocation failure, which should never happen |
| 342 | 502 when the server returns an empty, invalid or incomplete response, or |
| 343 | when an "rspdeny" filter blocks the response. |
| 344 | 503 when no server was available to handle the request, or in response to |
| 345 | monitoring requests which match the "monitor fail" condition |
| 346 | 504 when the response timeout strikes before the server responds |
| 347 | |
| 348 | The error 4xx and 5xx codes above may be customized (see "errorloc" in section |
| 349 | 4.2). |
| 350 | |
| 351 | |
| 352 | 1.3.2. The response headers |
| 353 | --------------------------- |
| 354 | |
| 355 | Response headers work exactly like request headers, and as such, HAProxy uses |
| 356 | the same parsing function for both. Please refer to paragraph 1.2.2 for more |
| 357 | details. |
| 358 | |
| 359 | |
| 360 | 2. Configuring HAProxy |
| 361 | ---------------------- |
| 362 | |
| 363 | 2.1. Configuration file format |
| 364 | ------------------------------ |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 365 | |
| 366 | HAProxy's configuration process involves 3 major sources of parameters : |
| 367 | |
| 368 | - the arguments from the command-line, which always take precedence |
| 369 | - the "global" section, which sets process-wide parameters |
| 370 | - the proxies sections which can take form of "defaults", "listen", |
| 371 | "frontend" and "backend". |
| 372 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 373 | The configuration file syntax consists in lines beginning with a keyword |
| 374 | referenced in this manual, optionally followed by one or several parameters |
William Lallemand | f9873ba | 2015-05-05 17:37:14 +0200 | [diff] [blame] | 375 | delimited by spaces. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 376 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 377 | |
William Lallemand | f9873ba | 2015-05-05 17:37:14 +0200 | [diff] [blame] | 378 | 2.2. Quoting and escaping |
| 379 | ------------------------- |
| 380 | |
| 381 | HAProxy's configuration introduces a quoting and escaping system similar to |
| 382 | many programming languages. The configuration file supports 3 types: escaping |
| 383 | with a backslash, weak quoting with double quotes, and strong quoting with |
| 384 | single quotes. |
| 385 | |
| 386 | If spaces have to be entered in strings, then they must be escaped by preceding |
| 387 | them by a backslash ('\') or by quoting them. Backslashes also have to be |
| 388 | escaped by doubling or strong quoting them. |
| 389 | |
| 390 | Escaping is achieved by preceding a special character by a backslash ('\'): |
| 391 | |
| 392 | \ to mark a space and differentiate it from a delimiter |
| 393 | \# to mark a hash and differentiate it from a comment |
| 394 | \\ to use a backslash |
| 395 | \' to use a single quote and differentiate it from strong quoting |
| 396 | \" to use a double quote and differentiate it from weak quoting |
| 397 | |
| 398 | Weak quoting is achieved by using double quotes (""). Weak quoting prevents |
| 399 | the interpretation of: |
| 400 | |
| 401 | space as a parameter separator |
| 402 | ' single quote as a strong quoting delimiter |
| 403 | # hash as a comment start |
| 404 | |
William Lallemand | b2f0745 | 2015-05-12 14:27:13 +0200 | [diff] [blame] | 405 | Weak quoting permits the interpretation of variables, if you want to use a non |
| 406 | -interpreted dollar within a double quoted string, you should escape it with a |
| 407 | backslash ("\$"), it does not work outside weak quoting. |
| 408 | |
| 409 | Interpretation of escaping and special characters are not prevented by weak |
William Lallemand | f9873ba | 2015-05-05 17:37:14 +0200 | [diff] [blame] | 410 | quoting. |
| 411 | |
| 412 | Strong quoting is achieved by using single quotes (''). Inside single quotes, |
| 413 | nothing is interpreted, it's the efficient way to quote regexes. |
| 414 | |
| 415 | Quoted and escaped strings are replaced in memory by their interpreted |
| 416 | equivalent, it allows you to perform concatenation. |
| 417 | |
| 418 | Example: |
| 419 | # those are equivalents: |
| 420 | log-format %{+Q}o\ %t\ %s\ %{-Q}r |
| 421 | log-format "%{+Q}o %t %s %{-Q}r" |
| 422 | log-format '%{+Q}o %t %s %{-Q}r' |
| 423 | log-format "%{+Q}o %t"' %s %{-Q}r' |
| 424 | log-format "%{+Q}o %t"' %s'\ %{-Q}r |
| 425 | |
| 426 | # those are equivalents: |
| 427 | reqrep "^([^\ :]*)\ /static/(.*)" \1\ /\2 |
| 428 | reqrep "^([^ :]*)\ /static/(.*)" '\1 /\2' |
| 429 | reqrep "^([^ :]*)\ /static/(.*)" "\1 /\2" |
| 430 | reqrep "^([^ :]*)\ /static/(.*)" "\1\ /\2" |
| 431 | |
| 432 | |
William Lallemand | b2f0745 | 2015-05-12 14:27:13 +0200 | [diff] [blame] | 433 | 2.3. Environment variables |
| 434 | -------------------------- |
| 435 | |
| 436 | HAProxy's configuration supports environment variables. Those variables are |
| 437 | interpreted only within double quotes. Variables are expanded during the |
| 438 | configuration parsing. Variable names must be preceded by a dollar ("$") and |
| 439 | optionally enclosed with braces ("{}") similarly to what is done in Bourne |
| 440 | shell. Variable names can contain alphanumerical characters or the character |
| 441 | underscore ("_") but should not start with a digit. |
| 442 | |
| 443 | Example: |
| 444 | |
| 445 | bind "fd@${FD_APP1}" |
| 446 | |
| 447 | log "${LOCAL_SYSLOG}:514" local0 notice # send to local server |
| 448 | |
| 449 | user "$HAPROXY_USER" |
| 450 | |
| 451 | |
| 452 | 2.4. Time format |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 453 | ---------------- |
| 454 | |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 455 | Some parameters involve values representing time, such as timeouts. These |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 456 | values are generally expressed in milliseconds (unless explicitly stated |
| 457 | otherwise) but may be expressed in any other unit by suffixing the unit to the |
| 458 | numeric value. It is important to consider this because it will not be repeated |
| 459 | for every keyword. Supported units are : |
| 460 | |
| 461 | - us : microseconds. 1 microsecond = 1/1000000 second |
| 462 | - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default. |
| 463 | - s : seconds. 1s = 1000ms |
| 464 | - m : minutes. 1m = 60s = 60000ms |
| 465 | - h : hours. 1h = 60m = 3600s = 3600000ms |
| 466 | - d : days. 1d = 24h = 1440m = 86400s = 86400000ms |
| 467 | |
| 468 | |
William Lallemand | f9873ba | 2015-05-05 17:37:14 +0200 | [diff] [blame] | 469 | 2.4. Examples |
Patrick Mezard | 35da19c | 2010-06-12 17:02:47 +0200 | [diff] [blame] | 470 | ------------- |
| 471 | |
| 472 | # Simple configuration for an HTTP proxy listening on port 80 on all |
| 473 | # interfaces and forwarding requests to a single backend "servers" with a |
| 474 | # single server "server1" listening on 127.0.0.1:8000 |
| 475 | global |
| 476 | daemon |
| 477 | maxconn 256 |
| 478 | |
| 479 | defaults |
| 480 | mode http |
| 481 | timeout connect 5000ms |
| 482 | timeout client 50000ms |
| 483 | timeout server 50000ms |
| 484 | |
| 485 | frontend http-in |
| 486 | bind *:80 |
| 487 | default_backend servers |
| 488 | |
| 489 | backend servers |
| 490 | server server1 127.0.0.1:8000 maxconn 32 |
| 491 | |
| 492 | |
| 493 | # The same configuration defined with a single listen block. Shorter but |
| 494 | # less expressive, especially in HTTP mode. |
| 495 | global |
| 496 | daemon |
| 497 | maxconn 256 |
| 498 | |
| 499 | defaults |
| 500 | mode http |
| 501 | timeout connect 5000ms |
| 502 | timeout client 50000ms |
| 503 | timeout server 50000ms |
| 504 | |
| 505 | listen http-in |
| 506 | bind *:80 |
| 507 | server server1 127.0.0.1:8000 maxconn 32 |
| 508 | |
| 509 | |
| 510 | Assuming haproxy is in $PATH, test these configurations in a shell with: |
| 511 | |
Willy Tarreau | ccb289d | 2010-12-11 20:19:38 +0100 | [diff] [blame] | 512 | $ sudo haproxy -f configuration.conf -c |
Patrick Mezard | 35da19c | 2010-06-12 17:02:47 +0200 | [diff] [blame] | 513 | |
| 514 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 515 | 3. Global parameters |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 516 | -------------------- |
| 517 | |
| 518 | Parameters in the "global" section are process-wide and often OS-specific. They |
| 519 | are generally set once for all and do not need being changed once correct. Some |
| 520 | of them have command-line equivalents. |
| 521 | |
| 522 | The following keywords are supported in the "global" section : |
| 523 | |
| 524 | * Process management and security |
Emeric Brun | c8e8d12 | 2012-10-02 18:42:10 +0200 | [diff] [blame] | 525 | - ca-base |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 526 | - chroot |
Emeric Brun | c8e8d12 | 2012-10-02 18:42:10 +0200 | [diff] [blame] | 527 | - crt-base |
Baptiste Assmann | 3493d0f | 2015-10-12 20:21:23 +0200 | [diff] [blame] | 528 | - cpu-map |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 529 | - daemon |
Baptiste Assmann | 3493d0f | 2015-10-12 20:21:23 +0200 | [diff] [blame] | 530 | - description |
| 531 | - deviceatlas-json-file |
| 532 | - deviceatlas-log-level |
| 533 | - deviceatlas-separator |
| 534 | - deviceatlas-properties-cookie |
Simon Horman | 98637e5 | 2014-06-20 12:30:16 +0900 | [diff] [blame] | 535 | - external-check |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 536 | - gid |
| 537 | - group |
| 538 | - log |
Baptiste Assmann | 3493d0f | 2015-10-12 20:21:23 +0200 | [diff] [blame] | 539 | - log-tag |
Joe Williams | df5b38f | 2010-12-29 17:05:48 +0100 | [diff] [blame] | 540 | - log-send-hostname |
Baptiste Assmann | 3493d0f | 2015-10-12 20:21:23 +0200 | [diff] [blame] | 541 | - lua-load |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 542 | - nbproc |
Baptiste Assmann | 3493d0f | 2015-10-12 20:21:23 +0200 | [diff] [blame] | 543 | - node |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 544 | - pidfile |
Willy Tarreau | 1d54972 | 2016-02-16 12:41:57 +0100 | [diff] [blame] | 545 | - presetenv |
| 546 | - resetenv |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 547 | - uid |
| 548 | - ulimit-n |
| 549 | - user |
Willy Tarreau | 1d54972 | 2016-02-16 12:41:57 +0100 | [diff] [blame] | 550 | - setenv |
Willy Tarreau | fbee713 | 2007-10-18 13:53:22 +0200 | [diff] [blame] | 551 | - stats |
Baptiste Assmann | 3493d0f | 2015-10-12 20:21:23 +0200 | [diff] [blame] | 552 | - ssl-default-bind-ciphers |
| 553 | - ssl-default-bind-options |
| 554 | - ssl-default-server-ciphers |
| 555 | - ssl-default-server-options |
| 556 | - ssl-dh-param-file |
Emeric Brun | 850efd5 | 2014-01-29 12:24:34 +0100 | [diff] [blame] | 557 | - ssl-server-verify |
Willy Tarreau | ceb24bc | 2010-11-09 12:46:41 +0100 | [diff] [blame] | 558 | - unix-bind |
Willy Tarreau | 1d54972 | 2016-02-16 12:41:57 +0100 | [diff] [blame] | 559 | - unsetenv |
Thomas Holmes | db04f19 | 2015-05-18 13:21:39 +0100 | [diff] [blame] | 560 | - 51degrees-data-file |
| 561 | - 51degrees-property-name-list |
Dragan Dosen | 93b38d9 | 2015-06-29 16:43:25 +0200 | [diff] [blame] | 562 | - 51degrees-property-separator |
Dragan Dosen | ae6d39a | 2015-06-29 16:43:27 +0200 | [diff] [blame] | 563 | - 51degrees-cache-size |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 564 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 565 | * Performance tuning |
Willy Tarreau | 1746eec | 2014-04-25 10:46:47 +0200 | [diff] [blame] | 566 | - max-spread-checks |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 567 | - maxconn |
Willy Tarreau | 81c25d0 | 2011-09-07 15:17:21 +0200 | [diff] [blame] | 568 | - maxconnrate |
William Lallemand | d85f917 | 2012-11-09 17:05:39 +0100 | [diff] [blame] | 569 | - maxcomprate |
William Lallemand | 072a2bf | 2012-11-20 17:01:01 +0100 | [diff] [blame] | 570 | - maxcompcpuusage |
Willy Tarreau | ff4f82d | 2009-02-06 11:28:13 +0100 | [diff] [blame] | 571 | - maxpipes |
Willy Tarreau | 93e7c00 | 2013-10-07 18:51:07 +0200 | [diff] [blame] | 572 | - maxsessrate |
Willy Tarreau | 403edff | 2012-09-06 11:58:37 +0200 | [diff] [blame] | 573 | - maxsslconn |
Willy Tarreau | e43d532 | 2013-10-07 20:01:52 +0200 | [diff] [blame] | 574 | - maxsslrate |
Baptiste Assmann | 3493d0f | 2015-10-12 20:21:23 +0200 | [diff] [blame] | 575 | - maxzlibmem |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 576 | - noepoll |
| 577 | - nokqueue |
| 578 | - nopoll |
Willy Tarreau | ff4f82d | 2009-02-06 11:28:13 +0100 | [diff] [blame] | 579 | - nosplice |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 580 | - nogetaddrinfo |
Willy Tarreau | fe255b7 | 2007-10-14 23:09:26 +0200 | [diff] [blame] | 581 | - spread-checks |
Baptiste Assmann | 5626f48 | 2015-08-23 10:00:10 +0200 | [diff] [blame] | 582 | - server-state-base |
Baptiste Assmann | ef1f0fc | 2015-08-23 10:06:39 +0200 | [diff] [blame] | 583 | - server-state-file |
Baptiste Assmann | 3493d0f | 2015-10-12 20:21:23 +0200 | [diff] [blame] | 584 | - tune.buffers.limit |
| 585 | - tune.buffers.reserve |
Willy Tarreau | 27a674e | 2009-08-17 07:23:33 +0200 | [diff] [blame] | 586 | - tune.bufsize |
Willy Tarreau | 43961d5 | 2010-10-04 20:39:20 +0200 | [diff] [blame] | 587 | - tune.chksize |
William Lallemand | f374783 | 2012-11-09 12:33:10 +0100 | [diff] [blame] | 588 | - tune.comp.maxlevel |
Willy Tarreau | 193b8c6 | 2012-11-22 00:17:38 +0100 | [diff] [blame] | 589 | - tune.http.cookielen |
Willy Tarreau | ac1932d | 2011-10-24 19:14:41 +0200 | [diff] [blame] | 590 | - tune.http.maxhdr |
Willy Tarreau | 7e31273 | 2014-02-12 16:35:14 +0100 | [diff] [blame] | 591 | - tune.idletimer |
Thierry FOURNIER | 90da191 | 2015-03-05 11:17:06 +0100 | [diff] [blame] | 592 | - tune.lua.forced-yield |
Willy Tarreau | 32f61e2 | 2015-03-18 17:54:59 +0100 | [diff] [blame] | 593 | - tune.lua.maxmem |
Thierry FOURNIER | 90da191 | 2015-03-05 11:17:06 +0100 | [diff] [blame] | 594 | - tune.lua.session-timeout |
| 595 | - tune.lua.task-timeout |
Thierry FOURNIER | 7dd784b | 2015-10-01 14:49:33 +0200 | [diff] [blame] | 596 | - tune.lua.service-timeout |
Willy Tarreau | a0250ba | 2008-01-06 11:22:57 +0100 | [diff] [blame] | 597 | - tune.maxaccept |
| 598 | - tune.maxpollevents |
Willy Tarreau | 27a674e | 2009-08-17 07:23:33 +0200 | [diff] [blame] | 599 | - tune.maxrewrite |
Willy Tarreau | f3045d2 | 2015-04-29 16:24:50 +0200 | [diff] [blame] | 600 | - tune.pattern.cache-size |
Willy Tarreau | bd9a0a7 | 2011-10-23 21:14:29 +0200 | [diff] [blame] | 601 | - tune.pipesize |
Willy Tarreau | e803de2 | 2010-01-21 17:43:04 +0100 | [diff] [blame] | 602 | - tune.rcvbuf.client |
| 603 | - tune.rcvbuf.server |
Willy Tarreau | b22fc30 | 2015-12-14 12:04:35 +0100 | [diff] [blame] | 604 | - tune.recv_enough |
Willy Tarreau | e803de2 | 2010-01-21 17:43:04 +0100 | [diff] [blame] | 605 | - tune.sndbuf.client |
| 606 | - tune.sndbuf.server |
Willy Tarreau | 6ec58db | 2012-11-16 16:32:15 +0100 | [diff] [blame] | 607 | - tune.ssl.cachesize |
Willy Tarreau | bfd5946 | 2013-02-21 07:46:09 +0100 | [diff] [blame] | 608 | - tune.ssl.lifetime |
Emeric Brun | 8dc6039 | 2014-05-09 13:52:00 +0200 | [diff] [blame] | 609 | - tune.ssl.force-private-cache |
Willy Tarreau | bfd5946 | 2013-02-21 07:46:09 +0100 | [diff] [blame] | 610 | - tune.ssl.maxrecord |
Remi Gacogne | f46cd6e | 2014-06-12 14:58:40 +0200 | [diff] [blame] | 611 | - tune.ssl.default-dh-param |
Christopher Faulet | 31af49d | 2015-06-09 17:29:50 +0200 | [diff] [blame] | 612 | - tune.ssl.ssl-ctx-cache-size |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 613 | - tune.vars.global-max-size |
| 614 | - tune.vars.reqres-max-size |
| 615 | - tune.vars.sess-max-size |
| 616 | - tune.vars.txn-max-size |
William Lallemand | a509e4c | 2012-11-07 16:54:34 +0100 | [diff] [blame] | 617 | - tune.zlib.memlevel |
| 618 | - tune.zlib.windowsize |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 619 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 620 | * Debugging |
| 621 | - debug |
| 622 | - quiet |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 623 | |
| 624 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 625 | 3.1. Process management and security |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 626 | ------------------------------------ |
| 627 | |
Emeric Brun | c8e8d12 | 2012-10-02 18:42:10 +0200 | [diff] [blame] | 628 | ca-base <dir> |
| 629 | Assigns a default directory to fetch SSL CA certificates and CRLs from when a |
Emeric Brun | fd33a26 | 2012-10-11 16:28:27 +0200 | [diff] [blame] | 630 | relative path is used with "ca-file" or "crl-file" directives. Absolute |
| 631 | locations specified in "ca-file" and "crl-file" prevail and ignore "ca-base". |
Emeric Brun | c8e8d12 | 2012-10-02 18:42:10 +0200 | [diff] [blame] | 632 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 633 | chroot <jail dir> |
| 634 | Changes current directory to <jail dir> and performs a chroot() there before |
| 635 | dropping privileges. This increases the security level in case an unknown |
| 636 | vulnerability would be exploited, since it would make it very hard for the |
| 637 | attacker to exploit the system. This only works when the process is started |
| 638 | with superuser privileges. It is important to ensure that <jail_dir> is both |
| 639 | empty and unwritable to anyone. |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 640 | |
Willy Tarreau | fc6c032 | 2012-11-16 16:12:27 +0100 | [diff] [blame] | 641 | cpu-map <"all"|"odd"|"even"|process_num> <cpu-set>... |
| 642 | On Linux 2.6 and above, it is possible to bind a process to a specific CPU |
| 643 | set. This means that the process will never run on other CPUs. The "cpu-map" |
| 644 | directive specifies CPU sets for process sets. The first argument is the |
Willy Tarreau | a9db57e | 2013-01-18 11:29:29 +0100 | [diff] [blame] | 645 | process number to bind. This process must have a number between 1 and 32 or |
| 646 | 64, depending on the machine's word size, and any process IDs above nbproc |
| 647 | are ignored. It is possible to specify all processes at once using "all", |
| 648 | only odd numbers using "odd" or even numbers using "even", just like with the |
| 649 | "bind-process" directive. The second and forthcoming arguments are CPU sets. |
| 650 | Each CPU set is either a unique number between 0 and 31 or 63 or a range with |
| 651 | two such numbers delimited by a dash ('-'). Multiple CPU numbers or ranges |
| 652 | may be specified, and the processes will be allowed to bind to all of them. |
| 653 | Obviously, multiple "cpu-map" directives may be specified. Each "cpu-map" |
| 654 | directive will replace the previous ones when they overlap. |
Willy Tarreau | fc6c032 | 2012-11-16 16:12:27 +0100 | [diff] [blame] | 655 | |
Emeric Brun | c8e8d12 | 2012-10-02 18:42:10 +0200 | [diff] [blame] | 656 | crt-base <dir> |
| 657 | Assigns a default directory to fetch SSL certificates from when a relative |
| 658 | path is used with "crtfile" directives. Absolute locations specified after |
| 659 | "crtfile" prevail and ignore "crt-base". |
| 660 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 661 | daemon |
| 662 | Makes the process fork into background. This is the recommended mode of |
| 663 | operation. It is equivalent to the command line "-D" argument. It can be |
| 664 | disabled by the command line "-db" argument. |
| 665 | |
David Carlier | 8167f30 | 2015-06-01 13:50:06 +0200 | [diff] [blame] | 666 | deviceatlas-json-file <path> |
| 667 | Sets the path of the DeviceAtlas JSON data file to be loaded by the API. |
| 668 | The path must be a valid JSON data file and accessible by Haproxy process. |
| 669 | |
| 670 | deviceatlas-log-level <value> |
| 671 | Sets the level of informations returned by the API. This directive is |
| 672 | optional and set to 0 by default if not set. |
| 673 | |
| 674 | deviceatlas-separator <char> |
| 675 | Sets the character separator for the API properties results. This directive |
| 676 | is optional and set to | by default if not set. |
| 677 | |
Cyril Bonté | 0306c4a | 2015-10-26 22:37:38 +0100 | [diff] [blame] | 678 | deviceatlas-properties-cookie <name> |
Cyril Bonté | 307ee1e | 2015-09-28 23:16:06 +0200 | [diff] [blame] | 679 | Sets the client cookie's name used for the detection if the DeviceAtlas |
| 680 | Client-side component was used during the request. This directive is optional |
| 681 | and set to DAPROPS by default if not set. |
David Carlier | 29b3ca3 | 2015-09-25 14:09:21 +0100 | [diff] [blame] | 682 | |
Simon Horman | 98637e5 | 2014-06-20 12:30:16 +0900 | [diff] [blame] | 683 | external-check |
| 684 | Allows the use of an external agent to perform health checks. |
| 685 | This is disabled by default as a security precaution. |
| 686 | See "option external-check". |
| 687 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 688 | gid <number> |
| 689 | Changes the process' group ID to <number>. It is recommended that the group |
| 690 | ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must |
| 691 | be started with a user belonging to this group, or with superuser privileges. |
Michael Scherer | ab012dd | 2013-01-12 18:35:19 +0100 | [diff] [blame] | 692 | Note that if haproxy is started from a user having supplementary groups, it |
| 693 | will only be able to drop these groups if started with superuser privileges. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 694 | See also "group" and "uid". |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 695 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 696 | group <group name> |
| 697 | Similar to "gid" but uses the GID of group name <group name> from /etc/group. |
| 698 | See also "gid" and "user". |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 699 | |
Dragan Dosen | 7ad3154 | 2015-09-28 17:16:47 +0200 | [diff] [blame] | 700 | log <address> [len <length>] [format <format>] <facility> [max level [min level]] |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 701 | Adds a global syslog server. Up to two global servers can be defined. They |
| 702 | will receive logs for startups and exits, as well as all logs from proxies |
Robert Tsai | 81ae195 | 2007-12-05 10:47:29 +0100 | [diff] [blame] | 703 | configured with "log global". |
| 704 | |
| 705 | <address> can be one of: |
| 706 | |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 707 | - An IPv4 address optionally followed by a colon and a UDP port. If |
Robert Tsai | 81ae195 | 2007-12-05 10:47:29 +0100 | [diff] [blame] | 708 | no port is specified, 514 is used by default (the standard syslog |
| 709 | port). |
| 710 | |
David du Colombier | 24bb5f5 | 2011-03-17 10:40:23 +0100 | [diff] [blame] | 711 | - An IPv6 address followed by a colon and optionally a UDP port. If |
| 712 | no port is specified, 514 is used by default (the standard syslog |
| 713 | port). |
| 714 | |
Robert Tsai | 81ae195 | 2007-12-05 10:47:29 +0100 | [diff] [blame] | 715 | - A filesystem path to a UNIX domain socket, keeping in mind |
| 716 | considerations for chroot (be sure the path is accessible inside |
| 717 | the chroot) and uid/gid (be sure the path is appropriately |
| 718 | writeable). |
| 719 | |
William Lallemand | b2f0745 | 2015-05-12 14:27:13 +0200 | [diff] [blame] | 720 | You may want to reference some environment variables in the address |
| 721 | parameter, see section 2.3 about environment variables. |
Willy Tarreau | dad36a3 | 2013-03-11 01:20:04 +0100 | [diff] [blame] | 722 | |
Willy Tarreau | 18324f5 | 2014-06-27 18:10:07 +0200 | [diff] [blame] | 723 | <length> is an optional maximum line length. Log lines larger than this value |
| 724 | will be truncated before being sent. The reason is that syslog |
| 725 | servers act differently on log line length. All servers support the |
| 726 | default value of 1024, but some servers simply drop larger lines |
| 727 | while others do log them. If a server supports long lines, it may |
| 728 | make sense to set this value here in order to avoid truncating long |
| 729 | lines. Similarly, if a server drops long lines, it is preferable to |
| 730 | truncate them before sending them. Accepted values are 80 to 65535 |
| 731 | inclusive. The default value of 1024 is generally fine for all |
| 732 | standard usages. Some specific cases of long captures or |
| 733 | JSON-formated logs may require larger values. |
| 734 | |
Dragan Dosen | 7ad3154 | 2015-09-28 17:16:47 +0200 | [diff] [blame] | 735 | <format> is the log format used when generating syslog messages. It may be |
| 736 | one of the following : |
| 737 | |
| 738 | rfc3164 The RFC3164 syslog message format. This is the default. |
| 739 | (https://tools.ietf.org/html/rfc3164) |
| 740 | |
| 741 | rfc5424 The RFC5424 syslog message format. |
| 742 | (https://tools.ietf.org/html/rfc5424) |
| 743 | |
Robert Tsai | 81ae195 | 2007-12-05 10:47:29 +0100 | [diff] [blame] | 744 | <facility> must be one of the 24 standard syslog facilities : |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 745 | |
| 746 | kern user mail daemon auth syslog lpr news |
| 747 | uucp cron auth2 ftp ntp audit alert cron2 |
| 748 | local0 local1 local2 local3 local4 local5 local6 local7 |
| 749 | |
| 750 | An optional level can be specified to filter outgoing messages. By default, |
Willy Tarreau | f7edefa | 2009-05-10 17:20:05 +0200 | [diff] [blame] | 751 | all messages are sent. If a maximum level is specified, only messages with a |
| 752 | severity at least as important as this level will be sent. An optional minimum |
| 753 | level can be specified. If it is set, logs emitted with a more severe level |
| 754 | than this one will be capped to this level. This is used to avoid sending |
| 755 | "emerg" messages on all terminals on some default syslog configurations. |
| 756 | Eight levels are known : |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 757 | |
Cyril Bonté | dc4d903 | 2012-04-08 21:57:39 +0200 | [diff] [blame] | 758 | emerg alert crit err warning notice info debug |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 759 | |
Joe Williams | df5b38f | 2010-12-29 17:05:48 +0100 | [diff] [blame] | 760 | log-send-hostname [<string>] |
| 761 | Sets the hostname field in the syslog header. If optional "string" parameter |
| 762 | is set the header is set to the string contents, otherwise uses the hostname |
| 763 | of the system. Generally used if one is not relaying logs through an |
| 764 | intermediate syslog server or for simply customizing the hostname printed in |
| 765 | the logs. |
| 766 | |
Kevinm | 48936af | 2010-12-22 16:08:21 +0000 | [diff] [blame] | 767 | log-tag <string> |
| 768 | Sets the tag field in the syslog header to this string. It defaults to the |
| 769 | program name as launched from the command line, which usually is "haproxy". |
| 770 | Sometimes it can be useful to differentiate between multiple processes |
Willy Tarreau | 094af4e | 2015-01-07 15:03:42 +0100 | [diff] [blame] | 771 | running on the same host. See also the per-proxy "log-tag" directive. |
Kevinm | 48936af | 2010-12-22 16:08:21 +0000 | [diff] [blame] | 772 | |
Thierry FOURNIER | 90da191 | 2015-03-05 11:17:06 +0100 | [diff] [blame] | 773 | lua-load <file> |
| 774 | This global directive loads and executes a Lua file. This directive can be |
| 775 | used multiple times. |
| 776 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 777 | nbproc <number> |
| 778 | Creates <number> processes when going daemon. This requires the "daemon" |
| 779 | mode. By default, only one process is created, which is the recommended mode |
| 780 | of operation. For systems limited to small sets of file descriptors per |
| 781 | process, it may be needed to fork multiple daemons. USING MULTIPLE PROCESSES |
| 782 | IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon". |
| 783 | |
| 784 | pidfile <pidfile> |
| 785 | Writes pids of all daemons into file <pidfile>. This option is equivalent to |
| 786 | the "-p" command line argument. The file must be accessible to the user |
| 787 | starting the process. See also "daemon". |
| 788 | |
Willy Tarreau | 1d54972 | 2016-02-16 12:41:57 +0100 | [diff] [blame] | 789 | presetenv <name> <value> |
| 790 | Sets environment variable <name> to value <value>. If the variable exists, it |
| 791 | is NOT overwritten. The changes immediately take effect so that the next line |
| 792 | in the configuration file sees the new value. See also "setenv", "resetenv", |
| 793 | and "unsetenv". |
| 794 | |
| 795 | resetenv [<name> ...] |
| 796 | Removes all environment variables except the ones specified in argument. It |
| 797 | allows to use a clean controlled environment before setting new values with |
| 798 | setenv or unsetenv. Please note that some internal functions may make use of |
| 799 | some environment variables, such as time manipulation functions, but also |
| 800 | OpenSSL or even external checks. This must be used with extreme care and only |
| 801 | after complete validation. The changes immediately take effect so that the |
| 802 | next line in the configuration file sees the new environment. See also |
| 803 | "setenv", "presetenv", and "unsetenv". |
| 804 | |
Willy Tarreau | a9db57e | 2013-01-18 11:29:29 +0100 | [diff] [blame] | 805 | stats bind-process [ all | odd | even | <number 1-64>[-<number 1-64>] ] ... |
Willy Tarreau | 35b7b16 | 2012-10-22 23:17:18 +0200 | [diff] [blame] | 806 | Limits the stats socket to a certain set of processes numbers. By default the |
| 807 | stats socket is bound to all processes, causing a warning to be emitted when |
| 808 | nbproc is greater than 1 because there is no way to select the target process |
| 809 | when connecting. However, by using this setting, it becomes possible to pin |
| 810 | the stats socket to a specific set of processes, typically the first one. The |
| 811 | warning will automatically be disabled when this setting is used, whatever |
Willy Tarreau | a9db57e | 2013-01-18 11:29:29 +0100 | [diff] [blame] | 812 | the number of processes used. The maximum process ID depends on the machine's |
Willy Tarreau | ae30253 | 2014-05-07 19:22:24 +0200 | [diff] [blame] | 813 | word size (32 or 64). A better option consists in using the "process" setting |
| 814 | of the "stats socket" line to force the process on each line. |
Willy Tarreau | 35b7b16 | 2012-10-22 23:17:18 +0200 | [diff] [blame] | 815 | |
Baptiste Assmann | 5626f48 | 2015-08-23 10:00:10 +0200 | [diff] [blame] | 816 | server-state-base <directory> |
| 817 | Specifies the directory prefix to be prepended in front of all servers state |
Baptiste Assmann | 01c6cc3 | 2015-08-23 11:45:29 +0200 | [diff] [blame] | 818 | file names which do not start with a '/'. See also "server-state-file", |
| 819 | "load-server-state-from-file" and "server-state-file-name". |
Baptiste Assmann | ef1f0fc | 2015-08-23 10:06:39 +0200 | [diff] [blame] | 820 | |
| 821 | server-state-file <file> |
| 822 | Specifies the path to the file containing state of servers. If the path starts |
| 823 | with a slash ('/'), it is considered absolute, otherwise it is considered |
| 824 | relative to the directory specified using "server-state-base" (if set) or to |
| 825 | the current directory. Before reloading HAProxy, it is possible to save the |
| 826 | servers' current state using the stats command "show servers state". The |
| 827 | output of this command must be written in the file pointed by <file>. When |
| 828 | starting up, before handling traffic, HAProxy will read, load and apply state |
| 829 | for each server found in the file and available in its current running |
Baptiste Assmann | 01c6cc3 | 2015-08-23 11:45:29 +0200 | [diff] [blame] | 830 | configuration. See also "server-state-base" and "show servers state", |
| 831 | "load-server-state-from-file" and "server-state-file-name" |
Baptiste Assmann | 5626f48 | 2015-08-23 10:00:10 +0200 | [diff] [blame] | 832 | |
Willy Tarreau | 1d54972 | 2016-02-16 12:41:57 +0100 | [diff] [blame] | 833 | setenv <name> <value> |
| 834 | Sets environment variable <name> to value <value>. If the variable exists, it |
| 835 | is overwritten. The changes immediately take effect so that the next line in |
| 836 | the configuration file sees the new value. See also "presetenv", "resetenv", |
| 837 | and "unsetenv". |
| 838 | |
Willy Tarreau | 610f04b | 2014-02-13 11:36:41 +0100 | [diff] [blame] | 839 | ssl-default-bind-ciphers <ciphers> |
| 840 | This setting is only available when support for OpenSSL was built in. It sets |
| 841 | the default string describing the list of cipher algorithms ("cipher suite") |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 842 | that are negotiated during the SSL/TLS handshake for all "bind" lines which |
Willy Tarreau | 610f04b | 2014-02-13 11:36:41 +0100 | [diff] [blame] | 843 | do not explicitly define theirs. The format of the string is defined in |
| 844 | "man 1 ciphers" from OpenSSL man pages, and can be for instance a string such |
| 845 | as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without quotes). Please check the |
| 846 | "bind" keyword for more information. |
| 847 | |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 848 | ssl-default-bind-options [<option>]... |
| 849 | This setting is only available when support for OpenSSL was built in. It sets |
| 850 | default ssl-options to force on all "bind" lines. Please check the "bind" |
| 851 | keyword to see available options. |
| 852 | |
| 853 | Example: |
| 854 | global |
| 855 | ssl-default-bind-options no-sslv3 no-tls-tickets |
| 856 | |
Willy Tarreau | 610f04b | 2014-02-13 11:36:41 +0100 | [diff] [blame] | 857 | ssl-default-server-ciphers <ciphers> |
| 858 | This setting is only available when support for OpenSSL was built in. It |
| 859 | sets the default string describing the list of cipher algorithms that are |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 860 | negotiated during the SSL/TLS handshake with the server, for all "server" |
Willy Tarreau | 610f04b | 2014-02-13 11:36:41 +0100 | [diff] [blame] | 861 | lines which do not explicitly define theirs. The format of the string is |
| 862 | defined in "man 1 ciphers". Please check the "server" keyword for more |
| 863 | information. |
| 864 | |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 865 | ssl-default-server-options [<option>]... |
| 866 | This setting is only available when support for OpenSSL was built in. It sets |
| 867 | default ssl-options to force on all "server" lines. Please check the "server" |
| 868 | keyword to see available options. |
| 869 | |
Remi Gacogne | 47783ef | 2015-05-29 15:53:22 +0200 | [diff] [blame] | 870 | ssl-dh-param-file <file> |
| 871 | This setting is only available when support for OpenSSL was built in. It sets |
| 872 | the default DH parameters that are used during the SSL/TLS handshake when |
| 873 | ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines |
| 874 | which do not explicitely define theirs. It will be overridden by custom DH |
| 875 | parameters found in a bind certificate file if any. If custom DH parameters |
Cyril Bonté | 307ee1e | 2015-09-28 23:16:06 +0200 | [diff] [blame] | 876 | are not specified either by using ssl-dh-param-file or by setting them |
| 877 | directly in the certificate file, pre-generated DH parameters of the size |
| 878 | specified by tune.ssl.default-dh-param will be used. Custom parameters are |
| 879 | known to be more secure and therefore their use is recommended. |
Remi Gacogne | 47783ef | 2015-05-29 15:53:22 +0200 | [diff] [blame] | 880 | Custom DH parameters may be generated by using the OpenSSL command |
| 881 | "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH |
| 882 | parameters should not be considered secure anymore. |
| 883 | |
Emeric Brun | 850efd5 | 2014-01-29 12:24:34 +0100 | [diff] [blame] | 884 | ssl-server-verify [none|required] |
| 885 | The default behavior for SSL verify on servers side. If specified to 'none', |
| 886 | servers certificates are not verified. The default is 'required' except if |
| 887 | forced using cmdline option '-dV'. |
| 888 | |
Willy Tarreau | abb175f | 2012-09-24 12:43:26 +0200 | [diff] [blame] | 889 | stats socket [<address:port>|<path>] [param*] |
| 890 | Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>. |
| 891 | Connections to this socket will return various statistics outputs and even |
| 892 | allow some commands to be issued to change some runtime settings. Please |
Kevin Decherf | 949c720 | 2015-10-13 23:26:44 +0200 | [diff] [blame] | 893 | consult section 9.2 "Unix Socket commands" of Management Guide for more |
| 894 | details. |
Willy Tarreau | 6162db2 | 2009-10-10 17:13:00 +0200 | [diff] [blame] | 895 | |
Willy Tarreau | abb175f | 2012-09-24 12:43:26 +0200 | [diff] [blame] | 896 | All parameters supported by "bind" lines are supported, for instance to |
| 897 | restrict access to some users or their access rights. Please consult |
| 898 | section 5.1 for more information. |
Willy Tarreau | fbee713 | 2007-10-18 13:53:22 +0200 | [diff] [blame] | 899 | |
| 900 | stats timeout <timeout, in milliseconds> |
| 901 | The default timeout on the stats socket is set to 10 seconds. It is possible |
| 902 | to change this value with "stats timeout". The value must be passed in |
Willy Tarreau | befdff1 | 2007-12-02 22:27:38 +0100 | [diff] [blame] | 903 | milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }. |
Willy Tarreau | fbee713 | 2007-10-18 13:53:22 +0200 | [diff] [blame] | 904 | |
| 905 | stats maxconn <connections> |
| 906 | By default, the stats socket is limited to 10 concurrent connections. It is |
| 907 | possible to change this value with "stats maxconn". |
| 908 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 909 | uid <number> |
| 910 | Changes the process' user ID to <number>. It is recommended that the user ID |
| 911 | is dedicated to HAProxy or to a small set of similar daemons. HAProxy must |
| 912 | be started with superuser privileges in order to be able to switch to another |
| 913 | one. See also "gid" and "user". |
| 914 | |
| 915 | ulimit-n <number> |
| 916 | Sets the maximum number of per-process file-descriptors to <number>. By |
| 917 | default, it is automatically computed, so it is recommended not to use this |
| 918 | option. |
| 919 | |
Willy Tarreau | ceb24bc | 2010-11-09 12:46:41 +0100 | [diff] [blame] | 920 | unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ] |
| 921 | [ group <group> ] [ gid <gid> ] |
| 922 | |
| 923 | Fixes common settings to UNIX listening sockets declared in "bind" statements. |
| 924 | This is mainly used to simplify declaration of those UNIX sockets and reduce |
| 925 | the risk of errors, since those settings are most commonly required but are |
| 926 | also process-specific. The <prefix> setting can be used to force all socket |
| 927 | path to be relative to that directory. This might be needed to access another |
| 928 | component's chroot. Note that those paths are resolved before haproxy chroots |
| 929 | itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid> |
| 930 | all have the same meaning as their homonyms used by the "bind" statement. If |
| 931 | both are specified, the "bind" statement has priority, meaning that the |
| 932 | "unix-bind" settings may be seen as process-wide default settings. |
| 933 | |
Willy Tarreau | 1d54972 | 2016-02-16 12:41:57 +0100 | [diff] [blame] | 934 | unsetenv [<name> ...] |
| 935 | Removes environment variables specified in arguments. This can be useful to |
| 936 | hide some sensitive information that are occasionally inherited from the |
| 937 | user's environment during some operations. Variables which did not exist are |
| 938 | silently ignored so that after the operation, it is certain that none of |
| 939 | these variables remain. The changes immediately take effect so that the next |
| 940 | line in the configuration file will not see these variables. See also |
| 941 | "setenv", "presetenv", and "resetenv". |
| 942 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 943 | user <user name> |
| 944 | Similar to "uid" but uses the UID of user name <user name> from /etc/passwd. |
| 945 | See also "uid" and "group". |
| 946 | |
Krzysztof Piotr Oledzki | 48cb2ae | 2009-10-02 22:51:14 +0200 | [diff] [blame] | 947 | node <name> |
| 948 | Only letters, digits, hyphen and underscore are allowed, like in DNS names. |
| 949 | |
| 950 | This statement is useful in HA configurations where two or more processes or |
| 951 | servers share the same IP address. By setting a different node-name on all |
| 952 | nodes, it becomes easy to immediately spot what server is handling the |
| 953 | traffic. |
| 954 | |
| 955 | description <text> |
| 956 | Add a text that describes the instance. |
| 957 | |
| 958 | Please note that it is required to escape certain characters (# for example) |
| 959 | and this text is inserted into a html page so you should avoid using |
| 960 | "<" and ">" characters. |
| 961 | |
Thomas Holmes | db04f19 | 2015-05-18 13:21:39 +0100 | [diff] [blame] | 962 | 51degrees-data-file <file path> |
| 963 | The path of the 51Degrees data file to provide device detection services. The |
| 964 | file should be unzipped and accessible by HAProxy with relevavnt permissions. |
| 965 | |
Dragan Dosen | ae6d39a | 2015-06-29 16:43:27 +0200 | [diff] [blame] | 966 | Please note that this option is only available when haproxy has been |
Thomas Holmes | db04f19 | 2015-05-18 13:21:39 +0100 | [diff] [blame] | 967 | compiled with USE_51DEGREES. |
| 968 | |
| 969 | 51degrees-property-name-list [<string>] |
| 970 | A list of 51Degrees property names to be load from the dataset. A full list |
| 971 | of names is available on the 51Degrees website: |
| 972 | https://51degrees.com/resources/property-dictionary |
| 973 | |
Dragan Dosen | ae6d39a | 2015-06-29 16:43:27 +0200 | [diff] [blame] | 974 | Please note that this option is only available when haproxy has been |
Thomas Holmes | db04f19 | 2015-05-18 13:21:39 +0100 | [diff] [blame] | 975 | compiled with USE_51DEGREES. |
| 976 | |
Dragan Dosen | 93b38d9 | 2015-06-29 16:43:25 +0200 | [diff] [blame] | 977 | 51degrees-property-separator <char> |
Thomas Holmes | db04f19 | 2015-05-18 13:21:39 +0100 | [diff] [blame] | 978 | A char that will be appended to every property value in a response header |
| 979 | containing 51Degrees results. If not set that will be set as ','. |
| 980 | |
Dragan Dosen | ae6d39a | 2015-06-29 16:43:27 +0200 | [diff] [blame] | 981 | Please note that this option is only available when haproxy has been |
| 982 | compiled with USE_51DEGREES. |
| 983 | |
| 984 | 51degrees-cache-size <number> |
| 985 | Sets the size of the 51Degrees converter cache to <number> entries. This |
| 986 | is an LRU cache which reminds previous device detections and their results. |
| 987 | By default, this cache is disabled. |
| 988 | |
| 989 | Please note that this option is only available when haproxy has been |
Thomas Holmes | db04f19 | 2015-05-18 13:21:39 +0100 | [diff] [blame] | 990 | compiled with USE_51DEGREES. |
| 991 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 992 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 993 | 3.2. Performance tuning |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 994 | ----------------------- |
| 995 | |
Willy Tarreau | 1746eec | 2014-04-25 10:46:47 +0200 | [diff] [blame] | 996 | max-spread-checks <delay in milliseconds> |
| 997 | By default, haproxy tries to spread the start of health checks across the |
| 998 | smallest health check interval of all the servers in a farm. The principle is |
| 999 | to avoid hammering services running on the same server. But when using large |
| 1000 | check intervals (10 seconds or more), the last servers in the farm take some |
| 1001 | time before starting to be tested, which can be a problem. This parameter is |
| 1002 | used to enforce an upper bound on delay between the first and the last check, |
| 1003 | even if the servers' check intervals are larger. When servers run with |
| 1004 | shorter intervals, their intervals will be respected though. |
| 1005 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 1006 | maxconn <number> |
| 1007 | Sets the maximum per-process number of concurrent connections to <number>. It |
| 1008 | is equivalent to the command-line argument "-n". Proxies will stop accepting |
| 1009 | connections when this limit is reached. The "ulimit-n" parameter is |
Willy Tarreau | 8274e10 | 2014-06-19 15:31:25 +0200 | [diff] [blame] | 1010 | automatically adjusted according to this value. See also "ulimit-n". Note: |
| 1011 | the "select" poller cannot reliably use more than 1024 file descriptors on |
| 1012 | some platforms. If your platform only supports select and reports "select |
| 1013 | FAILED" on startup, you need to reduce maxconn until it works (slightly |
Willy Tarreau | d025648 | 2015-01-15 21:45:22 +0100 | [diff] [blame] | 1014 | below 500 in general). If this value is not set, it will default to the value |
| 1015 | set in DEFAULT_MAXCONN at build time (reported in haproxy -vv) if no memory |
| 1016 | limit is enforced, or will be computed based on the memory limit, the buffer |
| 1017 | size, memory allocated to compression, SSL cache size, and use or not of SSL |
| 1018 | and the associated maxsslconn (which can also be automatic). |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 1019 | |
Willy Tarreau | 81c25d0 | 2011-09-07 15:17:21 +0200 | [diff] [blame] | 1020 | maxconnrate <number> |
| 1021 | Sets the maximum per-process number of connections per second to <number>. |
| 1022 | Proxies will stop accepting connections when this limit is reached. It can be |
| 1023 | used to limit the global capacity regardless of each frontend capacity. It is |
| 1024 | important to note that this can only be used as a service protection measure, |
| 1025 | as there will not necessarily be a fair share between frontends when the |
| 1026 | limit is reached, so it's a good idea to also limit each frontend to some |
| 1027 | value close to its expected share. Also, lowering tune.maxaccept can improve |
| 1028 | fairness. |
| 1029 | |
William Lallemand | d85f917 | 2012-11-09 17:05:39 +0100 | [diff] [blame] | 1030 | maxcomprate <number> |
| 1031 | Sets the maximum per-process input compression rate to <number> kilobytes |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 1032 | per second. For each session, if the maximum is reached, the compression |
William Lallemand | d85f917 | 2012-11-09 17:05:39 +0100 | [diff] [blame] | 1033 | level will be decreased during the session. If the maximum is reached at the |
| 1034 | beginning of a session, the session will not compress at all. If the maximum |
| 1035 | is not reached, the compression level will be increased up to |
| 1036 | tune.comp.maxlevel. A value of zero means there is no limit, this is the |
| 1037 | default value. |
| 1038 | |
William Lallemand | 072a2bf | 2012-11-20 17:01:01 +0100 | [diff] [blame] | 1039 | maxcompcpuusage <number> |
| 1040 | Sets the maximum CPU usage HAProxy can reach before stopping the compression |
| 1041 | for new requests or decreasing the compression level of current requests. |
| 1042 | It works like 'maxcomprate' but measures CPU usage instead of incoming data |
| 1043 | bandwidth. The value is expressed in percent of the CPU used by haproxy. In |
| 1044 | case of multiple processes (nbproc > 1), each process manages its individual |
| 1045 | usage. A value of 100 disable the limit. The default value is 100. Setting |
| 1046 | a lower value will prevent the compression work from slowing the whole |
| 1047 | process down and from introducing high latencies. |
| 1048 | |
Willy Tarreau | ff4f82d | 2009-02-06 11:28:13 +0100 | [diff] [blame] | 1049 | maxpipes <number> |
| 1050 | Sets the maximum per-process number of pipes to <number>. Currently, pipes |
| 1051 | are only used by kernel-based tcp splicing. Since a pipe contains two file |
| 1052 | descriptors, the "ulimit-n" value will be increased accordingly. The default |
| 1053 | value is maxconn/4, which seems to be more than enough for most heavy usages. |
| 1054 | The splice code dynamically allocates and releases pipes, and can fall back |
| 1055 | to standard copy, so setting this value too low may only impact performance. |
| 1056 | |
Willy Tarreau | 93e7c00 | 2013-10-07 18:51:07 +0200 | [diff] [blame] | 1057 | maxsessrate <number> |
| 1058 | Sets the maximum per-process number of sessions per second to <number>. |
| 1059 | Proxies will stop accepting connections when this limit is reached. It can be |
| 1060 | used to limit the global capacity regardless of each frontend capacity. It is |
| 1061 | important to note that this can only be used as a service protection measure, |
| 1062 | as there will not necessarily be a fair share between frontends when the |
| 1063 | limit is reached, so it's a good idea to also limit each frontend to some |
| 1064 | value close to its expected share. Also, lowering tune.maxaccept can improve |
| 1065 | fairness. |
| 1066 | |
Willy Tarreau | 403edff | 2012-09-06 11:58:37 +0200 | [diff] [blame] | 1067 | maxsslconn <number> |
| 1068 | Sets the maximum per-process number of concurrent SSL connections to |
| 1069 | <number>. By default there is no SSL-specific limit, which means that the |
| 1070 | global maxconn setting will apply to all connections. Setting this limit |
| 1071 | avoids having openssl use too much memory and crash when malloc returns NULL |
| 1072 | (since it unfortunately does not reliably check for such conditions). Note |
| 1073 | that the limit applies both to incoming and outgoing connections, so one |
| 1074 | connection which is deciphered then ciphered accounts for 2 SSL connections. |
Willy Tarreau | d025648 | 2015-01-15 21:45:22 +0100 | [diff] [blame] | 1075 | If this value is not set, but a memory limit is enforced, this value will be |
| 1076 | automatically computed based on the memory limit, maxconn, the buffer size, |
| 1077 | memory allocated to compression, SSL cache size, and use of SSL in either |
| 1078 | frontends, backends or both. If neither maxconn nor maxsslconn are specified |
| 1079 | when there is a memory limit, haproxy will automatically adjust these values |
| 1080 | so that 100% of the connections can be made over SSL with no risk, and will |
| 1081 | consider the sides where it is enabled (frontend, backend, both). |
Willy Tarreau | 403edff | 2012-09-06 11:58:37 +0200 | [diff] [blame] | 1082 | |
Willy Tarreau | e43d532 | 2013-10-07 20:01:52 +0200 | [diff] [blame] | 1083 | maxsslrate <number> |
| 1084 | Sets the maximum per-process number of SSL sessions per second to <number>. |
| 1085 | SSL listeners will stop accepting connections when this limit is reached. It |
| 1086 | can be used to limit the global SSL CPU usage regardless of each frontend |
| 1087 | capacity. It is important to note that this can only be used as a service |
| 1088 | protection measure, as there will not necessarily be a fair share between |
| 1089 | frontends when the limit is reached, so it's a good idea to also limit each |
| 1090 | frontend to some value close to its expected share. It is also important to |
| 1091 | note that the sessions are accounted before they enter the SSL stack and not |
| 1092 | after, which also protects the stack against bad handshakes. Also, lowering |
| 1093 | tune.maxaccept can improve fairness. |
| 1094 | |
William Lallemand | 9d5f548 | 2012-11-07 16:12:57 +0100 | [diff] [blame] | 1095 | maxzlibmem <number> |
| 1096 | Sets the maximum amount of RAM in megabytes per process usable by the zlib. |
| 1097 | When the maximum amount is reached, future sessions will not compress as long |
| 1098 | as RAM is unavailable. When sets to 0, there is no limit. |
William Lallemand | e3a7d99 | 2012-11-20 11:25:20 +0100 | [diff] [blame] | 1099 | The default value is 0. The value is available in bytes on the UNIX socket |
| 1100 | with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is |
| 1101 | "ZlibMemUsage" in bytes. |
| 1102 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 1103 | noepoll |
| 1104 | Disables the use of the "epoll" event polling system on Linux. It is |
| 1105 | equivalent to the command-line argument "-de". The next polling system |
Willy Tarreau | e9f49e7 | 2012-11-11 17:42:00 +0100 | [diff] [blame] | 1106 | used will generally be "poll". See also "nopoll". |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 1107 | |
| 1108 | nokqueue |
| 1109 | Disables the use of the "kqueue" event polling system on BSD. It is |
| 1110 | equivalent to the command-line argument "-dk". The next polling system |
| 1111 | used will generally be "poll". See also "nopoll". |
| 1112 | |
| 1113 | nopoll |
| 1114 | Disables the use of the "poll" event polling system. It is equivalent to the |
| 1115 | command-line argument "-dp". The next polling system used will be "select". |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1116 | It should never be needed to disable "poll" since it's available on all |
Willy Tarreau | e9f49e7 | 2012-11-11 17:42:00 +0100 | [diff] [blame] | 1117 | platforms supported by HAProxy. See also "nokqueue" and "noepoll". |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 1118 | |
Willy Tarreau | ff4f82d | 2009-02-06 11:28:13 +0100 | [diff] [blame] | 1119 | nosplice |
| 1120 | Disables the use of kernel tcp splicing between sockets on Linux. It is |
| 1121 | equivalent to the command line argument "-dS". Data will then be copied |
| 1122 | using conventional and more portable recv/send calls. Kernel tcp splicing is |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 1123 | limited to some very recent instances of kernel 2.6. Most versions between |
Willy Tarreau | ff4f82d | 2009-02-06 11:28:13 +0100 | [diff] [blame] | 1124 | 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not |
| 1125 | be used. This option makes it easier to globally disable kernel splicing in |
| 1126 | case of doubt. See also "option splice-auto", "option splice-request" and |
| 1127 | "option splice-response". |
| 1128 | |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 1129 | nogetaddrinfo |
| 1130 | Disables the use of getaddrinfo(3) for name resolving. It is equivalent to |
| 1131 | the command line argument "-dG". Deprecated gethostbyname(3) will be used. |
| 1132 | |
Willy Tarreau | fe255b7 | 2007-10-14 23:09:26 +0200 | [diff] [blame] | 1133 | spread-checks <0..50, in percent> |
Simon Horman | d60d691 | 2013-11-25 10:46:36 +0900 | [diff] [blame] | 1134 | Sometimes it is desirable to avoid sending agent and health checks to |
| 1135 | servers at exact intervals, for instance when many logical servers are |
| 1136 | located on the same physical server. With the help of this parameter, it |
| 1137 | becomes possible to add some randomness in the check interval between 0 |
| 1138 | and +/- 50%. A value between 2 and 5 seems to show good results. The |
| 1139 | default value remains at 0. |
Willy Tarreau | fe255b7 | 2007-10-14 23:09:26 +0200 | [diff] [blame] | 1140 | |
Willy Tarreau | 33cb065 | 2014-12-23 22:52:37 +0100 | [diff] [blame] | 1141 | tune.buffers.limit <number> |
| 1142 | Sets a hard limit on the number of buffers which may be allocated per process. |
| 1143 | The default value is zero which means unlimited. The minimum non-zero value |
| 1144 | will always be greater than "tune.buffers.reserve" and should ideally always |
| 1145 | be about twice as large. Forcing this value can be particularly useful to |
| 1146 | limit the amount of memory a process may take, while retaining a sane |
| 1147 | behaviour. When this limit is reached, sessions which need a buffer wait for |
| 1148 | another one to be released by another session. Since buffers are dynamically |
| 1149 | allocated and released, the waiting time is very short and not perceptible |
| 1150 | provided that limits remain reasonable. In fact sometimes reducing the limit |
| 1151 | may even increase performance by increasing the CPU cache's efficiency. Tests |
| 1152 | have shown good results on average HTTP traffic with a limit to 1/10 of the |
| 1153 | expected global maxconn setting, which also significantly reduces memory |
| 1154 | usage. The memory savings come from the fact that a number of connections |
| 1155 | will not allocate 2*tune.bufsize. It is best not to touch this value unless |
| 1156 | advised to do so by an haproxy core developer. |
| 1157 | |
Willy Tarreau | 1058ae7 | 2014-12-23 22:40:40 +0100 | [diff] [blame] | 1158 | tune.buffers.reserve <number> |
| 1159 | Sets the number of buffers which are pre-allocated and reserved for use only |
| 1160 | during memory shortage conditions resulting in failed memory allocations. The |
| 1161 | minimum value is 2 and is also the default. There is no reason a user would |
| 1162 | want to change this value, it's mostly aimed at haproxy core developers. |
| 1163 | |
Willy Tarreau | 27a674e | 2009-08-17 07:23:33 +0200 | [diff] [blame] | 1164 | tune.bufsize <number> |
| 1165 | Sets the buffer size to this size (in bytes). Lower values allow more |
| 1166 | sessions to coexist in the same amount of RAM, and higher values allow some |
| 1167 | applications with very large cookies to work. The default value is 16384 and |
| 1168 | can be changed at build time. It is strongly recommended not to change this |
| 1169 | from the default value, as very low values will break some services such as |
| 1170 | statistics, and values larger than default size will increase memory usage, |
| 1171 | possibly causing the system to run out of memory. At least the global maxconn |
| 1172 | parameter should be decreased by the same factor as this one is increased. |
Dmitry Sivachenko | f6f4f7b | 2012-10-21 18:10:25 +0400 | [diff] [blame] | 1173 | If HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will |
| 1174 | return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger |
| 1175 | than this size, haproxy will return HTTP 502 (Bad Gateway). |
Willy Tarreau | 27a674e | 2009-08-17 07:23:33 +0200 | [diff] [blame] | 1176 | |
Willy Tarreau | 43961d5 | 2010-10-04 20:39:20 +0200 | [diff] [blame] | 1177 | tune.chksize <number> |
| 1178 | Sets the check buffer size to this size (in bytes). Higher values may help |
| 1179 | find string or regex patterns in very large pages, though doing so may imply |
| 1180 | more memory and CPU usage. The default value is 16384 and can be changed at |
| 1181 | build time. It is not recommended to change this value, but to use better |
| 1182 | checks whenever possible. |
| 1183 | |
William Lallemand | f374783 | 2012-11-09 12:33:10 +0100 | [diff] [blame] | 1184 | tune.comp.maxlevel <number> |
| 1185 | Sets the maximum compression level. The compression level affects CPU |
| 1186 | usage during compression. This value affects CPU usage during compression. |
| 1187 | Each session using compression initializes the compression algorithm with |
| 1188 | this value. The default value is 1. |
| 1189 | |
Willy Tarreau | 193b8c6 | 2012-11-22 00:17:38 +0100 | [diff] [blame] | 1190 | tune.http.cookielen <number> |
| 1191 | Sets the maximum length of captured cookies. This is the maximum value that |
| 1192 | the "capture cookie xxx len yyy" will be allowed to take, and any upper value |
| 1193 | will automatically be truncated to this one. It is important not to set too |
| 1194 | high a value because all cookie captures still allocate this size whatever |
| 1195 | their configured value (they share a same pool). This value is per request |
| 1196 | per response, so the memory allocated is twice this value per connection. |
| 1197 | When not specified, the limit is set to 63 characters. It is recommended not |
| 1198 | to change this value. |
| 1199 | |
Willy Tarreau | ac1932d | 2011-10-24 19:14:41 +0200 | [diff] [blame] | 1200 | tune.http.maxhdr <number> |
| 1201 | Sets the maximum number of headers in a request. When a request comes with a |
| 1202 | number of headers greater than this value (including the first line), it is |
| 1203 | rejected with a "400 Bad Request" status code. Similarly, too large responses |
| 1204 | are blocked with "502 Bad Gateway". The default value is 101, which is enough |
| 1205 | for all usages, considering that the widely deployed Apache server uses the |
| 1206 | same limit. It can be useful to push this limit further to temporarily allow |
| 1207 | a buggy application to work by the time it gets fixed. Keep in mind that each |
| 1208 | new header consumes 32bits of memory for each session, so don't push this |
| 1209 | limit too high. |
| 1210 | |
Willy Tarreau | 7e31273 | 2014-02-12 16:35:14 +0100 | [diff] [blame] | 1211 | tune.idletimer <timeout> |
| 1212 | Sets the duration after which haproxy will consider that an empty buffer is |
| 1213 | probably associated with an idle stream. This is used to optimally adjust |
| 1214 | some packet sizes while forwarding large and small data alternatively. The |
| 1215 | decision to use splice() or to send large buffers in SSL is modulated by this |
| 1216 | parameter. The value is in milliseconds between 0 and 65535. A value of zero |
| 1217 | means that haproxy will not try to detect idle streams. The default is 1000, |
| 1218 | which seems to correctly detect end user pauses (eg: read a page before |
| 1219 | clicking). There should be not reason for changing this value. Please check |
| 1220 | tune.ssl.maxrecord below. |
| 1221 | |
Thierry FOURNIER | 90da191 | 2015-03-05 11:17:06 +0100 | [diff] [blame] | 1222 | tune.lua.forced-yield <number> |
| 1223 | This directive forces the Lua engine to execute a yield each <number> of |
| 1224 | instructions executed. This permits interruptng a long script and allows the |
| 1225 | HAProxy scheduler to process other tasks like accepting connections or |
| 1226 | forwarding traffic. The default value is 10000 instructions. If HAProxy often |
| 1227 | executes some Lua code but more reactivity is required, this value can be |
| 1228 | lowered. If the Lua code is quite long and its result is absolutely required |
| 1229 | to process the data, the <number> can be increased. |
| 1230 | |
Willy Tarreau | 32f61e2 | 2015-03-18 17:54:59 +0100 | [diff] [blame] | 1231 | tune.lua.maxmem |
| 1232 | Sets the maximum amount of RAM in megabytes per process usable by Lua. By |
| 1233 | default it is zero which means unlimited. It is important to set a limit to |
| 1234 | ensure that a bug in a script will not result in the system running out of |
| 1235 | memory. |
| 1236 | |
Thierry FOURNIER | 90da191 | 2015-03-05 11:17:06 +0100 | [diff] [blame] | 1237 | tune.lua.session-timeout <timeout> |
| 1238 | This is the execution timeout for the Lua sessions. This is useful for |
Thierry FOURNIER | 7dd784b | 2015-10-01 14:49:33 +0200 | [diff] [blame] | 1239 | preventing infinite loops or spending too much time in Lua. This timeout |
| 1240 | counts only the pure Lua runtime. If the Lua does a sleep, the sleep is |
| 1241 | not taked in account. The default timeout is 4s. |
Thierry FOURNIER | 90da191 | 2015-03-05 11:17:06 +0100 | [diff] [blame] | 1242 | |
| 1243 | tune.lua.task-timeout <timeout> |
| 1244 | Purpose is the same as "tune.lua.session-timeout", but this timeout is |
| 1245 | dedicated to the tasks. By default, this timeout isn't set because a task may |
| 1246 | remain alive during of the lifetime of HAProxy. For example, a task used to |
| 1247 | check servers. |
| 1248 | |
Thierry FOURNIER | 7dd784b | 2015-10-01 14:49:33 +0200 | [diff] [blame] | 1249 | tune.lua.service-timeout <timeout> |
| 1250 | This is the execution timeout for the Lua services. This is useful for |
| 1251 | preventing infinite loops or spending too much time in Lua. This timeout |
| 1252 | counts only the pure Lua runtime. If the Lua does a sleep, the sleep is |
| 1253 | not taked in account. The default timeout is 4s. |
| 1254 | |
Willy Tarreau | a0250ba | 2008-01-06 11:22:57 +0100 | [diff] [blame] | 1255 | tune.maxaccept <number> |
Willy Tarreau | 16a2147 | 2012-11-19 12:39:59 +0100 | [diff] [blame] | 1256 | Sets the maximum number of consecutive connections a process may accept in a |
| 1257 | row before switching to other work. In single process mode, higher numbers |
| 1258 | give better performance at high connection rates. However in multi-process |
| 1259 | modes, keeping a bit of fairness between processes generally is better to |
| 1260 | increase performance. This value applies individually to each listener, so |
| 1261 | that the number of processes a listener is bound to is taken into account. |
| 1262 | This value defaults to 64. In multi-process mode, it is divided by twice |
| 1263 | the number of processes the listener is bound to. Setting this value to -1 |
| 1264 | completely disables the limitation. It should normally not be needed to tweak |
| 1265 | this value. |
Willy Tarreau | a0250ba | 2008-01-06 11:22:57 +0100 | [diff] [blame] | 1266 | |
| 1267 | tune.maxpollevents <number> |
| 1268 | Sets the maximum amount of events that can be processed at once in a call to |
| 1269 | the polling system. The default value is adapted to the operating system. It |
| 1270 | has been noticed that reducing it below 200 tends to slightly decrease |
| 1271 | latency at the expense of network bandwidth, and increasing it above 200 |
| 1272 | tends to trade latency for slightly increased bandwidth. |
| 1273 | |
Willy Tarreau | 27a674e | 2009-08-17 07:23:33 +0200 | [diff] [blame] | 1274 | tune.maxrewrite <number> |
| 1275 | Sets the reserved buffer space to this size in bytes. The reserved space is |
| 1276 | used for header rewriting or appending. The first reads on sockets will never |
| 1277 | fill more than bufsize-maxrewrite. Historically it has defaulted to half of |
| 1278 | bufsize, though that does not make much sense since there are rarely large |
| 1279 | numbers of headers to add. Setting it too high prevents processing of large |
| 1280 | requests or responses. Setting it too low prevents addition of new headers |
| 1281 | to already large requests or to POST requests. It is generally wise to set it |
| 1282 | to about 1024. It is automatically readjusted to half of bufsize if it is |
| 1283 | larger than that. This means you don't have to worry about it when changing |
| 1284 | bufsize. |
| 1285 | |
Willy Tarreau | f3045d2 | 2015-04-29 16:24:50 +0200 | [diff] [blame] | 1286 | tune.pattern.cache-size <number> |
| 1287 | Sets the size of the pattern lookup cache to <number> entries. This is an LRU |
| 1288 | cache which reminds previous lookups and their results. It is used by ACLs |
| 1289 | and maps on slow pattern lookups, namely the ones using the "sub", "reg", |
| 1290 | "dir", "dom", "end", "bin" match methods as well as the case-insensitive |
| 1291 | strings. It applies to pattern expressions which means that it will be able |
| 1292 | to memorize the result of a lookup among all the patterns specified on a |
| 1293 | configuration line (including all those loaded from files). It automatically |
| 1294 | invalidates entries which are updated using HTTP actions or on the CLI. The |
| 1295 | default cache size is set to 10000 entries, which limits its footprint to |
| 1296 | about 5 MB on 32-bit systems and 8 MB on 64-bit systems. There is a very low |
| 1297 | risk of collision in this cache, which is in the order of the size of the |
| 1298 | cache divided by 2^64. Typically, at 10000 requests per second with the |
| 1299 | default cache size of 10000 entries, there's 1% chance that a brute force |
| 1300 | attack could cause a single collision after 60 years, or 0.1% after 6 years. |
| 1301 | This is considered much lower than the risk of a memory corruption caused by |
| 1302 | aging components. If this is not acceptable, the cache can be disabled by |
| 1303 | setting this parameter to 0. |
| 1304 | |
Willy Tarreau | bd9a0a7 | 2011-10-23 21:14:29 +0200 | [diff] [blame] | 1305 | tune.pipesize <number> |
| 1306 | Sets the kernel pipe buffer size to this size (in bytes). By default, pipes |
| 1307 | are the default size for the system. But sometimes when using TCP splicing, |
| 1308 | it can improve performance to increase pipe sizes, especially if it is |
| 1309 | suspected that pipes are not filled and that many calls to splice() are |
| 1310 | performed. This has an impact on the kernel's memory footprint, so this must |
| 1311 | not be changed if impacts are not understood. |
| 1312 | |
Willy Tarreau | e803de2 | 2010-01-21 17:43:04 +0100 | [diff] [blame] | 1313 | tune.rcvbuf.client <number> |
| 1314 | tune.rcvbuf.server <number> |
| 1315 | Forces the kernel socket receive buffer size on the client or the server side |
| 1316 | to the specified value in bytes. This value applies to all TCP/HTTP frontends |
| 1317 | and backends. It should normally never be set, and the default size (0) lets |
| 1318 | the kernel autotune this value depending on the amount of available memory. |
| 1319 | However it can sometimes help to set it to very low values (eg: 4096) in |
| 1320 | order to save kernel memory by preventing it from buffering too large amounts |
| 1321 | of received data. Lower values will significantly increase CPU usage though. |
| 1322 | |
Willy Tarreau | b22fc30 | 2015-12-14 12:04:35 +0100 | [diff] [blame] | 1323 | tune.recv_enough <number> |
| 1324 | Haproxy uses some hints to detect that a short read indicates the end of the |
| 1325 | socket buffers. One of them is that a read returns more than <recv_enough> |
| 1326 | bytes, which defaults to 10136 (7 segments of 1448 each). This default value |
| 1327 | may be changed by this setting to better deal with workloads involving lots |
| 1328 | of short messages such as telnet or SSH sessions. |
| 1329 | |
Willy Tarreau | e803de2 | 2010-01-21 17:43:04 +0100 | [diff] [blame] | 1330 | tune.sndbuf.client <number> |
| 1331 | tune.sndbuf.server <number> |
| 1332 | Forces the kernel socket send buffer size on the client or the server side to |
| 1333 | the specified value in bytes. This value applies to all TCP/HTTP frontends |
| 1334 | and backends. It should normally never be set, and the default size (0) lets |
| 1335 | the kernel autotune this value depending on the amount of available memory. |
| 1336 | However it can sometimes help to set it to very low values (eg: 4096) in |
| 1337 | order to save kernel memory by preventing it from buffering too large amounts |
| 1338 | of received data. Lower values will significantly increase CPU usage though. |
| 1339 | Another use case is to prevent write timeouts with extremely slow clients due |
| 1340 | to the kernel waiting for a large part of the buffer to be read before |
| 1341 | notifying haproxy again. |
| 1342 | |
Willy Tarreau | 6ec58db | 2012-11-16 16:32:15 +0100 | [diff] [blame] | 1343 | tune.ssl.cachesize <number> |
Emeric Brun | af9619d | 2012-11-28 18:47:52 +0100 | [diff] [blame] | 1344 | Sets the size of the global SSL session cache, in a number of blocks. A block |
| 1345 | is large enough to contain an encoded session without peer certificate. |
| 1346 | An encoded session with peer certificate is stored in multiple blocks |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 1347 | depending on the size of the peer certificate. A block uses approximately |
Emeric Brun | af9619d | 2012-11-28 18:47:52 +0100 | [diff] [blame] | 1348 | 200 bytes of memory. The default value may be forced at build time, otherwise |
| 1349 | defaults to 20000. When the cache is full, the most idle entries are purged |
| 1350 | and reassigned. Higher values reduce the occurrence of such a purge, hence |
| 1351 | the number of CPU-intensive SSL handshakes by ensuring that all users keep |
| 1352 | their session as long as possible. All entries are pre-allocated upon startup |
Emeric Brun | 22890a1 | 2012-12-28 14:41:32 +0100 | [diff] [blame] | 1353 | and are shared between all processes if "nbproc" is greater than 1. Setting |
| 1354 | this value to 0 disables the SSL session cache. |
Willy Tarreau | 6ec58db | 2012-11-16 16:32:15 +0100 | [diff] [blame] | 1355 | |
Emeric Brun | 8dc6039 | 2014-05-09 13:52:00 +0200 | [diff] [blame] | 1356 | tune.ssl.force-private-cache |
| 1357 | This boolean disables SSL session cache sharing between all processes. It |
| 1358 | should normally not be used since it will force many renegotiations due to |
| 1359 | clients hitting a random process. But it may be required on some operating |
| 1360 | systems where none of the SSL cache synchronization method may be used. In |
| 1361 | this case, adding a first layer of hash-based load balancing before the SSL |
| 1362 | layer might limit the impact of the lack of session sharing. |
| 1363 | |
Emeric Brun | 4f65bff | 2012-11-16 15:11:00 +0100 | [diff] [blame] | 1364 | tune.ssl.lifetime <timeout> |
| 1365 | Sets how long a cached SSL session may remain valid. This time is expressed |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 1366 | in seconds and defaults to 300 (5 min). It is important to understand that it |
Emeric Brun | 4f65bff | 2012-11-16 15:11:00 +0100 | [diff] [blame] | 1367 | does not guarantee that sessions will last that long, because if the cache is |
| 1368 | full, the longest idle sessions will be purged despite their configured |
| 1369 | lifetime. The real usefulness of this setting is to prevent sessions from |
| 1370 | being used for too long. |
| 1371 | |
Willy Tarreau | bfd5946 | 2013-02-21 07:46:09 +0100 | [diff] [blame] | 1372 | tune.ssl.maxrecord <number> |
| 1373 | Sets the maximum amount of bytes passed to SSL_write() at a time. Default |
| 1374 | value 0 means there is no limit. Over SSL/TLS, the client can decipher the |
| 1375 | data only once it has received a full record. With large records, it means |
| 1376 | that clients might have to download up to 16kB of data before starting to |
| 1377 | process them. Limiting the value can improve page load times on browsers |
| 1378 | located over high latency or low bandwidth networks. It is suggested to find |
| 1379 | optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over |
| 1380 | Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled), |
| 1381 | keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and |
| 1382 | 2859 gave good results during tests. Use "strace -e trace=write" to find the |
Willy Tarreau | 7e31273 | 2014-02-12 16:35:14 +0100 | [diff] [blame] | 1383 | best value. Haproxy will automatically switch to this setting after an idle |
| 1384 | stream has been detected (see tune.idletimer above). |
Willy Tarreau | bfd5946 | 2013-02-21 07:46:09 +0100 | [diff] [blame] | 1385 | |
Remi Gacogne | f46cd6e | 2014-06-12 14:58:40 +0200 | [diff] [blame] | 1386 | tune.ssl.default-dh-param <number> |
| 1387 | Sets the maximum size of the Diffie-Hellman parameters used for generating |
| 1388 | the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The |
| 1389 | final size will try to match the size of the server's RSA (or DSA) key (e.g, |
| 1390 | a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed |
| 1391 | this maximum value. Default value if 1024. Only 1024 or higher values are |
| 1392 | allowed. Higher values will increase the CPU load, and values greater than |
| 1393 | 1024 bits are not supported by Java 7 and earlier clients. This value is not |
Remi Gacogne | 47783ef | 2015-05-29 15:53:22 +0200 | [diff] [blame] | 1394 | used if static Diffie-Hellman parameters are supplied either directly |
| 1395 | in the certificate file or by using the ssl-dh-param-file parameter. |
Remi Gacogne | f46cd6e | 2014-06-12 14:58:40 +0200 | [diff] [blame] | 1396 | |
Christopher Faulet | 31af49d | 2015-06-09 17:29:50 +0200 | [diff] [blame] | 1397 | tune.ssl.ssl-ctx-cache-size <number> |
| 1398 | Sets the size of the cache used to store generated certificates to <number> |
| 1399 | entries. This is a LRU cache. Because generating a SSL certificate |
| 1400 | dynamically is expensive, they are cached. The default cache size is set to |
| 1401 | 1000 entries. |
| 1402 | |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 1403 | tune.vars.global-max-size <size> |
| 1404 | tune.vars.reqres-max-size <size> |
| 1405 | tune.vars.sess-max-size <size> |
| 1406 | tune.vars.txn-max-size <size> |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 1407 | These four tunes help to manage the maximum amount of memory used by the |
| 1408 | variables system. "global" limits the overall amount of memory available |
| 1409 | for all scopes. "sess" limits the memory for the session scope, "txn" for |
| 1410 | the transaction scope, and "reqres" limits the memory for each request or |
| 1411 | response processing. |
| 1412 | Memory accounting is hierarchical, meaning more coarse grained limits |
| 1413 | include the finer grained ones: "sess" includes "txn", and "txn" includes |
| 1414 | "reqres". |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 1415 | |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 1416 | For example, when "tune.vars.sess-max-size" is limited to 100, |
| 1417 | "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed |
| 1418 | 100 either. If we create a variable "txn.var" that contains 100 bytes, |
| 1419 | all available space is consumed. |
| 1420 | Notice that exceeding the limits at runtime will not result in an error |
| 1421 | message, but values might be cut off or corrupted. So make sure to accurately |
| 1422 | plan for the amount of space needed to store all your variables. |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 1423 | |
William Lallemand | a509e4c | 2012-11-07 16:54:34 +0100 | [diff] [blame] | 1424 | tune.zlib.memlevel <number> |
| 1425 | Sets the memLevel parameter in zlib initialization for each session. It |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 1426 | defines how much memory should be allocated for the internal compression |
William Lallemand | a509e4c | 2012-11-07 16:54:34 +0100 | [diff] [blame] | 1427 | state. A value of 1 uses minimum memory but is slow and reduces compression |
| 1428 | ratio, a value of 9 uses maximum memory for optimal speed. Can be a value |
| 1429 | between 1 and 9. The default value is 8. |
| 1430 | |
| 1431 | tune.zlib.windowsize <number> |
| 1432 | Sets the window size (the size of the history buffer) as a parameter of the |
| 1433 | zlib initialization for each session. Larger values of this parameter result |
| 1434 | in better compression at the expense of memory usage. Can be a value between |
| 1435 | 8 and 15. The default value is 15. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 1436 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 1437 | 3.3. Debugging |
| 1438 | -------------- |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 1439 | |
| 1440 | debug |
| 1441 | Enables debug mode which dumps to stdout all exchanges, and disables forking |
| 1442 | into background. It is the equivalent of the command-line argument "-d". It |
| 1443 | should never be used in a production configuration since it may prevent full |
| 1444 | system startup. |
| 1445 | |
| 1446 | quiet |
| 1447 | Do not display any message during startup. It is equivalent to the command- |
| 1448 | line argument "-q". |
| 1449 | |
Emeric Brun | f099e79 | 2010-09-27 12:05:28 +0200 | [diff] [blame] | 1450 | |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 1451 | 3.4. Userlists |
| 1452 | -------------- |
| 1453 | It is possible to control access to frontend/backend/listen sections or to |
| 1454 | http stats by allowing only authenticated and authorized users. To do this, |
| 1455 | it is required to create at least one userlist and to define users. |
| 1456 | |
| 1457 | userlist <listname> |
Cyril Bonté | 78caf84 | 2010-03-10 22:41:43 +0100 | [diff] [blame] | 1458 | Creates new userlist with name <listname>. Many independent userlists can be |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 1459 | used to store authentication & authorization data for independent customers. |
| 1460 | |
| 1461 | group <groupname> [users <user>,<user>,(...)] |
Cyril Bonté | 78caf84 | 2010-03-10 22:41:43 +0100 | [diff] [blame] | 1462 | Adds group <groupname> to the current userlist. It is also possible to |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 1463 | attach users to this group by using a comma separated list of names |
| 1464 | proceeded by "users" keyword. |
| 1465 | |
Cyril Bonté | f0c6061 | 2010-02-06 14:44:47 +0100 | [diff] [blame] | 1466 | user <username> [password|insecure-password <password>] |
| 1467 | [groups <group>,<group>,(...)] |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 1468 | Adds user <username> to the current userlist. Both secure (encrypted) and |
| 1469 | insecure (unencrypted) passwords can be used. Encrypted passwords are |
Cyril Bonté | 78caf84 | 2010-03-10 22:41:43 +0100 | [diff] [blame] | 1470 | evaluated using the crypt(3) function so depending of the system's |
| 1471 | capabilities, different algorithms are supported. For example modern Glibc |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 1472 | based Linux system supports MD5, SHA-256, SHA-512 and of course classic, |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 1473 | DES-based method of encrypting passwords. |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 1474 | |
| 1475 | |
| 1476 | Example: |
Cyril Bonté | f0c6061 | 2010-02-06 14:44:47 +0100 | [diff] [blame] | 1477 | userlist L1 |
| 1478 | group G1 users tiger,scott |
| 1479 | group G2 users xdb,scott |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 1480 | |
Cyril Bonté | f0c6061 | 2010-02-06 14:44:47 +0100 | [diff] [blame] | 1481 | user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91 |
| 1482 | user scott insecure-password elgato |
| 1483 | user xdb insecure-password hello |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 1484 | |
Cyril Bonté | f0c6061 | 2010-02-06 14:44:47 +0100 | [diff] [blame] | 1485 | userlist L2 |
| 1486 | group G1 |
| 1487 | group G2 |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 1488 | |
Cyril Bonté | f0c6061 | 2010-02-06 14:44:47 +0100 | [diff] [blame] | 1489 | user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1 |
| 1490 | user scott insecure-password elgato groups G1,G2 |
| 1491 | user xdb insecure-password hello groups G2 |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 1492 | |
| 1493 | Please note that both lists are functionally identical. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 1494 | |
Emeric Brun | f099e79 | 2010-09-27 12:05:28 +0200 | [diff] [blame] | 1495 | |
| 1496 | 3.5. Peers |
Cyril Bonté | dc4d903 | 2012-04-08 21:57:39 +0200 | [diff] [blame] | 1497 | ---------- |
Emeric Brun | 9490095 | 2015-06-11 18:25:54 +0200 | [diff] [blame] | 1498 | It is possible to propagate entries of any data-types in stick-tables between |
| 1499 | several haproxy instances over TCP connections in a multi-master fashion. Each |
| 1500 | instance pushes its local updates and insertions to remote peers. The pushed |
| 1501 | values overwrite remote ones without aggregation. Interrupted exchanges are |
| 1502 | automatically detected and recovered from the last known point. |
| 1503 | In addition, during a soft restart, the old process connects to the new one |
| 1504 | using such a TCP connection to push all its entries before the new process |
| 1505 | tries to connect to other peers. That ensures very fast replication during a |
| 1506 | reload, it typically takes a fraction of a second even for large tables. |
| 1507 | Note that Server IDs are used to identify servers remotely, so it is important |
| 1508 | that configurations look similar or at least that the same IDs are forced on |
| 1509 | each server on all participants. |
Emeric Brun | f099e79 | 2010-09-27 12:05:28 +0200 | [diff] [blame] | 1510 | |
| 1511 | peers <peersect> |
Jamie Gloudon | 801a0a3 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 1512 | Creates a new peer list with name <peersect>. It is an independent section, |
Emeric Brun | f099e79 | 2010-09-27 12:05:28 +0200 | [diff] [blame] | 1513 | which is referenced by one or more stick-tables. |
| 1514 | |
Willy Tarreau | 77e4bd1 | 2015-05-01 20:02:17 +0200 | [diff] [blame] | 1515 | disabled |
| 1516 | Disables a peers section. It disables both listening and any synchronization |
| 1517 | related to this section. This is provided to disable synchronization of stick |
| 1518 | tables without having to comment out all "peers" references. |
| 1519 | |
| 1520 | enable |
| 1521 | This re-enables a disabled peers section which was previously disabled. |
| 1522 | |
Emeric Brun | f099e79 | 2010-09-27 12:05:28 +0200 | [diff] [blame] | 1523 | peer <peername> <ip>:<port> |
| 1524 | Defines a peer inside a peers section. |
| 1525 | If <peername> is set to the local peer name (by default hostname, or forced |
| 1526 | using "-L" command line option), haproxy will listen for incoming remote peer |
| 1527 | connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to |
| 1528 | to join the remote peer, and <peername> is used at the protocol level to |
| 1529 | identify and validate the remote peer on the server side. |
| 1530 | |
| 1531 | During a soft restart, local peer <ip>:<port> is used by the old instance to |
| 1532 | connect the new one and initiate a complete replication (teaching process). |
| 1533 | |
| 1534 | It is strongly recommended to have the exact same peers declaration on all |
| 1535 | peers and to only rely on the "-L" command line argument to change the local |
| 1536 | peer name. This makes it easier to maintain coherent configuration files |
| 1537 | across all peers. |
| 1538 | |
William Lallemand | b2f0745 | 2015-05-12 14:27:13 +0200 | [diff] [blame] | 1539 | You may want to reference some environment variables in the address |
| 1540 | parameter, see section 2.3 about environment variables. |
Willy Tarreau | dad36a3 | 2013-03-11 01:20:04 +0100 | [diff] [blame] | 1541 | |
Cyril Bonté | dc4d903 | 2012-04-08 21:57:39 +0200 | [diff] [blame] | 1542 | Example: |
Emeric Brun | f099e79 | 2010-09-27 12:05:28 +0200 | [diff] [blame] | 1543 | peers mypeers |
Willy Tarreau | f7b30a9 | 2010-12-06 22:59:17 +0100 | [diff] [blame] | 1544 | peer haproxy1 192.168.0.1:1024 |
| 1545 | peer haproxy2 192.168.0.2:1024 |
| 1546 | peer haproxy3 10.2.0.1:1024 |
Emeric Brun | f099e79 | 2010-09-27 12:05:28 +0200 | [diff] [blame] | 1547 | |
| 1548 | backend mybackend |
| 1549 | mode tcp |
| 1550 | balance roundrobin |
| 1551 | stick-table type ip size 20k peers mypeers |
| 1552 | stick on src |
| 1553 | |
Willy Tarreau | f7b30a9 | 2010-12-06 22:59:17 +0100 | [diff] [blame] | 1554 | server srv1 192.168.0.30:80 |
| 1555 | server srv2 192.168.0.31:80 |
Emeric Brun | f099e79 | 2010-09-27 12:05:28 +0200 | [diff] [blame] | 1556 | |
| 1557 | |
Simon Horman | 51a1cf6 | 2015-02-03 13:00:44 +0900 | [diff] [blame] | 1558 | 3.6. Mailers |
| 1559 | ------------ |
| 1560 | It is possible to send email alerts when the state of servers changes. |
| 1561 | If configured email alerts are sent to each mailer that is configured |
| 1562 | in a mailers section. Email is sent to mailers using SMTP. |
| 1563 | |
Pieter Baauw | 386a127 | 2015-08-16 15:26:24 +0200 | [diff] [blame] | 1564 | mailers <mailersect> |
Simon Horman | 51a1cf6 | 2015-02-03 13:00:44 +0900 | [diff] [blame] | 1565 | Creates a new mailer list with the name <mailersect>. It is an |
| 1566 | independent section which is referenced by one or more proxies. |
| 1567 | |
| 1568 | mailer <mailername> <ip>:<port> |
| 1569 | Defines a mailer inside a mailers section. |
| 1570 | |
| 1571 | Example: |
| 1572 | mailers mymailers |
| 1573 | mailer smtp1 192.168.0.1:587 |
| 1574 | mailer smtp2 192.168.0.2:587 |
| 1575 | |
| 1576 | backend mybackend |
| 1577 | mode tcp |
| 1578 | balance roundrobin |
| 1579 | |
| 1580 | email-alert mailers mymailers |
| 1581 | email-alert from test1@horms.org |
| 1582 | email-alert to test2@horms.org |
| 1583 | |
| 1584 | server srv1 192.168.0.30:80 |
| 1585 | server srv2 192.168.0.31:80 |
| 1586 | |
Pieter Baauw | 235fcfc | 2016-02-13 15:33:40 +0100 | [diff] [blame] | 1587 | timeout mail <time> |
| 1588 | Defines the time available for a mail/connection to be made and send to |
| 1589 | the mail-server. If not defined the default value is 10 seconds. To allow |
| 1590 | for at least two SYN-ACK packets to be send during initial TCP handshake it |
| 1591 | is advised to keep this value above 4 seconds. |
| 1592 | |
| 1593 | Example: |
| 1594 | mailers mymailers |
| 1595 | timeout mail 20s |
| 1596 | mailer smtp1 192.168.0.1:587 |
Simon Horman | 51a1cf6 | 2015-02-03 13:00:44 +0900 | [diff] [blame] | 1597 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 1598 | 4. Proxies |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 1599 | ---------- |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1600 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 1601 | Proxy configuration can be located in a set of sections : |
William Lallemand | 6e62fb6 | 2015-04-28 16:55:23 +0200 | [diff] [blame] | 1602 | - defaults [<name>] |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 1603 | - frontend <name> |
| 1604 | - backend <name> |
| 1605 | - listen <name> |
| 1606 | |
| 1607 | A "defaults" section sets default parameters for all other sections following |
| 1608 | its declaration. Those default parameters are reset by the next "defaults" |
| 1609 | section. See below for the list of parameters which can be set in a "defaults" |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1610 | section. The name is optional but its use is encouraged for better readability. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 1611 | |
| 1612 | A "frontend" section describes a set of listening sockets accepting client |
| 1613 | connections. |
| 1614 | |
| 1615 | A "backend" section describes a set of servers to which the proxy will connect |
| 1616 | to forward incoming connections. |
| 1617 | |
| 1618 | A "listen" section defines a complete proxy with its frontend and backend |
| 1619 | parts combined in one section. It is generally useful for TCP-only traffic. |
| 1620 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1621 | All proxy names must be formed from upper and lower case letters, digits, |
| 1622 | '-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are |
| 1623 | case-sensitive, which means that "www" and "WWW" are two different proxies. |
| 1624 | |
| 1625 | Historically, all proxy names could overlap, it just caused troubles in the |
| 1626 | logs. Since the introduction of content switching, it is mandatory that two |
| 1627 | proxies with overlapping capabilities (frontend/backend) have different names. |
| 1628 | However, it is still permitted that a frontend and a backend share the same |
| 1629 | name, as this configuration seems to be commonly encountered. |
| 1630 | |
| 1631 | Right now, two major proxy modes are supported : "tcp", also known as layer 4, |
| 1632 | and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 1633 | bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1634 | protocol, and can interact with it by allowing, blocking, switching, adding, |
| 1635 | modifying, or removing arbitrary contents in requests or responses, based on |
| 1636 | arbitrary criteria. |
| 1637 | |
Willy Tarreau | 70dffda | 2014-01-30 03:07:23 +0100 | [diff] [blame] | 1638 | In HTTP mode, the processing applied to requests and responses flowing over |
| 1639 | a connection depends in the combination of the frontend's HTTP options and |
| 1640 | the backend's. HAProxy supports 5 connection modes : |
| 1641 | |
| 1642 | - KAL : keep alive ("option http-keep-alive") which is the default mode : all |
| 1643 | requests and responses are processed, and connections remain open but idle |
| 1644 | between responses and new requests. |
| 1645 | |
| 1646 | - TUN: tunnel ("option http-tunnel") : this was the default mode for versions |
| 1647 | 1.0 to 1.5-dev21 : only the first request and response are processed, and |
| 1648 | everything else is forwarded with no analysis at all. This mode should not |
| 1649 | be used as it creates lots of trouble with logging and HTTP processing. |
| 1650 | |
| 1651 | - PCL: passive close ("option httpclose") : exactly the same as tunnel mode, |
| 1652 | but with "Connection: close" appended in both directions to try to make |
| 1653 | both ends close after the first request/response exchange. |
| 1654 | |
| 1655 | - SCL: server close ("option http-server-close") : the server-facing |
| 1656 | connection is closed after the end of the response is received, but the |
| 1657 | client-facing connection remains open. |
| 1658 | |
| 1659 | - FCL: forced close ("option forceclose") : the connection is actively closed |
| 1660 | after the end of the response. |
| 1661 | |
| 1662 | The effective mode that will be applied to a connection passing through a |
| 1663 | frontend and a backend can be determined by both proxy modes according to the |
| 1664 | following matrix, but in short, the modes are symmetric, keep-alive is the |
| 1665 | weakest option and force close is the strongest. |
| 1666 | |
| 1667 | Backend mode |
| 1668 | |
| 1669 | | KAL | TUN | PCL | SCL | FCL |
| 1670 | ----+-----+-----+-----+-----+---- |
| 1671 | KAL | KAL | TUN | PCL | SCL | FCL |
| 1672 | ----+-----+-----+-----+-----+---- |
| 1673 | TUN | TUN | TUN | PCL | SCL | FCL |
| 1674 | Frontend ----+-----+-----+-----+-----+---- |
| 1675 | mode PCL | PCL | PCL | PCL | FCL | FCL |
| 1676 | ----+-----+-----+-----+-----+---- |
| 1677 | SCL | SCL | SCL | FCL | SCL | FCL |
| 1678 | ----+-----+-----+-----+-----+---- |
| 1679 | FCL | FCL | FCL | FCL | FCL | FCL |
| 1680 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1681 | |
Willy Tarreau | 70dffda | 2014-01-30 03:07:23 +0100 | [diff] [blame] | 1682 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 1683 | 4.1. Proxy keywords matrix |
| 1684 | -------------------------- |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1685 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 1686 | The following list of keywords is supported. Most of them may only be used in a |
| 1687 | limited set of section types. Some of them are marked as "deprecated" because |
| 1688 | they are inherited from an old syntax which may be confusing or functionally |
| 1689 | limited, and there are new recommended keywords to replace them. Keywords |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1690 | marked with "(*)" can be optionally inverted using the "no" prefix, eg. "no |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 1691 | option contstats". This makes sense when the option has been enabled by default |
Willy Tarreau | 3842f00 | 2009-06-14 11:39:52 +0200 | [diff] [blame] | 1692 | and must be disabled for a specific instance. Such options may also be prefixed |
| 1693 | with "default" in order to restore default settings regardless of what has been |
| 1694 | specified in a previous "defaults" section. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1695 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 1696 | |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1697 | keyword defaults frontend listen backend |
| 1698 | ------------------------------------+----------+----------+---------+--------- |
| 1699 | acl - X X X |
Willy Tarreau | 294d0f0 | 2015-08-10 19:40:12 +0200 | [diff] [blame] | 1700 | appsession - - - - |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1701 | backlog X X X - |
| 1702 | balance X - X X |
| 1703 | bind - X X - |
| 1704 | bind-process X X X X |
| 1705 | block - X X X |
| 1706 | capture cookie - X X - |
| 1707 | capture request header - X X - |
| 1708 | capture response header - X X - |
| 1709 | clitimeout (deprecated) X X X - |
William Lallemand | 82fe75c | 2012-10-23 10:25:10 +0200 | [diff] [blame] | 1710 | compression X X X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1711 | contimeout (deprecated) X - X X |
| 1712 | cookie X - X X |
Thierry FOURNIER | a0a1b75 | 2015-05-26 17:44:32 +0200 | [diff] [blame] | 1713 | declare capture - X X - |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1714 | default-server X - X X |
| 1715 | default_backend X X X - |
| 1716 | description - X X X |
| 1717 | disabled X X X X |
| 1718 | dispatch - - X X |
Simon Horman | 51a1cf6 | 2015-02-03 13:00:44 +0900 | [diff] [blame] | 1719 | email-alert from X X X X |
Simon Horman | 64e3416 | 2015-02-06 11:11:57 +0900 | [diff] [blame] | 1720 | email-alert level X X X X |
Simon Horman | 51a1cf6 | 2015-02-03 13:00:44 +0900 | [diff] [blame] | 1721 | email-alert mailers X X X X |
| 1722 | email-alert myhostname X X X X |
| 1723 | email-alert to X X X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1724 | enabled X X X X |
| 1725 | errorfile X X X X |
| 1726 | errorloc X X X X |
| 1727 | errorloc302 X X X X |
| 1728 | -- keyword -------------------------- defaults - frontend - listen -- backend - |
| 1729 | errorloc303 X X X X |
Cyril Bonté | 0d4bf01 | 2010-04-25 23:21:46 +0200 | [diff] [blame] | 1730 | force-persist - X X X |
Christopher Faulet | c3fe533 | 2016-04-07 15:30:10 +0200 | [diff] [blame] | 1731 | filter - X X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1732 | fullconn X - X X |
| 1733 | grace X X X X |
| 1734 | hash-type X - X X |
| 1735 | http-check disable-on-404 X - X X |
Willy Tarreau | bd74154 | 2010-03-16 18:46:54 +0100 | [diff] [blame] | 1736 | http-check expect - - X X |
Willy Tarreau | 7ab6aff | 2010-10-12 06:30:16 +0200 | [diff] [blame] | 1737 | http-check send-state X - X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1738 | http-request - X X X |
Willy Tarreau | e365c0b | 2013-06-11 16:06:12 +0200 | [diff] [blame] | 1739 | http-response - X X X |
Willy Tarreau | 3063195 | 2015-08-06 15:05:24 +0200 | [diff] [blame] | 1740 | http-reuse X - X X |
Baptiste Assmann | 2c42ef5 | 2013-10-09 21:57:02 +0200 | [diff] [blame] | 1741 | http-send-name-header - - X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1742 | id - X X X |
Cyril Bonté | 0d4bf01 | 2010-04-25 23:21:46 +0200 | [diff] [blame] | 1743 | ignore-persist - X X X |
Baptiste Assmann | 01c6cc3 | 2015-08-23 11:45:29 +0200 | [diff] [blame] | 1744 | load-server-state-from-file X - X X |
William Lallemand | 0f99e34 | 2011-10-12 17:50:54 +0200 | [diff] [blame] | 1745 | log (*) X X X X |
Willy Tarreau | fb4e7ea | 2015-01-07 14:55:17 +0100 | [diff] [blame] | 1746 | log-format X X X - |
Dragan Dosen | 7ad3154 | 2015-09-28 17:16:47 +0200 | [diff] [blame] | 1747 | log-format-sd X X X - |
Willy Tarreau | 094af4e | 2015-01-07 15:03:42 +0100 | [diff] [blame] | 1748 | log-tag X X X X |
Willy Tarreau | c35362a | 2014-04-25 13:58:37 +0200 | [diff] [blame] | 1749 | max-keep-alive-queue X - X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1750 | maxconn X X X - |
| 1751 | mode X X X X |
| 1752 | monitor fail - X X - |
| 1753 | monitor-net X X X - |
| 1754 | monitor-uri X X X - |
| 1755 | option abortonclose (*) X - X X |
| 1756 | option accept-invalid-http-request (*) X X X - |
| 1757 | option accept-invalid-http-response (*) X - X X |
| 1758 | option allbackups (*) X - X X |
| 1759 | option checkcache (*) X - X X |
| 1760 | option clitcpka (*) X X X - |
| 1761 | option contstats (*) X X X - |
| 1762 | option dontlog-normal (*) X X X - |
| 1763 | option dontlognull (*) X X X - |
| 1764 | option forceclose (*) X X X X |
| 1765 | -- keyword -------------------------- defaults - frontend - listen -- backend - |
| 1766 | option forwardfor X X X X |
Willy Tarreau | 9fbe18e | 2015-05-01 22:42:08 +0200 | [diff] [blame] | 1767 | option http-buffer-request (*) X X X X |
Willy Tarreau | 82649f9 | 2015-05-01 22:40:51 +0200 | [diff] [blame] | 1768 | option http-ignore-probes (*) X X X - |
Willy Tarreau | 16bfb02 | 2010-01-16 19:48:41 +0100 | [diff] [blame] | 1769 | option http-keep-alive (*) X X X X |
Willy Tarreau | 96e3121 | 2011-05-30 18:10:30 +0200 | [diff] [blame] | 1770 | option http-no-delay (*) X X X X |
Willy Tarreau | 8a8e1d9 | 2010-04-05 16:15:16 +0200 | [diff] [blame] | 1771 | option http-pretend-keepalive (*) X X X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1772 | option http-server-close (*) X X X X |
Willy Tarreau | 02bce8b | 2014-01-30 00:15:28 +0100 | [diff] [blame] | 1773 | option http-tunnel (*) X X X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1774 | option http-use-proxy-header (*) X X X - |
| 1775 | option httpchk X - X X |
| 1776 | option httpclose (*) X X X X |
| 1777 | option httplog X X X X |
| 1778 | option http_proxy (*) X X X X |
Jamie Gloudon | 801a0a3 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 1779 | option independent-streams (*) X X X X |
Gabor Lekeny | b4c81e4 | 2010-09-29 18:17:05 +0200 | [diff] [blame] | 1780 | option ldap-check X - X X |
Simon Horman | 98637e5 | 2014-06-20 12:30:16 +0900 | [diff] [blame] | 1781 | option external-check X - X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1782 | option log-health-checks (*) X - X X |
| 1783 | option log-separate-errors (*) X X X - |
| 1784 | option logasap (*) X X X - |
| 1785 | option mysql-check X - X X |
| 1786 | option nolinger (*) X X X X |
| 1787 | option originalto X X X X |
| 1788 | option persist (*) X - X X |
Baptiste Assmann | 809e22a | 2015-10-12 20:22:55 +0200 | [diff] [blame] | 1789 | option pgsql-check X - X X |
| 1790 | option prefer-last-server (*) X - X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1791 | option redispatch (*) X - X X |
Hervé COMMOWICK | ec032d6 | 2011-08-05 16:23:48 +0200 | [diff] [blame] | 1792 | option redis-check X - X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1793 | option smtpchk X - X X |
| 1794 | option socket-stats (*) X X X - |
| 1795 | option splice-auto (*) X X X X |
| 1796 | option splice-request (*) X X X X |
| 1797 | option splice-response (*) X X X X |
| 1798 | option srvtcpka (*) X - X X |
| 1799 | option ssl-hello-chk X - X X |
| 1800 | -- keyword -------------------------- defaults - frontend - listen -- backend - |
Willy Tarreau | ed17985 | 2013-12-16 01:07:00 +0100 | [diff] [blame] | 1801 | option tcp-check X - X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1802 | option tcp-smart-accept (*) X X X - |
| 1803 | option tcp-smart-connect (*) X - X X |
| 1804 | option tcpka X X X X |
| 1805 | option tcplog X X X X |
| 1806 | option transparent (*) X - X X |
Simon Horman | 98637e5 | 2014-06-20 12:30:16 +0900 | [diff] [blame] | 1807 | external-check command X - X X |
| 1808 | external-check path X - X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1809 | persist rdp-cookie X - X X |
| 1810 | rate-limit sessions X X X - |
| 1811 | redirect - X X X |
| 1812 | redisp (deprecated) X - X X |
| 1813 | redispatch (deprecated) X - X X |
| 1814 | reqadd - X X X |
| 1815 | reqallow - X X X |
| 1816 | reqdel - X X X |
| 1817 | reqdeny - X X X |
| 1818 | reqiallow - X X X |
| 1819 | reqidel - X X X |
| 1820 | reqideny - X X X |
| 1821 | reqipass - X X X |
| 1822 | reqirep - X X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1823 | reqitarpit - X X X |
| 1824 | reqpass - X X X |
| 1825 | reqrep - X X X |
| 1826 | -- keyword -------------------------- defaults - frontend - listen -- backend - |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1827 | reqtarpit - X X X |
| 1828 | retries X - X X |
| 1829 | rspadd - X X X |
| 1830 | rspdel - X X X |
| 1831 | rspdeny - X X X |
| 1832 | rspidel - X X X |
| 1833 | rspideny - X X X |
| 1834 | rspirep - X X X |
| 1835 | rsprep - X X X |
| 1836 | server - - X X |
Baptiste Assmann | 01c6cc3 | 2015-08-23 11:45:29 +0200 | [diff] [blame] | 1837 | server-state-file-name X - X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1838 | source X - X X |
| 1839 | srvtimeout (deprecated) X - X X |
Baptiste Assmann | 5a54921 | 2015-10-12 20:30:24 +0200 | [diff] [blame] | 1840 | stats admin - X X X |
| 1841 | stats auth X X X X |
| 1842 | stats enable X X X X |
| 1843 | stats hide-version X X X X |
| 1844 | stats http-request - X X X |
| 1845 | stats realm X X X X |
| 1846 | stats refresh X X X X |
| 1847 | stats scope X X X X |
| 1848 | stats show-desc X X X X |
| 1849 | stats show-legends X X X X |
| 1850 | stats show-node X X X X |
| 1851 | stats uri X X X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1852 | -- keyword -------------------------- defaults - frontend - listen -- backend - |
| 1853 | stick match - - X X |
| 1854 | stick on - - X X |
| 1855 | stick store-request - - X X |
Willy Tarreau | d8dc99f | 2011-07-01 11:33:25 +0200 | [diff] [blame] | 1856 | stick store-response - - X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1857 | stick-table - - X X |
Willy Tarreau | 938c7fe | 2014-04-25 14:21:39 +0200 | [diff] [blame] | 1858 | tcp-check connect - - X X |
| 1859 | tcp-check expect - - X X |
| 1860 | tcp-check send - - X X |
| 1861 | tcp-check send-binary - - X X |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 1862 | tcp-request connection - X X - |
| 1863 | tcp-request content - X X X |
Willy Tarreau | a56235c | 2010-09-14 11:31:36 +0200 | [diff] [blame] | 1864 | tcp-request inspect-delay - X X X |
Emeric Brun | 0a3b67f | 2010-09-24 15:34:53 +0200 | [diff] [blame] | 1865 | tcp-response content - - X X |
| 1866 | tcp-response inspect-delay - - X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1867 | timeout check X - X X |
| 1868 | timeout client X X X - |
Willy Tarreau | 05cdd96 | 2014-05-10 14:30:07 +0200 | [diff] [blame] | 1869 | timeout client-fin X X X - |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1870 | timeout clitimeout (deprecated) X X X - |
| 1871 | timeout connect X - X X |
| 1872 | timeout contimeout (deprecated) X - X X |
| 1873 | timeout http-keep-alive X X X X |
| 1874 | timeout http-request X X X X |
| 1875 | timeout queue X - X X |
| 1876 | timeout server X - X X |
Willy Tarreau | 05cdd96 | 2014-05-10 14:30:07 +0200 | [diff] [blame] | 1877 | timeout server-fin X - X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1878 | timeout srvtimeout (deprecated) X - X X |
| 1879 | timeout tarpit X X X X |
Willy Tarreau | ce887fd | 2012-05-12 12:50:00 +0200 | [diff] [blame] | 1880 | timeout tunnel X - X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1881 | transparent (deprecated) X - X X |
William Lallemand | a73203e | 2012-03-12 12:48:57 +0100 | [diff] [blame] | 1882 | unique-id-format X X X - |
| 1883 | unique-id-header X X X - |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1884 | use_backend - X X - |
Willy Tarreau | 4a5cade | 2012-04-05 21:09:48 +0200 | [diff] [blame] | 1885 | use-server - - X X |
Willy Tarreau | 5c6f7b3 | 2010-02-26 13:34:29 +0100 | [diff] [blame] | 1886 | ------------------------------------+----------+----------+---------+--------- |
| 1887 | keyword defaults frontend listen backend |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 1888 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1889 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 1890 | 4.2. Alphabetically sorted keywords reference |
| 1891 | --------------------------------------------- |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1892 | |
| 1893 | This section provides a description of each keyword and its usage. |
| 1894 | |
| 1895 | |
| 1896 | acl <aclname> <criterion> [flags] [operator] <value> ... |
| 1897 | Declare or complete an access list. |
| 1898 | May be used in sections : defaults | frontend | listen | backend |
| 1899 | no | yes | yes | yes |
| 1900 | Example: |
| 1901 | acl invalid_src src 0.0.0.0/7 224.0.0.0/3 |
| 1902 | acl invalid_src src_port 0:1023 |
| 1903 | acl local_dst hdr(host) -i localhost |
| 1904 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 1905 | See section 7 about ACL usage. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1906 | |
| 1907 | |
Cyril Bonté | b21570a | 2009-11-29 20:04:48 +0100 | [diff] [blame] | 1908 | appsession <cookie> len <length> timeout <holdtime> |
| 1909 | [request-learn] [prefix] [mode <path-parameters|query-string>] |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1910 | Define session stickiness on an existing application cookie. |
| 1911 | May be used in sections : defaults | frontend | listen | backend |
| 1912 | no | no | yes | yes |
| 1913 | Arguments : |
| 1914 | <cookie> this is the name of the cookie used by the application and which |
| 1915 | HAProxy will have to learn for each new session. |
| 1916 | |
Cyril Bonté | b21570a | 2009-11-29 20:04:48 +0100 | [diff] [blame] | 1917 | <length> this is the max number of characters that will be memorized and |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1918 | checked in each cookie value. |
| 1919 | |
| 1920 | <holdtime> this is the time after which the cookie will be removed from |
| 1921 | memory if unused. If no unit is specified, this time is in |
| 1922 | milliseconds. |
| 1923 | |
Cyril Bonté | bf47aeb | 2009-10-15 00:15:40 +0200 | [diff] [blame] | 1924 | request-learn |
| 1925 | If this option is specified, then haproxy will be able to learn |
| 1926 | the cookie found in the request in case the server does not |
| 1927 | specify any in response. This is typically what happens with |
| 1928 | PHPSESSID cookies, or when haproxy's session expires before |
| 1929 | the application's session and the correct server is selected. |
| 1930 | It is recommended to specify this option to improve reliability. |
| 1931 | |
Cyril Bonté | b21570a | 2009-11-29 20:04:48 +0100 | [diff] [blame] | 1932 | prefix When this option is specified, haproxy will match on the cookie |
| 1933 | prefix (or URL parameter prefix). The appsession value is the |
| 1934 | data following this prefix. |
| 1935 | |
| 1936 | Example : |
| 1937 | appsession ASPSESSIONID len 64 timeout 3h prefix |
| 1938 | |
| 1939 | This will match the cookie ASPSESSIONIDXXXX=XXXXX, |
| 1940 | the appsession value will be XXXX=XXXXX. |
| 1941 | |
| 1942 | mode This option allows to change the URL parser mode. |
| 1943 | 2 modes are currently supported : |
| 1944 | - path-parameters : |
| 1945 | The parser looks for the appsession in the path parameters |
| 1946 | part (each parameter is separated by a semi-colon), which is |
| 1947 | convenient for JSESSIONID for example. |
| 1948 | This is the default mode if the option is not set. |
| 1949 | - query-string : |
| 1950 | In this mode, the parser will look for the appsession in the |
| 1951 | query string. |
| 1952 | |
Willy Tarreau | 294d0f0 | 2015-08-10 19:40:12 +0200 | [diff] [blame] | 1953 | As of version 1.6, appsessions was removed. It is more flexible and more |
| 1954 | convenient to use stick-tables instead, and stick-tables support multi-master |
| 1955 | replication and data conservation across reloads, which appsessions did not. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1956 | |
Cyril Bonté | 02ff8ef | 2010-12-14 22:48:49 +0100 | [diff] [blame] | 1957 | See also : "cookie", "capture cookie", "balance", "stick", "stick-table", |
| 1958 | "ignore-persist", "nbproc" and "bind-process". |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1959 | |
| 1960 | |
Willy Tarreau | c73ce2b | 2008-01-06 10:55:10 +0100 | [diff] [blame] | 1961 | backlog <conns> |
| 1962 | Give hints to the system about the approximate listen backlog desired size |
| 1963 | May be used in sections : defaults | frontend | listen | backend |
| 1964 | yes | yes | yes | no |
| 1965 | Arguments : |
| 1966 | <conns> is the number of pending connections. Depending on the operating |
| 1967 | system, it may represent the number of already acknowledged |
Cyril Bonté | dc4d903 | 2012-04-08 21:57:39 +0200 | [diff] [blame] | 1968 | connections, of non-acknowledged ones, or both. |
Willy Tarreau | c73ce2b | 2008-01-06 10:55:10 +0100 | [diff] [blame] | 1969 | |
| 1970 | In order to protect against SYN flood attacks, one solution is to increase |
| 1971 | the system's SYN backlog size. Depending on the system, sometimes it is just |
| 1972 | tunable via a system parameter, sometimes it is not adjustable at all, and |
| 1973 | sometimes the system relies on hints given by the application at the time of |
| 1974 | the listen() syscall. By default, HAProxy passes the frontend's maxconn value |
| 1975 | to the listen() syscall. On systems which can make use of this value, it can |
| 1976 | sometimes be useful to be able to specify a different value, hence this |
| 1977 | backlog parameter. |
| 1978 | |
| 1979 | On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is |
| 1980 | used as a hint and the system accepts up to the smallest greater power of |
| 1981 | two, and never more than some limits (usually 32768). |
| 1982 | |
| 1983 | See also : "maxconn" and the target operating system's tuning guide. |
| 1984 | |
| 1985 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1986 | balance <algorithm> [ <arguments> ] |
Willy Tarreau | 226071e | 2014-04-10 11:55:45 +0200 | [diff] [blame] | 1987 | balance url_param <param> [check_post] |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 1988 | Define the load balancing algorithm to be used in a backend. |
| 1989 | May be used in sections : defaults | frontend | listen | backend |
| 1990 | yes | no | yes | yes |
| 1991 | Arguments : |
| 1992 | <algorithm> is the algorithm used to select a server when doing load |
| 1993 | balancing. This only applies when no persistence information |
| 1994 | is available, or when a connection is redispatched to another |
| 1995 | server. <algorithm> may be one of the following : |
| 1996 | |
| 1997 | roundrobin Each server is used in turns, according to their weights. |
| 1998 | This is the smoothest and fairest algorithm when the server's |
| 1999 | processing time remains equally distributed. This algorithm |
| 2000 | is dynamic, which means that server weights may be adjusted |
Willy Tarreau | 9757a38 | 2009-10-03 12:56:50 +0200 | [diff] [blame] | 2001 | on the fly for slow starts for instance. It is limited by |
Godbach | a34bdc0 | 2013-07-22 07:44:53 +0800 | [diff] [blame] | 2002 | design to 4095 active servers per backend. Note that in some |
Willy Tarreau | 9757a38 | 2009-10-03 12:56:50 +0200 | [diff] [blame] | 2003 | large farms, when a server becomes up after having been down |
| 2004 | for a very short time, it may sometimes take a few hundreds |
| 2005 | requests for it to be re-integrated into the farm and start |
| 2006 | receiving traffic. This is normal, though very rare. It is |
| 2007 | indicated here in case you would have the chance to observe |
| 2008 | it, so that you don't worry. |
| 2009 | |
| 2010 | static-rr Each server is used in turns, according to their weights. |
| 2011 | This algorithm is as similar to roundrobin except that it is |
| 2012 | static, which means that changing a server's weight on the |
| 2013 | fly will have no effect. On the other hand, it has no design |
| 2014 | limitation on the number of servers, and when a server goes |
| 2015 | up, it is always immediately reintroduced into the farm, once |
| 2016 | the full map is recomputed. It also uses slightly less CPU to |
| 2017 | run (around -1%). |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2018 | |
Willy Tarreau | 2d2a7f8 | 2008-03-17 12:07:56 +0100 | [diff] [blame] | 2019 | leastconn The server with the lowest number of connections receives the |
| 2020 | connection. Round-robin is performed within groups of servers |
| 2021 | of the same load to ensure that all servers will be used. Use |
| 2022 | of this algorithm is recommended where very long sessions are |
| 2023 | expected, such as LDAP, SQL, TSE, etc... but is not very well |
| 2024 | suited for protocols using short sessions such as HTTP. This |
| 2025 | algorithm is dynamic, which means that server weights may be |
| 2026 | adjusted on the fly for slow starts for instance. |
| 2027 | |
Willy Tarreau | f09c660 | 2012-02-13 17:12:08 +0100 | [diff] [blame] | 2028 | first The first server with available connection slots receives the |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 2029 | connection. The servers are chosen from the lowest numeric |
Willy Tarreau | f09c660 | 2012-02-13 17:12:08 +0100 | [diff] [blame] | 2030 | identifier to the highest (see server parameter "id"), which |
| 2031 | defaults to the server's position in the farm. Once a server |
Willy Tarreau | 64559c5 | 2012-04-07 09:08:45 +0200 | [diff] [blame] | 2032 | reaches its maxconn value, the next server is used. It does |
Willy Tarreau | f09c660 | 2012-02-13 17:12:08 +0100 | [diff] [blame] | 2033 | not make sense to use this algorithm without setting maxconn. |
| 2034 | The purpose of this algorithm is to always use the smallest |
| 2035 | number of servers so that extra servers can be powered off |
| 2036 | during non-intensive hours. This algorithm ignores the server |
| 2037 | weight, and brings more benefit to long session such as RDP |
Willy Tarreau | 64559c5 | 2012-04-07 09:08:45 +0200 | [diff] [blame] | 2038 | or IMAP than HTTP, though it can be useful there too. In |
| 2039 | order to use this algorithm efficiently, it is recommended |
| 2040 | that a cloud controller regularly checks server usage to turn |
| 2041 | them off when unused, and regularly checks backend queue to |
| 2042 | turn new servers on when the queue inflates. Alternatively, |
| 2043 | using "http-check send-state" may inform servers on the load. |
Willy Tarreau | f09c660 | 2012-02-13 17:12:08 +0100 | [diff] [blame] | 2044 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2045 | source The source IP address is hashed and divided by the total |
| 2046 | weight of the running servers to designate which server will |
| 2047 | receive the request. This ensures that the same client IP |
| 2048 | address will always reach the same server as long as no |
| 2049 | server goes down or up. If the hash result changes due to the |
| 2050 | number of running servers changing, many clients will be |
| 2051 | directed to a different server. This algorithm is generally |
| 2052 | used in TCP mode where no cookie may be inserted. It may also |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 2053 | be used on the Internet to provide a best-effort stickiness |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2054 | to clients which refuse session cookies. This algorithm is |
Willy Tarreau | 6b2e11b | 2009-10-01 07:52:15 +0200 | [diff] [blame] | 2055 | static by default, which means that changing a server's |
| 2056 | weight on the fly will have no effect, but this can be |
| 2057 | changed using "hash-type". |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2058 | |
Oskar Stolc | 8dc4184 | 2012-05-19 10:19:54 +0100 | [diff] [blame] | 2059 | uri This algorithm hashes either the left part of the URI (before |
| 2060 | the question mark) or the whole URI (if the "whole" parameter |
| 2061 | is present) and divides the hash value by the total weight of |
| 2062 | the running servers. The result designates which server will |
| 2063 | receive the request. This ensures that the same URI will |
| 2064 | always be directed to the same server as long as no server |
| 2065 | goes up or down. This is used with proxy caches and |
| 2066 | anti-virus proxies in order to maximize the cache hit rate. |
| 2067 | Note that this algorithm may only be used in an HTTP backend. |
| 2068 | This algorithm is static by default, which means that |
| 2069 | changing a server's weight on the fly will have no effect, |
| 2070 | but this can be changed using "hash-type". |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2071 | |
Oskar Stolc | 8dc4184 | 2012-05-19 10:19:54 +0100 | [diff] [blame] | 2072 | This algorithm supports two optional parameters "len" and |
Marek Majkowski | 9c30fc1 | 2008-04-27 23:25:55 +0200 | [diff] [blame] | 2073 | "depth", both followed by a positive integer number. These |
| 2074 | options may be helpful when it is needed to balance servers |
| 2075 | based on the beginning of the URI only. The "len" parameter |
| 2076 | indicates that the algorithm should only consider that many |
| 2077 | characters at the beginning of the URI to compute the hash. |
| 2078 | Note that having "len" set to 1 rarely makes sense since most |
| 2079 | URIs start with a leading "/". |
| 2080 | |
| 2081 | The "depth" parameter indicates the maximum directory depth |
| 2082 | to be used to compute the hash. One level is counted for each |
| 2083 | slash in the request. If both parameters are specified, the |
| 2084 | evaluation stops when either is reached. |
| 2085 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2086 | url_param The URL parameter specified in argument will be looked up in |
matt.farnsworth@nokia.com | 1c2ab96 | 2008-04-14 20:47:37 +0200 | [diff] [blame] | 2087 | the query string of each HTTP GET request. |
| 2088 | |
| 2089 | If the modifier "check_post" is used, then an HTTP POST |
Cyril Bonté | dc4d903 | 2012-04-08 21:57:39 +0200 | [diff] [blame] | 2090 | request entity will be searched for the parameter argument, |
| 2091 | when it is not found in a query string after a question mark |
Willy Tarreau | 226071e | 2014-04-10 11:55:45 +0200 | [diff] [blame] | 2092 | ('?') in the URL. The message body will only start to be |
| 2093 | analyzed once either the advertised amount of data has been |
| 2094 | received or the request buffer is full. In the unlikely event |
| 2095 | that chunked encoding is used, only the first chunk is |
Cyril Bonté | dc4d903 | 2012-04-08 21:57:39 +0200 | [diff] [blame] | 2096 | scanned. Parameter values separated by a chunk boundary, may |
Willy Tarreau | 226071e | 2014-04-10 11:55:45 +0200 | [diff] [blame] | 2097 | be randomly balanced if at all. This keyword used to support |
| 2098 | an optional <max_wait> parameter which is now ignored. |
matt.farnsworth@nokia.com | 1c2ab96 | 2008-04-14 20:47:37 +0200 | [diff] [blame] | 2099 | |
| 2100 | If the parameter is found followed by an equal sign ('=') and |
| 2101 | a value, then the value is hashed and divided by the total |
| 2102 | weight of the running servers. The result designates which |
| 2103 | server will receive the request. |
| 2104 | |
| 2105 | This is used to track user identifiers in requests and ensure |
| 2106 | that a same user ID will always be sent to the same server as |
| 2107 | long as no server goes up or down. If no value is found or if |
| 2108 | the parameter is not found, then a round robin algorithm is |
| 2109 | applied. Note that this algorithm may only be used in an HTTP |
Willy Tarreau | 6b2e11b | 2009-10-01 07:52:15 +0200 | [diff] [blame] | 2110 | backend. This algorithm is static by default, which means |
| 2111 | that changing a server's weight on the fly will have no |
| 2112 | effect, but this can be changed using "hash-type". |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2113 | |
Cyril Bonté | dc4d903 | 2012-04-08 21:57:39 +0200 | [diff] [blame] | 2114 | hdr(<name>) The HTTP header <name> will be looked up in each HTTP |
| 2115 | request. Just as with the equivalent ACL 'hdr()' function, |
| 2116 | the header name in parenthesis is not case sensitive. If the |
| 2117 | header is absent or if it does not contain any value, the |
| 2118 | roundrobin algorithm is applied instead. |
Benoit | affb481 | 2009-03-25 13:02:10 +0100 | [diff] [blame] | 2119 | |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 2120 | An optional 'use_domain_only' parameter is available, for |
Benoit | affb481 | 2009-03-25 13:02:10 +0100 | [diff] [blame] | 2121 | reducing the hash algorithm to the main domain part with some |
| 2122 | specific headers such as 'Host'. For instance, in the Host |
| 2123 | value "haproxy.1wt.eu", only "1wt" will be considered. |
| 2124 | |
Willy Tarreau | 6b2e11b | 2009-10-01 07:52:15 +0200 | [diff] [blame] | 2125 | This algorithm is static by default, which means that |
| 2126 | changing a server's weight on the fly will have no effect, |
| 2127 | but this can be changed using "hash-type". |
| 2128 | |
Emeric Brun | 736aa23 | 2009-06-30 17:56:00 +0200 | [diff] [blame] | 2129 | rdp-cookie |
Hervé COMMOWICK | a3eb39c | 2011-08-05 18:48:51 +0200 | [diff] [blame] | 2130 | rdp-cookie(<name>) |
Emeric Brun | 736aa23 | 2009-06-30 17:56:00 +0200 | [diff] [blame] | 2131 | The RDP cookie <name> (or "mstshash" if omitted) will be |
| 2132 | looked up and hashed for each incoming TCP request. Just as |
| 2133 | with the equivalent ACL 'req_rdp_cookie()' function, the name |
| 2134 | is not case-sensitive. This mechanism is useful as a degraded |
| 2135 | persistence mode, as it makes it possible to always send the |
| 2136 | same user (or the same session ID) to the same server. If the |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 2137 | cookie is not found, the normal roundrobin algorithm is |
Emeric Brun | 736aa23 | 2009-06-30 17:56:00 +0200 | [diff] [blame] | 2138 | used instead. |
| 2139 | |
| 2140 | Note that for this to work, the frontend must ensure that an |
| 2141 | RDP cookie is already present in the request buffer. For this |
| 2142 | you must use 'tcp-request content accept' rule combined with |
| 2143 | a 'req_rdp_cookie_cnt' ACL. |
| 2144 | |
Willy Tarreau | 6b2e11b | 2009-10-01 07:52:15 +0200 | [diff] [blame] | 2145 | This algorithm is static by default, which means that |
| 2146 | changing a server's weight on the fly will have no effect, |
| 2147 | but this can be changed using "hash-type". |
| 2148 | |
Cyril Bonté | dc4d903 | 2012-04-08 21:57:39 +0200 | [diff] [blame] | 2149 | See also the rdp_cookie pattern fetch function. |
Simon Horman | ab814e0 | 2011-06-24 14:50:20 +0900 | [diff] [blame] | 2150 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2151 | <arguments> is an optional list of arguments which may be needed by some |
Marek Majkowski | 9c30fc1 | 2008-04-27 23:25:55 +0200 | [diff] [blame] | 2152 | algorithms. Right now, only "url_param" and "uri" support an |
| 2153 | optional argument. |
matt.farnsworth@nokia.com | 1c2ab96 | 2008-04-14 20:47:37 +0200 | [diff] [blame] | 2154 | |
Willy Tarreau | 3cd9af2 | 2009-03-15 14:06:41 +0100 | [diff] [blame] | 2155 | The load balancing algorithm of a backend is set to roundrobin when no other |
| 2156 | algorithm, mode nor option have been set. The algorithm may only be set once |
| 2157 | for each backend. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2158 | |
| 2159 | Examples : |
| 2160 | balance roundrobin |
| 2161 | balance url_param userid |
matt.farnsworth@nokia.com | 1c2ab96 | 2008-04-14 20:47:37 +0200 | [diff] [blame] | 2162 | balance url_param session_id check_post 64 |
Benoit | affb481 | 2009-03-25 13:02:10 +0100 | [diff] [blame] | 2163 | balance hdr(User-Agent) |
| 2164 | balance hdr(host) |
| 2165 | balance hdr(Host) use_domain_only |
matt.farnsworth@nokia.com | 1c2ab96 | 2008-04-14 20:47:37 +0200 | [diff] [blame] | 2166 | |
| 2167 | Note: the following caveats and limitations on using the "check_post" |
| 2168 | extension with "url_param" must be considered : |
| 2169 | |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 2170 | - all POST requests are eligible for consideration, because there is no way |
matt.farnsworth@nokia.com | 1c2ab96 | 2008-04-14 20:47:37 +0200 | [diff] [blame] | 2171 | to determine if the parameters will be found in the body or entity which |
| 2172 | may contain binary data. Therefore another method may be required to |
| 2173 | restrict consideration of POST requests that have no URL parameters in |
| 2174 | the body. (see acl reqideny http_end) |
| 2175 | |
| 2176 | - using a <max_wait> value larger than the request buffer size does not |
| 2177 | make sense and is useless. The buffer size is set at build time, and |
| 2178 | defaults to 16 kB. |
| 2179 | |
| 2180 | - Content-Encoding is not supported, the parameter search will probably |
| 2181 | fail; and load balancing will fall back to Round Robin. |
| 2182 | |
| 2183 | - Expect: 100-continue is not supported, load balancing will fall back to |
| 2184 | Round Robin. |
| 2185 | |
| 2186 | - Transfer-Encoding (RFC2616 3.6.1) is only supported in the first chunk. |
| 2187 | If the entire parameter value is not present in the first chunk, the |
| 2188 | selection of server is undefined (actually, defined by how little |
| 2189 | actually appeared in the first chunk). |
| 2190 | |
| 2191 | - This feature does not support generation of a 100, 411 or 501 response. |
| 2192 | |
| 2193 | - In some cases, requesting "check_post" MAY attempt to scan the entire |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 2194 | contents of a message body. Scanning normally terminates when linear |
matt.farnsworth@nokia.com | 1c2ab96 | 2008-04-14 20:47:37 +0200 | [diff] [blame] | 2195 | white space or control characters are found, indicating the end of what |
| 2196 | might be a URL parameter list. This is probably not a concern with SGML |
| 2197 | type message bodies. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2198 | |
Willy Tarreau | 294d0f0 | 2015-08-10 19:40:12 +0200 | [diff] [blame] | 2199 | See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy". |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2200 | |
| 2201 | |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 2202 | bind [<address>]:<port_range> [, ...] [param*] |
| 2203 | bind /<path> [, ...] [param*] |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2204 | Define one or several listening addresses and/or ports in a frontend. |
| 2205 | May be used in sections : defaults | frontend | listen | backend |
| 2206 | no | yes | yes | no |
| 2207 | Arguments : |
Willy Tarreau | b1e52e8 | 2008-01-13 14:49:51 +0100 | [diff] [blame] | 2208 | <address> is optional and can be a host name, an IPv4 address, an IPv6 |
| 2209 | address, or '*'. It designates the address the frontend will |
| 2210 | listen on. If unset, all IPv4 addresses of the system will be |
| 2211 | listened on. The same will apply for '*' or the system's |
David du Colombier | 9c938da | 2011-03-17 10:40:27 +0100 | [diff] [blame] | 2212 | special address "0.0.0.0". The IPv6 equivalent is '::'. |
Willy Tarreau | 2470928 | 2013-03-10 21:32:12 +0100 | [diff] [blame] | 2213 | Optionally, an address family prefix may be used before the |
| 2214 | address to force the family regardless of the address format, |
| 2215 | which can be useful to specify a path to a unix socket with |
| 2216 | no slash ('/'). Currently supported prefixes are : |
| 2217 | - 'ipv4@' -> address is always IPv4 |
| 2218 | - 'ipv6@' -> address is always IPv6 |
| 2219 | - 'unix@' -> address is a path to a local unix socket |
Willy Tarreau | 70f72e0 | 2014-07-08 00:37:50 +0200 | [diff] [blame] | 2220 | - 'abns@' -> address is in abstract namespace (Linux only). |
| 2221 | Note: since abstract sockets are not "rebindable", they |
| 2222 | do not cope well with multi-process mode during |
| 2223 | soft-restart, so it is better to avoid them if |
| 2224 | nbproc is greater than 1. The effect is that if the |
| 2225 | new process fails to start, only one of the old ones |
| 2226 | will be able to rebind to the socket. |
Willy Tarreau | 40aa070 | 2013-03-10 23:51:38 +0100 | [diff] [blame] | 2227 | - 'fd@<n>' -> use file descriptor <n> inherited from the |
| 2228 | parent. The fd must be bound and may or may not already |
| 2229 | be listening. |
William Lallemand | b2f0745 | 2015-05-12 14:27:13 +0200 | [diff] [blame] | 2230 | You may want to reference some environment variables in the |
| 2231 | address parameter, see section 2.3 about environment |
| 2232 | variables. |
Willy Tarreau | b1e52e8 | 2008-01-13 14:49:51 +0100 | [diff] [blame] | 2233 | |
Willy Tarreau | c5011ca | 2010-03-22 11:53:56 +0100 | [diff] [blame] | 2234 | <port_range> is either a unique TCP port, or a port range for which the |
| 2235 | proxy will accept connections for the IP address specified |
Willy Tarreau | ceb24bc | 2010-11-09 12:46:41 +0100 | [diff] [blame] | 2236 | above. The port is mandatory for TCP listeners. Note that in |
| 2237 | the case of an IPv6 address, the port is always the number |
| 2238 | after the last colon (':'). A range can either be : |
Willy Tarreau | c5011ca | 2010-03-22 11:53:56 +0100 | [diff] [blame] | 2239 | - a numerical port (ex: '80') |
| 2240 | - a dash-delimited ports range explicitly stating the lower |
| 2241 | and upper bounds (ex: '2000-2100') which are included in |
| 2242 | the range. |
| 2243 | |
| 2244 | Particular care must be taken against port ranges, because |
| 2245 | every <address:port> couple consumes one socket (= a file |
| 2246 | descriptor), so it's easy to consume lots of descriptors |
| 2247 | with a simple range, and to run out of sockets. Also, each |
| 2248 | <address:port> couple must be used only once among all |
| 2249 | instances running on a same system. Please note that binding |
| 2250 | to ports lower than 1024 generally require particular |
Jamie Gloudon | 801a0a3 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 2251 | privileges to start the program, which are independent of |
Willy Tarreau | c5011ca | 2010-03-22 11:53:56 +0100 | [diff] [blame] | 2252 | the 'uid' parameter. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2253 | |
Willy Tarreau | ceb24bc | 2010-11-09 12:46:41 +0100 | [diff] [blame] | 2254 | <path> is a UNIX socket path beginning with a slash ('/'). This is |
| 2255 | alternative to the TCP listening port. Haproxy will then |
| 2256 | receive UNIX connections on the socket located at this place. |
| 2257 | The path must begin with a slash and by default is absolute. |
| 2258 | It can be relative to the prefix defined by "unix-bind" in |
| 2259 | the global section. Note that the total length of the prefix |
| 2260 | followed by the socket path cannot exceed some system limits |
| 2261 | for UNIX sockets, which commonly are set to 107 characters. |
| 2262 | |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 2263 | <param*> is a list of parameters common to all sockets declared on the |
| 2264 | same line. These numerous parameters depend on OS and build |
| 2265 | options and have a complete section dedicated to them. Please |
| 2266 | refer to section 5 to for more details. |
Willy Tarreau | a0ee1d0 | 2012-09-10 09:01:23 +0200 | [diff] [blame] | 2267 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2268 | It is possible to specify a list of address:port combinations delimited by |
| 2269 | commas. The frontend will then listen on all of these addresses. There is no |
| 2270 | fixed limit to the number of addresses and ports which can be listened on in |
| 2271 | a frontend, as well as there is no limit to the number of "bind" statements |
| 2272 | in a frontend. |
| 2273 | |
| 2274 | Example : |
| 2275 | listen http_proxy |
| 2276 | bind :80,:443 |
| 2277 | bind 10.0.0.1:10080,10.0.0.1:10443 |
Willy Tarreau | ceb24bc | 2010-11-09 12:46:41 +0100 | [diff] [blame] | 2278 | bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2279 | |
Willy Tarreau | a0ee1d0 | 2012-09-10 09:01:23 +0200 | [diff] [blame] | 2280 | listen http_https_proxy |
| 2281 | bind :80 |
Cyril Bonté | 0d44fc6 | 2012-10-09 22:45:33 +0200 | [diff] [blame] | 2282 | bind :443 ssl crt /etc/haproxy/site.pem |
Willy Tarreau | a0ee1d0 | 2012-09-10 09:01:23 +0200 | [diff] [blame] | 2283 | |
Willy Tarreau | 2470928 | 2013-03-10 21:32:12 +0100 | [diff] [blame] | 2284 | listen http_https_proxy_explicit |
| 2285 | bind ipv6@:80 |
| 2286 | bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem |
| 2287 | bind unix@ssl-frontend.sock user root mode 600 accept-proxy |
| 2288 | |
Willy Tarreau | dad36a3 | 2013-03-11 01:20:04 +0100 | [diff] [blame] | 2289 | listen external_bind_app1 |
William Lallemand | b2f0745 | 2015-05-12 14:27:13 +0200 | [diff] [blame] | 2290 | bind "fd@${FD_APP1}" |
Willy Tarreau | dad36a3 | 2013-03-11 01:20:04 +0100 | [diff] [blame] | 2291 | |
Willy Tarreau | 55dcaf6 | 2015-09-27 15:03:15 +0200 | [diff] [blame] | 2292 | Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole |
| 2293 | sun_path length is used for the address length. Some other programs |
| 2294 | such as socat use the string length only by default. Pass the option |
| 2295 | ",unix-tightsocklen=0" to any abstract socket definition in socat to |
| 2296 | make it compatible with HAProxy's. |
| 2297 | |
Willy Tarreau | ceb24bc | 2010-11-09 12:46:41 +0100 | [diff] [blame] | 2298 | See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 2299 | documentation, and section 5 about bind options. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2300 | |
| 2301 | |
Willy Tarreau | a9db57e | 2013-01-18 11:29:29 +0100 | [diff] [blame] | 2302 | bind-process [ all | odd | even | <number 1-64>[-<number 1-64>] ] ... |
Willy Tarreau | 0b9c02c | 2009-02-04 22:05:05 +0100 | [diff] [blame] | 2303 | Limit visibility of an instance to a certain set of processes numbers. |
| 2304 | May be used in sections : defaults | frontend | listen | backend |
| 2305 | yes | yes | yes | yes |
| 2306 | Arguments : |
| 2307 | all All process will see this instance. This is the default. It |
| 2308 | may be used to override a default value. |
| 2309 | |
Willy Tarreau | a9db57e | 2013-01-18 11:29:29 +0100 | [diff] [blame] | 2310 | odd This instance will be enabled on processes 1,3,5,...63. This |
Willy Tarreau | 0b9c02c | 2009-02-04 22:05:05 +0100 | [diff] [blame] | 2311 | option may be combined with other numbers. |
| 2312 | |
Willy Tarreau | a9db57e | 2013-01-18 11:29:29 +0100 | [diff] [blame] | 2313 | even This instance will be enabled on processes 2,4,6,...64. This |
Willy Tarreau | 0b9c02c | 2009-02-04 22:05:05 +0100 | [diff] [blame] | 2314 | option may be combined with other numbers. Do not use it |
| 2315 | with less than 2 processes otherwise some instances might be |
| 2316 | missing from all processes. |
| 2317 | |
Willy Tarreau | 110ecc1 | 2012-11-15 17:50:01 +0100 | [diff] [blame] | 2318 | number The instance will be enabled on this process number or range, |
Willy Tarreau | a9db57e | 2013-01-18 11:29:29 +0100 | [diff] [blame] | 2319 | whose values must all be between 1 and 32 or 64 depending on |
Willy Tarreau | 102df61 | 2014-05-07 23:56:38 +0200 | [diff] [blame] | 2320 | the machine's word size. If a proxy is bound to process |
| 2321 | numbers greater than the configured global.nbproc, it will |
| 2322 | either be forced to process #1 if a single process was |
| 2323 | specified, or to all processes otherwise. |
Willy Tarreau | 0b9c02c | 2009-02-04 22:05:05 +0100 | [diff] [blame] | 2324 | |
| 2325 | This keyword limits binding of certain instances to certain processes. This |
| 2326 | is useful in order not to have too many processes listening to the same |
| 2327 | ports. For instance, on a dual-core machine, it might make sense to set |
| 2328 | 'nbproc 2' in the global section, then distributes the listeners among 'odd' |
| 2329 | and 'even' instances. |
| 2330 | |
Willy Tarreau | a9db57e | 2013-01-18 11:29:29 +0100 | [diff] [blame] | 2331 | At the moment, it is not possible to reference more than 32 or 64 processes |
| 2332 | using this keyword, but this should be more than enough for most setups. |
| 2333 | Please note that 'all' really means all processes regardless of the machine's |
| 2334 | word size, and is not limited to the first 32 or 64. |
Willy Tarreau | 0b9c02c | 2009-02-04 22:05:05 +0100 | [diff] [blame] | 2335 | |
Willy Tarreau | 6ae1ba6 | 2014-05-07 19:01:58 +0200 | [diff] [blame] | 2336 | Each "bind" line may further be limited to a subset of the proxy's processes, |
| 2337 | please consult the "process" bind keyword in section 5.1. |
| 2338 | |
Willy Tarreau | b369a04 | 2014-09-16 13:21:03 +0200 | [diff] [blame] | 2339 | When a frontend has no explicit "bind-process" line, it tries to bind to all |
| 2340 | the processes referenced by its "bind" lines. That means that frontends can |
| 2341 | easily adapt to their listeners' processes. |
| 2342 | |
Willy Tarreau | 0b9c02c | 2009-02-04 22:05:05 +0100 | [diff] [blame] | 2343 | If some backends are referenced by frontends bound to other processes, the |
| 2344 | backend automatically inherits the frontend's processes. |
| 2345 | |
| 2346 | Example : |
| 2347 | listen app_ip1 |
| 2348 | bind 10.0.0.1:80 |
Willy Tarreau | bfcd311 | 2010-10-23 11:22:08 +0200 | [diff] [blame] | 2349 | bind-process odd |
Willy Tarreau | 0b9c02c | 2009-02-04 22:05:05 +0100 | [diff] [blame] | 2350 | |
| 2351 | listen app_ip2 |
| 2352 | bind 10.0.0.2:80 |
Willy Tarreau | bfcd311 | 2010-10-23 11:22:08 +0200 | [diff] [blame] | 2353 | bind-process even |
Willy Tarreau | 0b9c02c | 2009-02-04 22:05:05 +0100 | [diff] [blame] | 2354 | |
| 2355 | listen management |
| 2356 | bind 10.0.0.3:80 |
Willy Tarreau | bfcd311 | 2010-10-23 11:22:08 +0200 | [diff] [blame] | 2357 | bind-process 1 2 3 4 |
Willy Tarreau | 0b9c02c | 2009-02-04 22:05:05 +0100 | [diff] [blame] | 2358 | |
Willy Tarreau | 110ecc1 | 2012-11-15 17:50:01 +0100 | [diff] [blame] | 2359 | listen management |
| 2360 | bind 10.0.0.4:80 |
| 2361 | bind-process 1-4 |
| 2362 | |
Willy Tarreau | 6ae1ba6 | 2014-05-07 19:01:58 +0200 | [diff] [blame] | 2363 | See also : "nbproc" in global section, and "process" in section 5.1. |
Willy Tarreau | 0b9c02c | 2009-02-04 22:05:05 +0100 | [diff] [blame] | 2364 | |
| 2365 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2366 | block { if | unless } <condition> |
| 2367 | Block a layer 7 request if/unless a condition is matched |
| 2368 | May be used in sections : defaults | frontend | listen | backend |
| 2369 | no | yes | yes | yes |
| 2370 | |
| 2371 | The HTTP request will be blocked very early in the layer 7 processing |
| 2372 | if/unless <condition> is matched. A 403 error will be returned if the request |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 2373 | is blocked. The condition has to reference ACLs (see section 7). This is |
Willy Tarreau | 3c92c5f | 2011-08-28 09:45:47 +0200 | [diff] [blame] | 2374 | typically used to deny access to certain sensitive resources if some |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2375 | conditions are met or not met. There is no fixed limit to the number of |
| 2376 | "block" statements per instance. |
| 2377 | |
| 2378 | Example: |
| 2379 | acl invalid_src src 0.0.0.0/7 224.0.0.0/3 |
| 2380 | acl invalid_src src_port 0:1023 |
| 2381 | acl local_dst hdr(host) -i localhost |
| 2382 | block if invalid_src || local_dst |
| 2383 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 2384 | See section 7 about ACL usage. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2385 | |
| 2386 | |
| 2387 | capture cookie <name> len <length> |
| 2388 | Capture and log a cookie in the request and in the response. |
| 2389 | May be used in sections : defaults | frontend | listen | backend |
| 2390 | no | yes | yes | no |
| 2391 | Arguments : |
| 2392 | <name> is the beginning of the name of the cookie to capture. In order |
| 2393 | to match the exact name, simply suffix the name with an equal |
| 2394 | sign ('='). The full name will appear in the logs, which is |
| 2395 | useful with application servers which adjust both the cookie name |
| 2396 | and value (eg: ASPSESSIONXXXXX). |
| 2397 | |
| 2398 | <length> is the maximum number of characters to report in the logs, which |
| 2399 | include the cookie name, the equal sign and the value, all in the |
| 2400 | standard "name=value" form. The string will be truncated on the |
| 2401 | right if it exceeds <length>. |
| 2402 | |
| 2403 | Only the first cookie is captured. Both the "cookie" request headers and the |
| 2404 | "set-cookie" response headers are monitored. This is particularly useful to |
| 2405 | check for application bugs causing session crossing or stealing between |
| 2406 | users, because generally the user's cookies can only change on a login page. |
| 2407 | |
| 2408 | When the cookie was not presented by the client, the associated log column |
| 2409 | will report "-". When a request does not cause a cookie to be assigned by the |
| 2410 | server, a "-" is reported in the response column. |
| 2411 | |
| 2412 | The capture is performed in the frontend only because it is necessary that |
| 2413 | the log format does not change for a given frontend depending on the |
| 2414 | backends. This may change in the future. Note that there can be only one |
Willy Tarreau | 193b8c6 | 2012-11-22 00:17:38 +0100 | [diff] [blame] | 2415 | "capture cookie" statement in a frontend. The maximum capture length is set |
| 2416 | by the global "tune.http.cookielen" setting and defaults to 63 characters. It |
| 2417 | is not possible to specify a capture in a "defaults" section. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2418 | |
| 2419 | Example: |
| 2420 | capture cookie ASPSESSION len 32 |
| 2421 | |
| 2422 | See also : "capture request header", "capture response header" as well as |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 2423 | section 8 about logging. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2424 | |
| 2425 | |
| 2426 | capture request header <name> len <length> |
Willy Tarreau | 4460d03 | 2012-11-21 23:37:37 +0100 | [diff] [blame] | 2427 | Capture and log the last occurrence of the specified request header. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2428 | May be used in sections : defaults | frontend | listen | backend |
| 2429 | no | yes | yes | no |
| 2430 | Arguments : |
| 2431 | <name> is the name of the header to capture. The header names are not |
Willy Tarreau | d2a4aa2 | 2008-01-31 15:28:22 +0100 | [diff] [blame] | 2432 | case-sensitive, but it is a common practice to write them as they |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2433 | appear in the requests, with the first letter of each word in |
| 2434 | upper case. The header name will not appear in the logs, only the |
| 2435 | value is reported, but the position in the logs is respected. |
| 2436 | |
| 2437 | <length> is the maximum number of characters to extract from the value and |
| 2438 | report in the logs. The string will be truncated on the right if |
| 2439 | it exceeds <length>. |
| 2440 | |
Willy Tarreau | 4460d03 | 2012-11-21 23:37:37 +0100 | [diff] [blame] | 2441 | The complete value of the last occurrence of the header is captured. The |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2442 | value will be added to the logs between braces ('{}'). If multiple headers |
| 2443 | are captured, they will be delimited by a vertical bar ('|') and will appear |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 2444 | in the same order they were declared in the configuration. Non-existent |
| 2445 | headers will be logged just as an empty string. Common uses for request |
| 2446 | header captures include the "Host" field in virtual hosting environments, the |
| 2447 | "Content-length" when uploads are supported, "User-agent" to quickly |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 2448 | differentiate between real users and robots, and "X-Forwarded-For" in proxied |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 2449 | environments to find where the request came from. |
| 2450 | |
| 2451 | Note that when capturing headers such as "User-agent", some spaces may be |
| 2452 | logged, making the log analysis more difficult. Thus be careful about what |
| 2453 | you log if you know your log parser is not smart enough to rely on the |
| 2454 | braces. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2455 | |
Willy Tarreau | 0900abb | 2012-11-22 00:21:46 +0100 | [diff] [blame] | 2456 | There is no limit to the number of captured request headers nor to their |
| 2457 | length, though it is wise to keep them low to limit memory usage per session. |
| 2458 | In order to keep log format consistent for a same frontend, header captures |
| 2459 | can only be declared in a frontend. It is not possible to specify a capture |
| 2460 | in a "defaults" section. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2461 | |
| 2462 | Example: |
| 2463 | capture request header Host len 15 |
| 2464 | capture request header X-Forwarded-For len 15 |
Cyril Bonté | d1b0f7c | 2015-10-26 22:37:39 +0100 | [diff] [blame] | 2465 | capture request header Referer len 15 |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2466 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 2467 | See also : "capture cookie", "capture response header" as well as section 8 |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2468 | about logging. |
| 2469 | |
| 2470 | |
| 2471 | capture response header <name> len <length> |
Willy Tarreau | 4460d03 | 2012-11-21 23:37:37 +0100 | [diff] [blame] | 2472 | Capture and log the last occurrence of the specified response header. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2473 | May be used in sections : defaults | frontend | listen | backend |
| 2474 | no | yes | yes | no |
| 2475 | Arguments : |
| 2476 | <name> is the name of the header to capture. The header names are not |
Willy Tarreau | d2a4aa2 | 2008-01-31 15:28:22 +0100 | [diff] [blame] | 2477 | case-sensitive, but it is a common practice to write them as they |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2478 | appear in the response, with the first letter of each word in |
| 2479 | upper case. The header name will not appear in the logs, only the |
| 2480 | value is reported, but the position in the logs is respected. |
| 2481 | |
| 2482 | <length> is the maximum number of characters to extract from the value and |
| 2483 | report in the logs. The string will be truncated on the right if |
| 2484 | it exceeds <length>. |
| 2485 | |
Willy Tarreau | 4460d03 | 2012-11-21 23:37:37 +0100 | [diff] [blame] | 2486 | The complete value of the last occurrence of the header is captured. The |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2487 | result will be added to the logs between braces ('{}') after the captured |
| 2488 | request headers. If multiple headers are captured, they will be delimited by |
| 2489 | a vertical bar ('|') and will appear in the same order they were declared in |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 2490 | the configuration. Non-existent headers will be logged just as an empty |
| 2491 | string. Common uses for response header captures include the "Content-length" |
| 2492 | header which indicates how many bytes are expected to be returned, the |
| 2493 | "Location" header to track redirections. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2494 | |
Willy Tarreau | 0900abb | 2012-11-22 00:21:46 +0100 | [diff] [blame] | 2495 | There is no limit to the number of captured response headers nor to their |
| 2496 | length, though it is wise to keep them low to limit memory usage per session. |
| 2497 | In order to keep log format consistent for a same frontend, header captures |
| 2498 | can only be declared in a frontend. It is not possible to specify a capture |
| 2499 | in a "defaults" section. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2500 | |
| 2501 | Example: |
| 2502 | capture response header Content-length len 9 |
| 2503 | capture response header Location len 15 |
| 2504 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 2505 | See also : "capture cookie", "capture request header" as well as section 8 |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2506 | about logging. |
| 2507 | |
| 2508 | |
Cyril Bonté | f0c6061 | 2010-02-06 14:44:47 +0100 | [diff] [blame] | 2509 | clitimeout <timeout> (deprecated) |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2510 | Set the maximum inactivity time on the client side. |
| 2511 | May be used in sections : defaults | frontend | listen | backend |
| 2512 | yes | yes | yes | no |
| 2513 | Arguments : |
| 2514 | <timeout> is the timeout value is specified in milliseconds by default, but |
| 2515 | can be in any other unit if the number is suffixed by the unit, |
| 2516 | as explained at the top of this document. |
| 2517 | |
| 2518 | The inactivity timeout applies when the client is expected to acknowledge or |
| 2519 | send data. In HTTP mode, this timeout is particularly important to consider |
| 2520 | during the first phase, when the client sends the request, and during the |
| 2521 | response while it is reading data sent by the server. The value is specified |
| 2522 | in milliseconds by default, but can be in any other unit if the number is |
| 2523 | suffixed by the unit, as specified at the top of this document. In TCP mode |
| 2524 | (and to a lesser extent, in HTTP mode), it is highly recommended that the |
| 2525 | client timeout remains equal to the server timeout in order to avoid complex |
Willy Tarreau | d2a4aa2 | 2008-01-31 15:28:22 +0100 | [diff] [blame] | 2526 | situations to debug. It is a good practice to cover one or several TCP packet |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2527 | losses by specifying timeouts that are slightly above multiples of 3 seconds |
| 2528 | (eg: 4 or 5 seconds). |
| 2529 | |
| 2530 | This parameter is specific to frontends, but can be specified once for all in |
| 2531 | "defaults" sections. This is in fact one of the easiest solutions not to |
| 2532 | forget about it. An unspecified timeout results in an infinite timeout, which |
| 2533 | is not recommended. Such a usage is accepted and works but reports a warning |
| 2534 | during startup because it may results in accumulation of expired sessions in |
| 2535 | the system if the system's timeouts are not configured either. |
| 2536 | |
| 2537 | This parameter is provided for compatibility but is currently deprecated. |
| 2538 | Please use "timeout client" instead. |
| 2539 | |
Willy Tarreau | 036fae0 | 2008-01-06 13:24:40 +0100 | [diff] [blame] | 2540 | See also : "timeout client", "timeout http-request", "timeout server", and |
| 2541 | "srvtimeout". |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2542 | |
Cyril Bonté | 316a8cf | 2012-11-11 13:38:27 +0100 | [diff] [blame] | 2543 | compression algo <algorithm> ... |
| 2544 | compression type <mime type> ... |
Willy Tarreau | 70737d1 | 2012-10-27 00:34:28 +0200 | [diff] [blame] | 2545 | compression offload |
William Lallemand | 82fe75c | 2012-10-23 10:25:10 +0200 | [diff] [blame] | 2546 | Enable HTTP compression. |
| 2547 | May be used in sections : defaults | frontend | listen | backend |
| 2548 | yes | yes | yes | yes |
| 2549 | Arguments : |
Cyril Bonté | 316a8cf | 2012-11-11 13:38:27 +0100 | [diff] [blame] | 2550 | algo is followed by the list of supported compression algorithms. |
| 2551 | type is followed by the list of MIME types that will be compressed. |
| 2552 | offload makes haproxy work as a compression offloader only (see notes). |
| 2553 | |
| 2554 | The currently supported algorithms are : |
Willy Tarreau | c91840a | 2015-03-28 17:00:39 +0100 | [diff] [blame] | 2555 | identity this is mostly for debugging, and it was useful for developing |
| 2556 | the compression feature. Identity does not apply any change on |
| 2557 | data. |
Cyril Bonté | 316a8cf | 2012-11-11 13:38:27 +0100 | [diff] [blame] | 2558 | |
Willy Tarreau | c91840a | 2015-03-28 17:00:39 +0100 | [diff] [blame] | 2559 | gzip applies gzip compression. This setting is only available when |
Baptiste Assmann | f085d63 | 2015-12-21 17:57:32 +0100 | [diff] [blame] | 2560 | support for zlib or libslz was built in. |
Willy Tarreau | c91840a | 2015-03-28 17:00:39 +0100 | [diff] [blame] | 2561 | |
| 2562 | deflate same as "gzip", but with deflate algorithm and zlib format. |
| 2563 | Note that this algorithm has ambiguous support on many |
| 2564 | browsers and no support at all from recent ones. It is |
| 2565 | strongly recommended not to use it for anything else than |
| 2566 | experimentation. This setting is only available when support |
Baptiste Assmann | f085d63 | 2015-12-21 17:57:32 +0100 | [diff] [blame] | 2567 | for zlib or libslz was built in. |
Cyril Bonté | 316a8cf | 2012-11-11 13:38:27 +0100 | [diff] [blame] | 2568 | |
Willy Tarreau | c91840a | 2015-03-28 17:00:39 +0100 | [diff] [blame] | 2569 | raw-deflate same as "deflate" without the zlib wrapper, and used as an |
| 2570 | alternative when the browser wants "deflate". All major |
| 2571 | browsers understand it and despite violating the standards, |
| 2572 | it is known to work better than "deflate", at least on MSIE |
| 2573 | and some versions of Safari. Do not use it in conjunction |
| 2574 | with "deflate", use either one or the other since both react |
| 2575 | to the same Accept-Encoding token. This setting is only |
Baptiste Assmann | f085d63 | 2015-12-21 17:57:32 +0100 | [diff] [blame] | 2576 | available when support for zlib or libslz was built in. |
Cyril Bonté | 316a8cf | 2012-11-11 13:38:27 +0100 | [diff] [blame] | 2577 | |
Dmitry Sivachenko | 87c208b | 2012-11-22 20:03:26 +0400 | [diff] [blame] | 2578 | Compression will be activated depending on the Accept-Encoding request |
Cyril Bonté | 316a8cf | 2012-11-11 13:38:27 +0100 | [diff] [blame] | 2579 | header. With identity, it does not take care of that header. |
Dmitry Sivachenko | c9f3b45 | 2012-11-28 17:47:11 +0400 | [diff] [blame] | 2580 | If backend servers support HTTP compression, these directives |
| 2581 | will be no-op: haproxy will see the compressed response and will not |
| 2582 | compress again. If backend servers do not support HTTP compression and |
| 2583 | there is Accept-Encoding header in request, haproxy will compress the |
| 2584 | matching response. |
Willy Tarreau | 70737d1 | 2012-10-27 00:34:28 +0200 | [diff] [blame] | 2585 | |
| 2586 | The "offload" setting makes haproxy remove the Accept-Encoding header to |
| 2587 | prevent backend servers from compressing responses. It is strongly |
| 2588 | recommended not to do this because this means that all the compression work |
| 2589 | will be done on the single point where haproxy is located. However in some |
| 2590 | deployment scenarios, haproxy may be installed in front of a buggy gateway |
Dmitry Sivachenko | c9f3b45 | 2012-11-28 17:47:11 +0400 | [diff] [blame] | 2591 | with broken HTTP compression implementation which can't be turned off. |
| 2592 | In that case haproxy can be used to prevent that gateway from emitting |
| 2593 | invalid payloads. In this case, simply removing the header in the |
| 2594 | configuration does not work because it applies before the header is parsed, |
| 2595 | so that prevents haproxy from compressing. The "offload" setting should |
Willy Tarreau | ffea9fd | 2014-07-12 16:37:02 +0200 | [diff] [blame] | 2596 | then be used for such scenarios. Note: for now, the "offload" setting is |
| 2597 | ignored when set in a defaults section. |
William Lallemand | 82fe75c | 2012-10-23 10:25:10 +0200 | [diff] [blame] | 2598 | |
William Lallemand | 0509744 | 2012-11-20 12:14:28 +0100 | [diff] [blame] | 2599 | Compression is disabled when: |
Baptiste Assmann | 650d53d | 2013-01-05 15:44:44 +0100 | [diff] [blame] | 2600 | * the request does not advertise a supported compression algorithm in the |
| 2601 | "Accept-Encoding" header |
| 2602 | * the response message is not HTTP/1.1 |
William Lallemand | d300261 | 2012-11-26 14:34:47 +0100 | [diff] [blame] | 2603 | * HTTP status code is not 200 |
William Lallemand | 8bb4e34 | 2013-12-10 17:28:48 +0100 | [diff] [blame] | 2604 | * response header "Transfer-Encoding" contains "chunked" (Temporary |
| 2605 | Workaround) |
Baptiste Assmann | 650d53d | 2013-01-05 15:44:44 +0100 | [diff] [blame] | 2606 | * response contain neither a "Content-Length" header nor a |
| 2607 | "Transfer-Encoding" whose last value is "chunked" |
| 2608 | * response contains a "Content-Type" header whose first value starts with |
| 2609 | "multipart" |
| 2610 | * the response contains the "no-transform" value in the "Cache-control" |
| 2611 | header |
| 2612 | * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7 |
| 2613 | and later |
| 2614 | * The response contains a "Content-Encoding" header, indicating that the |
| 2615 | response is already compressed (see compression offload) |
William Lallemand | 0509744 | 2012-11-20 12:14:28 +0100 | [diff] [blame] | 2616 | |
Baptiste Assmann | 650d53d | 2013-01-05 15:44:44 +0100 | [diff] [blame] | 2617 | Note: The compression does not rewrite Etag headers, and does not emit the |
| 2618 | Warning header. |
William Lallemand | 0509744 | 2012-11-20 12:14:28 +0100 | [diff] [blame] | 2619 | |
William Lallemand | 82fe75c | 2012-10-23 10:25:10 +0200 | [diff] [blame] | 2620 | Examples : |
| 2621 | compression algo gzip |
| 2622 | compression type text/html text/plain |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2623 | |
Christopher Faulet | c3fe533 | 2016-04-07 15:30:10 +0200 | [diff] [blame] | 2624 | |
Cyril Bonté | f0c6061 | 2010-02-06 14:44:47 +0100 | [diff] [blame] | 2625 | contimeout <timeout> (deprecated) |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2626 | Set the maximum time to wait for a connection attempt to a server to succeed. |
| 2627 | May be used in sections : defaults | frontend | listen | backend |
| 2628 | yes | no | yes | yes |
| 2629 | Arguments : |
| 2630 | <timeout> is the timeout value is specified in milliseconds by default, but |
| 2631 | can be in any other unit if the number is suffixed by the unit, |
| 2632 | as explained at the top of this document. |
| 2633 | |
| 2634 | If the server is located on the same LAN as haproxy, the connection should be |
Willy Tarreau | d2a4aa2 | 2008-01-31 15:28:22 +0100 | [diff] [blame] | 2635 | immediate (less than a few milliseconds). Anyway, it is a good practice to |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 2636 | cover one or several TCP packet losses by specifying timeouts that are |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2637 | slightly above multiples of 3 seconds (eg: 4 or 5 seconds). By default, the |
| 2638 | connect timeout also presets the queue timeout to the same value if this one |
| 2639 | has not been specified. Historically, the contimeout was also used to set the |
| 2640 | tarpit timeout in a listen section, which is not possible in a pure frontend. |
| 2641 | |
| 2642 | This parameter is specific to backends, but can be specified once for all in |
| 2643 | "defaults" sections. This is in fact one of the easiest solutions not to |
| 2644 | forget about it. An unspecified timeout results in an infinite timeout, which |
| 2645 | is not recommended. Such a usage is accepted and works but reports a warning |
| 2646 | during startup because it may results in accumulation of failed sessions in |
| 2647 | the system if the system's timeouts are not configured either. |
| 2648 | |
| 2649 | This parameter is provided for backwards compatibility but is currently |
| 2650 | deprecated. Please use "timeout connect", "timeout queue" or "timeout tarpit" |
| 2651 | instead. |
| 2652 | |
| 2653 | See also : "timeout connect", "timeout queue", "timeout tarpit", |
| 2654 | "timeout server", "contimeout". |
| 2655 | |
| 2656 | |
Willy Tarreau | 55165fe | 2009-05-10 12:02:55 +0200 | [diff] [blame] | 2657 | cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ] |
Willy Tarreau | 4992dd2 | 2012-05-31 21:02:17 +0200 | [diff] [blame] | 2658 | [ postonly ] [ preserve ] [ httponly ] [ secure ] |
| 2659 | [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ] |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2660 | Enable cookie-based persistence in a backend. |
| 2661 | May be used in sections : defaults | frontend | listen | backend |
| 2662 | yes | no | yes | yes |
| 2663 | Arguments : |
| 2664 | <name> is the name of the cookie which will be monitored, modified or |
| 2665 | inserted in order to bring persistence. This cookie is sent to |
| 2666 | the client via a "Set-Cookie" header in the response, and is |
| 2667 | brought back by the client in a "Cookie" header in all requests. |
| 2668 | Special care should be taken to choose a name which does not |
| 2669 | conflict with any likely application cookie. Also, if the same |
| 2670 | backends are subject to be used by the same clients (eg: |
| 2671 | HTTP/HTTPS), care should be taken to use different cookie names |
| 2672 | between all backends if persistence between them is not desired. |
| 2673 | |
| 2674 | rewrite This keyword indicates that the cookie will be provided by the |
| 2675 | server and that haproxy will have to modify its value to set the |
| 2676 | server's identifier in it. This mode is handy when the management |
| 2677 | of complex combinations of "Set-cookie" and "Cache-control" |
| 2678 | headers is left to the application. The application can then |
| 2679 | decide whether or not it is appropriate to emit a persistence |
Lukas Tribus | f01a9cd | 2016-02-03 18:09:37 +0100 | [diff] [blame] | 2680 | cookie. Since all responses should be monitored, this mode |
| 2681 | doesn't work in HTTP tunnel mode. Unless the application |
| 2682 | behaviour is very complex and/or broken, it is advised not to |
| 2683 | start with this mode for new deployments. This keyword is |
| 2684 | incompatible with "insert" and "prefix". |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2685 | |
| 2686 | insert This keyword indicates that the persistence cookie will have to |
Willy Tarreau | a79094d | 2010-08-31 22:54:15 +0200 | [diff] [blame] | 2687 | be inserted by haproxy in server responses if the client did not |
Willy Tarreau | ba4c5be | 2010-10-23 12:46:42 +0200 | [diff] [blame] | 2688 | |
Willy Tarreau | a79094d | 2010-08-31 22:54:15 +0200 | [diff] [blame] | 2689 | already have a cookie that would have permitted it to access this |
Willy Tarreau | ba4c5be | 2010-10-23 12:46:42 +0200 | [diff] [blame] | 2690 | server. When used without the "preserve" option, if the server |
| 2691 | emits a cookie with the same name, it will be remove before |
| 2692 | processing. For this reason, this mode can be used to upgrade |
| 2693 | existing configurations running in the "rewrite" mode. The cookie |
| 2694 | will only be a session cookie and will not be stored on the |
| 2695 | client's disk. By default, unless the "indirect" option is added, |
| 2696 | the server will see the cookies emitted by the client. Due to |
| 2697 | caching effects, it is generally wise to add the "nocache" or |
| 2698 | "postonly" keywords (see below). The "insert" keyword is not |
| 2699 | compatible with "rewrite" and "prefix". |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2700 | |
| 2701 | prefix This keyword indicates that instead of relying on a dedicated |
| 2702 | cookie for the persistence, an existing one will be completed. |
| 2703 | This may be needed in some specific environments where the client |
| 2704 | does not support more than one single cookie and the application |
| 2705 | already needs it. In this case, whenever the server sets a cookie |
| 2706 | named <name>, it will be prefixed with the server's identifier |
| 2707 | and a delimiter. The prefix will be removed from all client |
| 2708 | requests so that the server still finds the cookie it emitted. |
| 2709 | Since all requests and responses are subject to being modified, |
Lukas Tribus | f01a9cd | 2016-02-03 18:09:37 +0100 | [diff] [blame] | 2710 | this mode doesn't work with tunnel mode. The "prefix" keyword is |
Willy Tarreau | 37229df | 2011-10-17 12:24:55 +0200 | [diff] [blame] | 2711 | not compatible with "rewrite" and "insert". Note: it is highly |
| 2712 | recommended not to use "indirect" with "prefix", otherwise server |
| 2713 | cookie updates would not be sent to clients. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2714 | |
Willy Tarreau | a79094d | 2010-08-31 22:54:15 +0200 | [diff] [blame] | 2715 | indirect When this option is specified, no cookie will be emitted to a |
| 2716 | client which already has a valid one for the server which has |
| 2717 | processed the request. If the server sets such a cookie itself, |
Willy Tarreau | ba4c5be | 2010-10-23 12:46:42 +0200 | [diff] [blame] | 2718 | it will be removed, unless the "preserve" option is also set. In |
| 2719 | "insert" mode, this will additionally remove cookies from the |
| 2720 | requests transmitted to the server, making the persistence |
| 2721 | mechanism totally transparent from an application point of view. |
Willy Tarreau | 37229df | 2011-10-17 12:24:55 +0200 | [diff] [blame] | 2722 | Note: it is highly recommended not to use "indirect" with |
| 2723 | "prefix", otherwise server cookie updates would not be sent to |
| 2724 | clients. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2725 | |
| 2726 | nocache This option is recommended in conjunction with the insert mode |
| 2727 | when there is a cache between the client and HAProxy, as it |
| 2728 | ensures that a cacheable response will be tagged non-cacheable if |
| 2729 | a cookie needs to be inserted. This is important because if all |
| 2730 | persistence cookies are added on a cacheable home page for |
| 2731 | instance, then all customers will then fetch the page from an |
| 2732 | outer cache and will all share the same persistence cookie, |
| 2733 | leading to one server receiving much more traffic than others. |
| 2734 | See also the "insert" and "postonly" options. |
| 2735 | |
| 2736 | postonly This option ensures that cookie insertion will only be performed |
| 2737 | on responses to POST requests. It is an alternative to the |
| 2738 | "nocache" option, because POST responses are not cacheable, so |
| 2739 | this ensures that the persistence cookie will never get cached. |
| 2740 | Since most sites do not need any sort of persistence before the |
| 2741 | first POST which generally is a login request, this is a very |
| 2742 | efficient method to optimize caching without risking to find a |
| 2743 | persistence cookie in the cache. |
| 2744 | See also the "insert" and "nocache" options. |
| 2745 | |
Willy Tarreau | ba4c5be | 2010-10-23 12:46:42 +0200 | [diff] [blame] | 2746 | preserve This option may only be used with "insert" and/or "indirect". It |
| 2747 | allows the server to emit the persistence cookie itself. In this |
| 2748 | case, if a cookie is found in the response, haproxy will leave it |
| 2749 | untouched. This is useful in order to end persistence after a |
| 2750 | logout request for instance. For this, the server just has to |
| 2751 | emit a cookie with an invalid value (eg: empty) or with a date in |
| 2752 | the past. By combining this mechanism with the "disable-on-404" |
| 2753 | check option, it is possible to perform a completely graceful |
| 2754 | shutdown because users will definitely leave the server after |
| 2755 | they logout. |
| 2756 | |
Willy Tarreau | 4992dd2 | 2012-05-31 21:02:17 +0200 | [diff] [blame] | 2757 | httponly This option tells haproxy to add an "HttpOnly" cookie attribute |
| 2758 | when a cookie is inserted. This attribute is used so that a |
| 2759 | user agent doesn't share the cookie with non-HTTP components. |
| 2760 | Please check RFC6265 for more information on this attribute. |
| 2761 | |
| 2762 | secure This option tells haproxy to add a "Secure" cookie attribute when |
| 2763 | a cookie is inserted. This attribute is used so that a user agent |
| 2764 | never emits this cookie over non-secure channels, which means |
| 2765 | that a cookie learned with this flag will be presented only over |
| 2766 | SSL/TLS connections. Please check RFC6265 for more information on |
| 2767 | this attribute. |
| 2768 | |
Krzysztof Piotr Oledzki | efe3b6f | 2008-05-23 23:49:32 +0200 | [diff] [blame] | 2769 | domain This option allows to specify the domain at which a cookie is |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 2770 | inserted. It requires exactly one parameter: a valid domain |
Willy Tarreau | 68a897b | 2009-12-03 23:28:34 +0100 | [diff] [blame] | 2771 | name. If the domain begins with a dot, the browser is allowed to |
| 2772 | use it for any host ending with that name. It is also possible to |
| 2773 | specify several domain names by invoking this option multiple |
| 2774 | times. Some browsers might have small limits on the number of |
| 2775 | domains, so be careful when doing that. For the record, sending |
| 2776 | 10 domains to MSIE 6 or Firefox 2 works as expected. |
Krzysztof Piotr Oledzki | efe3b6f | 2008-05-23 23:49:32 +0200 | [diff] [blame] | 2777 | |
Willy Tarreau | 996a92c | 2010-10-13 19:30:47 +0200 | [diff] [blame] | 2778 | maxidle This option allows inserted cookies to be ignored after some idle |
| 2779 | time. It only works with insert-mode cookies. When a cookie is |
| 2780 | sent to the client, the date this cookie was emitted is sent too. |
| 2781 | Upon further presentations of this cookie, if the date is older |
| 2782 | than the delay indicated by the parameter (in seconds), it will |
| 2783 | be ignored. Otherwise, it will be refreshed if needed when the |
| 2784 | response is sent to the client. This is particularly useful to |
| 2785 | prevent users who never close their browsers from remaining for |
| 2786 | too long on the same server (eg: after a farm size change). When |
| 2787 | this option is set and a cookie has no date, it is always |
| 2788 | accepted, but gets refreshed in the response. This maintains the |
| 2789 | ability for admins to access their sites. Cookies that have a |
| 2790 | date in the future further than 24 hours are ignored. Doing so |
| 2791 | lets admins fix timezone issues without risking kicking users off |
| 2792 | the site. |
| 2793 | |
| 2794 | maxlife This option allows inserted cookies to be ignored after some life |
| 2795 | time, whether they're in use or not. It only works with insert |
| 2796 | mode cookies. When a cookie is first sent to the client, the date |
| 2797 | this cookie was emitted is sent too. Upon further presentations |
| 2798 | of this cookie, if the date is older than the delay indicated by |
| 2799 | the parameter (in seconds), it will be ignored. If the cookie in |
| 2800 | the request has no date, it is accepted and a date will be set. |
| 2801 | Cookies that have a date in the future further than 24 hours are |
| 2802 | ignored. Doing so lets admins fix timezone issues without risking |
| 2803 | kicking users off the site. Contrary to maxidle, this value is |
| 2804 | not refreshed, only the first visit date counts. Both maxidle and |
| 2805 | maxlife may be used at the time. This is particularly useful to |
| 2806 | prevent users who never close their browsers from remaining for |
| 2807 | too long on the same server (eg: after a farm size change). This |
| 2808 | is stronger than the maxidle method in that it forces a |
| 2809 | redispatch after some absolute delay. |
| 2810 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2811 | There can be only one persistence cookie per HTTP backend, and it can be |
| 2812 | declared in a defaults section. The value of the cookie will be the value |
| 2813 | indicated after the "cookie" keyword in a "server" statement. If no cookie |
| 2814 | is declared for a given server, the cookie is not set. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 2815 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2816 | Examples : |
| 2817 | cookie JSESSIONID prefix |
| 2818 | cookie SRV insert indirect nocache |
| 2819 | cookie SRV insert postonly indirect |
Willy Tarreau | 996a92c | 2010-10-13 19:30:47 +0200 | [diff] [blame] | 2820 | cookie SRV insert indirect nocache maxidle 30m maxlife 8h |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2821 | |
Willy Tarreau | 294d0f0 | 2015-08-10 19:40:12 +0200 | [diff] [blame] | 2822 | See also : "balance source", "capture cookie", "server" and "ignore-persist". |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2823 | |
Willy Tarreau | 983e01e | 2010-01-11 18:42:06 +0100 | [diff] [blame] | 2824 | |
Thierry FOURNIER | a0a1b75 | 2015-05-26 17:44:32 +0200 | [diff] [blame] | 2825 | declare capture [ request | response ] len <length> |
| 2826 | Declares a capture slot. |
| 2827 | May be used in sections : defaults | frontend | listen | backend |
| 2828 | no | yes | yes | no |
| 2829 | Arguments: |
| 2830 | <length> is the length allowed for the capture. |
| 2831 | |
| 2832 | This declaration is only available in the frontend or listen section, but the |
| 2833 | reserved slot can be used in the backends. The "request" keyword allocates a |
| 2834 | capture slot for use in the request, and "response" allocates a capture slot |
| 2835 | for use in the response. |
| 2836 | |
| 2837 | See also: "capture-req", "capture-res" (sample converters), |
Baptiste Assmann | 5ac425c | 2015-10-21 23:13:46 +0200 | [diff] [blame] | 2838 | "capture.req.hdr", "capture.res.hdr" (sample fetches), |
Thierry FOURNIER | a0a1b75 | 2015-05-26 17:44:32 +0200 | [diff] [blame] | 2839 | "http-request capture" and "http-response capture". |
| 2840 | |
| 2841 | |
Krzysztof Piotr Oledzki | c6df066 | 2010-01-05 16:38:49 +0100 | [diff] [blame] | 2842 | default-server [param*] |
| 2843 | Change default options for a server in a backend |
| 2844 | May be used in sections : defaults | frontend | listen | backend |
| 2845 | yes | no | yes | yes |
| 2846 | Arguments: |
Willy Tarreau | 983e01e | 2010-01-11 18:42:06 +0100 | [diff] [blame] | 2847 | <param*> is a list of parameters for this server. The "default-server" |
| 2848 | keyword accepts an important number of options and has a complete |
| 2849 | section dedicated to it. Please refer to section 5 for more |
| 2850 | details. |
Krzysztof Piotr Oledzki | c6df066 | 2010-01-05 16:38:49 +0100 | [diff] [blame] | 2851 | |
Willy Tarreau | 983e01e | 2010-01-11 18:42:06 +0100 | [diff] [blame] | 2852 | Example : |
Krzysztof Piotr Oledzki | c6df066 | 2010-01-05 16:38:49 +0100 | [diff] [blame] | 2853 | default-server inter 1000 weight 13 |
| 2854 | |
| 2855 | See also: "server" and section 5 about server options |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2856 | |
Willy Tarreau | 983e01e | 2010-01-11 18:42:06 +0100 | [diff] [blame] | 2857 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2858 | default_backend <backend> |
| 2859 | Specify the backend to use when no "use_backend" rule has been matched. |
| 2860 | May be used in sections : defaults | frontend | listen | backend |
| 2861 | yes | yes | yes | no |
| 2862 | Arguments : |
| 2863 | <backend> is the name of the backend to use. |
| 2864 | |
| 2865 | When doing content-switching between frontend and backends using the |
| 2866 | "use_backend" keyword, it is often useful to indicate which backend will be |
| 2867 | used when no rule has matched. It generally is the dynamic backend which |
| 2868 | will catch all undetermined requests. |
| 2869 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2870 | Example : |
| 2871 | |
| 2872 | use_backend dynamic if url_dyn |
| 2873 | use_backend static if url_css url_img extension_img |
| 2874 | default_backend dynamic |
| 2875 | |
Willy Tarreau | 98d0485 | 2015-05-26 12:18:29 +0200 | [diff] [blame] | 2876 | See also : "use_backend" |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 2877 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2878 | |
Baptiste Assmann | 27f5134 | 2013-10-09 06:51:49 +0200 | [diff] [blame] | 2879 | description <string> |
| 2880 | Describe a listen, frontend or backend. |
| 2881 | May be used in sections : defaults | frontend | listen | backend |
| 2882 | no | yes | yes | yes |
| 2883 | Arguments : string |
| 2884 | |
| 2885 | Allows to add a sentence to describe the related object in the HAProxy HTML |
| 2886 | stats page. The description will be printed on the right of the object name |
| 2887 | it describes. |
| 2888 | No need to backslash spaces in the <string> arguments. |
| 2889 | |
| 2890 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2891 | disabled |
| 2892 | Disable a proxy, frontend or backend. |
| 2893 | May be used in sections : defaults | frontend | listen | backend |
| 2894 | yes | yes | yes | yes |
| 2895 | Arguments : none |
| 2896 | |
| 2897 | The "disabled" keyword is used to disable an instance, mainly in order to |
| 2898 | liberate a listening port or to temporarily disable a service. The instance |
| 2899 | will still be created and its configuration will be checked, but it will be |
| 2900 | created in the "stopped" state and will appear as such in the statistics. It |
| 2901 | will not receive any traffic nor will it send any health-checks or logs. It |
| 2902 | is possible to disable many instances at once by adding the "disabled" |
| 2903 | keyword in a "defaults" section. |
| 2904 | |
| 2905 | See also : "enabled" |
| 2906 | |
| 2907 | |
Willy Tarreau | 5ce9457 | 2010-06-07 14:35:41 +0200 | [diff] [blame] | 2908 | dispatch <address>:<port> |
| 2909 | Set a default server address |
| 2910 | May be used in sections : defaults | frontend | listen | backend |
| 2911 | no | no | yes | yes |
Cyril Bonté | 108cf6e | 2012-04-21 23:30:29 +0200 | [diff] [blame] | 2912 | Arguments : |
Willy Tarreau | 5ce9457 | 2010-06-07 14:35:41 +0200 | [diff] [blame] | 2913 | |
| 2914 | <address> is the IPv4 address of the default server. Alternatively, a |
| 2915 | resolvable hostname is supported, but this name will be resolved |
| 2916 | during start-up. |
| 2917 | |
| 2918 | <ports> is a mandatory port specification. All connections will be sent |
| 2919 | to this port, and it is not permitted to use port offsets as is |
| 2920 | possible with normal servers. |
| 2921 | |
Willy Tarreau | 787aed5 | 2011-04-15 06:45:37 +0200 | [diff] [blame] | 2922 | The "dispatch" keyword designates a default server for use when no other |
Willy Tarreau | 5ce9457 | 2010-06-07 14:35:41 +0200 | [diff] [blame] | 2923 | server can take the connection. In the past it was used to forward non |
| 2924 | persistent connections to an auxiliary load balancer. Due to its simple |
| 2925 | syntax, it has also been used for simple TCP relays. It is recommended not to |
| 2926 | use it for more clarity, and to use the "server" directive instead. |
| 2927 | |
| 2928 | See also : "server" |
| 2929 | |
| 2930 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2931 | enabled |
| 2932 | Enable a proxy, frontend or backend. |
| 2933 | May be used in sections : defaults | frontend | listen | backend |
| 2934 | yes | yes | yes | yes |
| 2935 | Arguments : none |
| 2936 | |
| 2937 | The "enabled" keyword is used to explicitly enable an instance, when the |
| 2938 | defaults has been set to "disabled". This is very rarely used. |
| 2939 | |
| 2940 | See also : "disabled" |
| 2941 | |
| 2942 | |
| 2943 | errorfile <code> <file> |
| 2944 | Return a file contents instead of errors generated by HAProxy |
| 2945 | May be used in sections : defaults | frontend | listen | backend |
| 2946 | yes | yes | yes | yes |
| 2947 | Arguments : |
| 2948 | <code> is the HTTP status code. Currently, HAProxy is capable of |
CJ Ess | 108b1dd | 2015-04-07 12:03:37 -0400 | [diff] [blame] | 2949 | generating codes 200, 400, 403, 405, 408, 429, 500, 502, 503, and |
| 2950 | 504. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2951 | |
| 2952 | <file> designates a file containing the full HTTP response. It is |
Willy Tarreau | d2a4aa2 | 2008-01-31 15:28:22 +0100 | [diff] [blame] | 2953 | recommended to follow the common practice of appending ".http" to |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2954 | the filename so that people do not confuse the response with HTML |
Willy Tarreau | 59140a2 | 2009-02-22 12:02:30 +0100 | [diff] [blame] | 2955 | error pages, and to use absolute paths, since files are read |
| 2956 | before any chroot is performed. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2957 | |
| 2958 | It is important to understand that this keyword is not meant to rewrite |
| 2959 | errors returned by the server, but errors detected and returned by HAProxy. |
| 2960 | This is why the list of supported errors is limited to a small set. |
| 2961 | |
Willy Tarreau | ae94d4d | 2011-05-11 16:28:49 +0200 | [diff] [blame] | 2962 | Code 200 is emitted in response to requests matching a "monitor-uri" rule. |
| 2963 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2964 | The files are returned verbatim on the TCP socket. This allows any trick such |
| 2965 | as redirections to another URL or site, as well as tricks to clean cookies, |
| 2966 | force enable or disable caching, etc... The package provides default error |
| 2967 | files returning the same contents as default errors. |
| 2968 | |
Willy Tarreau | 59140a2 | 2009-02-22 12:02:30 +0100 | [diff] [blame] | 2969 | The files should not exceed the configured buffer size (BUFSIZE), which |
| 2970 | generally is 8 or 16 kB, otherwise they will be truncated. It is also wise |
| 2971 | not to put any reference to local contents (eg: images) in order to avoid |
| 2972 | loops between the client and HAProxy when all servers are down, causing an |
| 2973 | error to be returned instead of an image. For better HTTP compliance, it is |
| 2974 | recommended that all header lines end with CR-LF and not LF alone. |
| 2975 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2976 | The files are read at the same time as the configuration and kept in memory. |
| 2977 | For this reason, the errors continue to be returned even when the process is |
| 2978 | chrooted, and no file change is considered while the process is running. A |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 2979 | simple method for developing those files consists in associating them to the |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 2980 | 403 status code and interrogating a blocked URL. |
| 2981 | |
| 2982 | See also : "errorloc", "errorloc302", "errorloc303" |
| 2983 | |
Willy Tarreau | 59140a2 | 2009-02-22 12:02:30 +0100 | [diff] [blame] | 2984 | Example : |
| 2985 | errorfile 400 /etc/haproxy/errorfiles/400badreq.http |
Willy Tarreau | 989222a | 2016-01-15 10:26:26 +0100 | [diff] [blame] | 2986 | errorfile 408 /dev/null # work around Chrome pre-connect bug |
Willy Tarreau | 59140a2 | 2009-02-22 12:02:30 +0100 | [diff] [blame] | 2987 | errorfile 403 /etc/haproxy/errorfiles/403forbid.http |
| 2988 | errorfile 503 /etc/haproxy/errorfiles/503sorry.http |
| 2989 | |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 2990 | |
| 2991 | errorloc <code> <url> |
| 2992 | errorloc302 <code> <url> |
| 2993 | Return an HTTP redirection to a URL instead of errors generated by HAProxy |
| 2994 | May be used in sections : defaults | frontend | listen | backend |
| 2995 | yes | yes | yes | yes |
| 2996 | Arguments : |
| 2997 | <code> is the HTTP status code. Currently, HAProxy is capable of |
Willy Tarreau | ae94d4d | 2011-05-11 16:28:49 +0200 | [diff] [blame] | 2998 | generating codes 200, 400, 403, 408, 500, 502, 503, and 504. |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 2999 | |
| 3000 | <url> it is the exact contents of the "Location" header. It may contain |
| 3001 | either a relative URI to an error page hosted on the same site, |
| 3002 | or an absolute URI designating an error page on another site. |
| 3003 | Special care should be given to relative URIs to avoid redirect |
| 3004 | loops if the URI itself may generate the same error (eg: 500). |
| 3005 | |
| 3006 | It is important to understand that this keyword is not meant to rewrite |
| 3007 | errors returned by the server, but errors detected and returned by HAProxy. |
| 3008 | This is why the list of supported errors is limited to a small set. |
| 3009 | |
Willy Tarreau | ae94d4d | 2011-05-11 16:28:49 +0200 | [diff] [blame] | 3010 | Code 200 is emitted in response to requests matching a "monitor-uri" rule. |
| 3011 | |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 3012 | Note that both keyword return the HTTP 302 status code, which tells the |
| 3013 | client to fetch the designated URL using the same HTTP method. This can be |
| 3014 | quite problematic in case of non-GET methods such as POST, because the URL |
| 3015 | sent to the client might not be allowed for something other than GET. To |
Willy Tarreau | 989222a | 2016-01-15 10:26:26 +0100 | [diff] [blame] | 3016 | work around this problem, please use "errorloc303" which send the HTTP 303 |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 3017 | status code, indicating to the client that the URL must be fetched with a GET |
| 3018 | request. |
| 3019 | |
| 3020 | See also : "errorfile", "errorloc303" |
| 3021 | |
| 3022 | |
| 3023 | errorloc303 <code> <url> |
| 3024 | Return an HTTP redirection to a URL instead of errors generated by HAProxy |
| 3025 | May be used in sections : defaults | frontend | listen | backend |
| 3026 | yes | yes | yes | yes |
| 3027 | Arguments : |
| 3028 | <code> is the HTTP status code. Currently, HAProxy is capable of |
| 3029 | generating codes 400, 403, 408, 500, 502, 503, and 504. |
| 3030 | |
| 3031 | <url> it is the exact contents of the "Location" header. It may contain |
| 3032 | either a relative URI to an error page hosted on the same site, |
| 3033 | or an absolute URI designating an error page on another site. |
| 3034 | Special care should be given to relative URIs to avoid redirect |
| 3035 | loops if the URI itself may generate the same error (eg: 500). |
| 3036 | |
| 3037 | It is important to understand that this keyword is not meant to rewrite |
| 3038 | errors returned by the server, but errors detected and returned by HAProxy. |
| 3039 | This is why the list of supported errors is limited to a small set. |
| 3040 | |
Willy Tarreau | ae94d4d | 2011-05-11 16:28:49 +0200 | [diff] [blame] | 3041 | Code 200 is emitted in response to requests matching a "monitor-uri" rule. |
| 3042 | |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 3043 | Note that both keyword return the HTTP 303 status code, which tells the |
| 3044 | client to fetch the designated URL using the same HTTP GET method. This |
| 3045 | solves the usual problems associated with "errorloc" and the 302 code. It is |
| 3046 | possible that some very old browsers designed before HTTP/1.1 do not support |
Willy Tarreau | d2a4aa2 | 2008-01-31 15:28:22 +0100 | [diff] [blame] | 3047 | it, but no such problem has been reported till now. |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 3048 | |
| 3049 | See also : "errorfile", "errorloc", "errorloc302" |
| 3050 | |
| 3051 | |
Simon Horman | 51a1cf6 | 2015-02-03 13:00:44 +0900 | [diff] [blame] | 3052 | email-alert from <emailaddr> |
| 3053 | Declare the from email address to be used in both the envelope and header |
| 3054 | of email alerts. This is the address that email alerts are sent from. |
| 3055 | May be used in sections: defaults | frontend | listen | backend |
| 3056 | yes | yes | yes | yes |
| 3057 | |
| 3058 | Arguments : |
| 3059 | |
| 3060 | <emailaddr> is the from email address to use when sending email alerts |
| 3061 | |
| 3062 | Also requires "email-alert mailers" and "email-alert to" to be set |
| 3063 | and if so sending email alerts is enabled for the proxy. |
| 3064 | |
Simon Horman | 64e3416 | 2015-02-06 11:11:57 +0900 | [diff] [blame] | 3065 | See also : "email-alert level", "email-alert mailers", |
Cyril Bonté | 307ee1e | 2015-09-28 23:16:06 +0200 | [diff] [blame] | 3066 | "email-alert myhostname", "email-alert to", section 3.6 about |
| 3067 | mailers. |
Simon Horman | 64e3416 | 2015-02-06 11:11:57 +0900 | [diff] [blame] | 3068 | |
| 3069 | |
| 3070 | email-alert level <level> |
| 3071 | Declare the maximum log level of messages for which email alerts will be |
| 3072 | sent. This acts as a filter on the sending of email alerts. |
| 3073 | May be used in sections: defaults | frontend | listen | backend |
| 3074 | yes | yes | yes | yes |
| 3075 | |
| 3076 | Arguments : |
| 3077 | |
| 3078 | <level> One of the 8 syslog levels: |
| 3079 | emerg alert crit err warning notice info debug |
| 3080 | The above syslog levels are ordered from lowest to highest. |
| 3081 | |
| 3082 | By default level is alert |
| 3083 | |
| 3084 | Also requires "email-alert from", "email-alert mailers" and |
| 3085 | "email-alert to" to be set and if so sending email alerts is enabled |
| 3086 | for the proxy. |
| 3087 | |
Simon Horman | 1421e21 | 2015-04-30 13:10:35 +0900 | [diff] [blame] | 3088 | Alerts are sent when : |
| 3089 | |
| 3090 | * An un-paused server is marked as down and <level> is alert or lower |
| 3091 | * A paused server is marked as down and <level> is notice or lower |
| 3092 | * A server is marked as up or enters the drain state and <level> |
| 3093 | is notice or lower |
| 3094 | * "option log-health-checks" is enabled, <level> is info or lower, |
| 3095 | and a health check status update occurs |
| 3096 | |
Simon Horman | 64e3416 | 2015-02-06 11:11:57 +0900 | [diff] [blame] | 3097 | See also : "email-alert from", "email-alert mailers", |
| 3098 | "email-alert myhostname", "email-alert to", |
Simon Horman | 51a1cf6 | 2015-02-03 13:00:44 +0900 | [diff] [blame] | 3099 | section 3.6 about mailers. |
| 3100 | |
| 3101 | |
| 3102 | email-alert mailers <mailersect> |
| 3103 | Declare the mailers to be used when sending email alerts |
| 3104 | May be used in sections: defaults | frontend | listen | backend |
| 3105 | yes | yes | yes | yes |
| 3106 | |
| 3107 | Arguments : |
| 3108 | |
| 3109 | <mailersect> is the name of the mailers section to send email alerts. |
| 3110 | |
| 3111 | Also requires "email-alert from" and "email-alert to" to be set |
| 3112 | and if so sending email alerts is enabled for the proxy. |
| 3113 | |
Simon Horman | 64e3416 | 2015-02-06 11:11:57 +0900 | [diff] [blame] | 3114 | See also : "email-alert from", "email-alert level", "email-alert myhostname", |
| 3115 | "email-alert to", section 3.6 about mailers. |
Simon Horman | 51a1cf6 | 2015-02-03 13:00:44 +0900 | [diff] [blame] | 3116 | |
| 3117 | |
| 3118 | email-alert myhostname <hostname> |
| 3119 | Declare the to hostname address to be used when communicating with |
| 3120 | mailers. |
| 3121 | May be used in sections: defaults | frontend | listen | backend |
| 3122 | yes | yes | yes | yes |
| 3123 | |
| 3124 | Arguments : |
| 3125 | |
Baptiste Assmann | 738bad9 | 2015-12-21 15:27:53 +0100 | [diff] [blame] | 3126 | <hostname> is the hostname to use when communicating with mailers |
Simon Horman | 51a1cf6 | 2015-02-03 13:00:44 +0900 | [diff] [blame] | 3127 | |
| 3128 | By default the systems hostname is used. |
| 3129 | |
| 3130 | Also requires "email-alert from", "email-alert mailers" and |
| 3131 | "email-alert to" to be set and if so sending email alerts is enabled |
| 3132 | for the proxy. |
| 3133 | |
Simon Horman | 64e3416 | 2015-02-06 11:11:57 +0900 | [diff] [blame] | 3134 | See also : "email-alert from", "email-alert level", "email-alert mailers", |
| 3135 | "email-alert to", section 3.6 about mailers. |
Simon Horman | 51a1cf6 | 2015-02-03 13:00:44 +0900 | [diff] [blame] | 3136 | |
| 3137 | |
| 3138 | email-alert to <emailaddr> |
| 3139 | Declare both the recipent address in the envelope and to address in the |
| 3140 | header of email alerts. This is the address that email alerts are sent to. |
| 3141 | May be used in sections: defaults | frontend | listen | backend |
| 3142 | yes | yes | yes | yes |
| 3143 | |
| 3144 | Arguments : |
| 3145 | |
| 3146 | <emailaddr> is the to email address to use when sending email alerts |
| 3147 | |
| 3148 | Also requires "email-alert mailers" and "email-alert to" to be set |
| 3149 | and if so sending email alerts is enabled for the proxy. |
| 3150 | |
Simon Horman | 64e3416 | 2015-02-06 11:11:57 +0900 | [diff] [blame] | 3151 | See also : "email-alert from", "email-alert level", "email-alert mailers", |
Simon Horman | 51a1cf6 | 2015-02-03 13:00:44 +0900 | [diff] [blame] | 3152 | "email-alert myhostname", section 3.6 about mailers. |
| 3153 | |
| 3154 | |
Willy Tarreau | 4de9149 | 2010-01-22 19:10:05 +0100 | [diff] [blame] | 3155 | force-persist { if | unless } <condition> |
| 3156 | Declare a condition to force persistence on down servers |
| 3157 | May be used in sections: defaults | frontend | listen | backend |
| 3158 | no | yes | yes | yes |
| 3159 | |
| 3160 | By default, requests are not dispatched to down servers. It is possible to |
| 3161 | force this using "option persist", but it is unconditional and redispatches |
| 3162 | to a valid server if "option redispatch" is set. That leaves with very little |
| 3163 | possibilities to force some requests to reach a server which is artificially |
| 3164 | marked down for maintenance operations. |
| 3165 | |
| 3166 | The "force-persist" statement allows one to declare various ACL-based |
| 3167 | conditions which, when met, will cause a request to ignore the down status of |
| 3168 | a server and still try to connect to it. That makes it possible to start a |
| 3169 | server, still replying an error to the health checks, and run a specially |
| 3170 | configured browser to test the service. Among the handy methods, one could |
| 3171 | use a specific source IP address, or a specific cookie. The cookie also has |
| 3172 | the advantage that it can easily be added/removed on the browser from a test |
| 3173 | page. Once the service is validated, it is then possible to open the service |
| 3174 | to the world by returning a valid response to health checks. |
| 3175 | |
| 3176 | The forced persistence is enabled when an "if" condition is met, or unless an |
| 3177 | "unless" condition is met. The final redispatch is always disabled when this |
| 3178 | is used. |
| 3179 | |
Cyril Bonté | 0d4bf01 | 2010-04-25 23:21:46 +0200 | [diff] [blame] | 3180 | See also : "option redispatch", "ignore-persist", "persist", |
Cyril Bonté | a8e7bbc | 2010-04-25 22:29:29 +0200 | [diff] [blame] | 3181 | and section 7 about ACL usage. |
Willy Tarreau | 4de9149 | 2010-01-22 19:10:05 +0100 | [diff] [blame] | 3182 | |
Christopher Faulet | c3fe533 | 2016-04-07 15:30:10 +0200 | [diff] [blame] | 3183 | |
| 3184 | filter <name> [param*] |
| 3185 | Add the filter <name> in the filter list attached to the proxy. |
| 3186 | May be used in sections : defaults | frontend | listen | backend |
| 3187 | no | yes | yes | yes |
| 3188 | Arguments : |
| 3189 | <name> is the name of the filter. Officially supported filters are |
| 3190 | referenced in section 9. |
| 3191 | |
| 3192 | <param*> is a list of parameters accpeted by the filter <name>. The |
| 3193 | parsing of these parameters are the responsibility of the |
| 3194 | filter. Please refer to the documention of the corresponding |
| 3195 | filter (section 9) from all details on the supported parameters. |
| 3196 | |
| 3197 | Multiple occurrences of the filter line can be used for the same proxy. The |
| 3198 | same filter can be referenced many times if needed. |
| 3199 | |
| 3200 | Example: |
| 3201 | listen |
| 3202 | bind *:80 |
| 3203 | |
| 3204 | filter trace name BEFORE-HTTP-COMP |
| 3205 | filter compression |
| 3206 | filter trace name AFTER-HTTP-COMP |
| 3207 | |
| 3208 | compression algo gzip |
| 3209 | compression offload |
| 3210 | |
| 3211 | server srv1 192.168.0.1:80 |
| 3212 | |
| 3213 | See also : section 9. |
| 3214 | |
Willy Tarreau | 4de9149 | 2010-01-22 19:10:05 +0100 | [diff] [blame] | 3215 | |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 3216 | fullconn <conns> |
| 3217 | Specify at what backend load the servers will reach their maxconn |
| 3218 | May be used in sections : defaults | frontend | listen | backend |
| 3219 | yes | no | yes | yes |
| 3220 | Arguments : |
| 3221 | <conns> is the number of connections on the backend which will make the |
| 3222 | servers use the maximal number of connections. |
| 3223 | |
Willy Tarreau | 198a744 | 2008-01-17 12:05:32 +0100 | [diff] [blame] | 3224 | When a server has a "maxconn" parameter specified, it means that its number |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 3225 | of concurrent connections will never go higher. Additionally, if it has a |
Willy Tarreau | 198a744 | 2008-01-17 12:05:32 +0100 | [diff] [blame] | 3226 | "minconn" parameter, it indicates a dynamic limit following the backend's |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 3227 | load. The server will then always accept at least <minconn> connections, |
| 3228 | never more than <maxconn>, and the limit will be on the ramp between both |
| 3229 | values when the backend has less than <conns> concurrent connections. This |
| 3230 | makes it possible to limit the load on the servers during normal loads, but |
| 3231 | push it further for important loads without overloading the servers during |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 3232 | exceptional loads. |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 3233 | |
Willy Tarreau | fbb7842 | 2011-06-05 15:38:35 +0200 | [diff] [blame] | 3234 | Since it's hard to get this value right, haproxy automatically sets it to |
| 3235 | 10% of the sum of the maxconns of all frontends that may branch to this |
Bertrand Jacquin | 702d44f | 2013-11-19 11:43:06 +0100 | [diff] [blame] | 3236 | backend (based on "use_backend" and "default_backend" rules). That way it's |
| 3237 | safe to leave it unset. However, "use_backend" involving dynamic names are |
| 3238 | not counted since there is no way to know if they could match or not. |
Willy Tarreau | fbb7842 | 2011-06-05 15:38:35 +0200 | [diff] [blame] | 3239 | |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 3240 | Example : |
| 3241 | # The servers will accept between 100 and 1000 concurrent connections each |
| 3242 | # and the maximum of 1000 will be reached when the backend reaches 10000 |
| 3243 | # connections. |
| 3244 | backend dynamic |
| 3245 | fullconn 10000 |
| 3246 | server srv1 dyn1:80 minconn 100 maxconn 1000 |
| 3247 | server srv2 dyn2:80 minconn 100 maxconn 1000 |
| 3248 | |
| 3249 | See also : "maxconn", "server" |
| 3250 | |
| 3251 | |
| 3252 | grace <time> |
| 3253 | Maintain a proxy operational for some time after a soft stop |
| 3254 | May be used in sections : defaults | frontend | listen | backend |
Cyril Bonté | 99ed327 | 2010-01-24 23:29:44 +0100 | [diff] [blame] | 3255 | yes | yes | yes | yes |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 3256 | Arguments : |
| 3257 | <time> is the time (by default in milliseconds) for which the instance |
| 3258 | will remain operational with the frontend sockets still listening |
| 3259 | when a soft-stop is received via the SIGUSR1 signal. |
| 3260 | |
| 3261 | This may be used to ensure that the services disappear in a certain order. |
| 3262 | This was designed so that frontends which are dedicated to monitoring by an |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 3263 | external equipment fail immediately while other ones remain up for the time |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 3264 | needed by the equipment to detect the failure. |
| 3265 | |
| 3266 | Note that currently, there is very little benefit in using this parameter, |
| 3267 | and it may in fact complicate the soft-reconfiguration process more than |
| 3268 | simplify it. |
| 3269 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 3270 | |
Bhaskar Maddala | b6c0ac9 | 2013-11-05 11:54:02 -0500 | [diff] [blame] | 3271 | hash-type <method> <function> <modifier> |
Willy Tarreau | 6b2e11b | 2009-10-01 07:52:15 +0200 | [diff] [blame] | 3272 | Specify a method to use for mapping hashes to servers |
| 3273 | May be used in sections : defaults | frontend | listen | backend |
| 3274 | yes | no | yes | yes |
| 3275 | Arguments : |
Bhaskar | 98634f0 | 2013-10-29 23:30:51 -0400 | [diff] [blame] | 3276 | <method> is the method used to select a server from the hash computed by |
| 3277 | the <function> : |
Willy Tarreau | 6b2e11b | 2009-10-01 07:52:15 +0200 | [diff] [blame] | 3278 | |
Bhaskar | 98634f0 | 2013-10-29 23:30:51 -0400 | [diff] [blame] | 3279 | map-based the hash table is a static array containing all alive servers. |
| 3280 | The hashes will be very smooth, will consider weights, but |
| 3281 | will be static in that weight changes while a server is up |
| 3282 | will be ignored. This means that there will be no slow start. |
| 3283 | Also, since a server is selected by its position in the array, |
| 3284 | most mappings are changed when the server count changes. This |
| 3285 | means that when a server goes up or down, or when a server is |
| 3286 | added to a farm, most connections will be redistributed to |
| 3287 | different servers. This can be inconvenient with caches for |
| 3288 | instance. |
Willy Tarreau | 798a39c | 2010-11-24 15:04:29 +0100 | [diff] [blame] | 3289 | |
Bhaskar | 98634f0 | 2013-10-29 23:30:51 -0400 | [diff] [blame] | 3290 | consistent the hash table is a tree filled with many occurrences of each |
| 3291 | server. The hash key is looked up in the tree and the closest |
| 3292 | server is chosen. This hash is dynamic, it supports changing |
| 3293 | weights while the servers are up, so it is compatible with the |
| 3294 | slow start feature. It has the advantage that when a server |
| 3295 | goes up or down, only its associations are moved. When a |
| 3296 | server is added to the farm, only a few part of the mappings |
| 3297 | are redistributed, making it an ideal method for caches. |
| 3298 | However, due to its principle, the distribution will never be |
| 3299 | very smooth and it may sometimes be necessary to adjust a |
| 3300 | server's weight or its ID to get a more balanced distribution. |
| 3301 | In order to get the same distribution on multiple load |
| 3302 | balancers, it is important that all servers have the exact |
Bhaskar Maddala | b6c0ac9 | 2013-11-05 11:54:02 -0500 | [diff] [blame] | 3303 | same IDs. Note: consistent hash uses sdbm and avalanche if no |
| 3304 | hash function is specified. |
Bhaskar | 98634f0 | 2013-10-29 23:30:51 -0400 | [diff] [blame] | 3305 | |
| 3306 | <function> is the hash function to be used : |
| 3307 | |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 3308 | sdbm this function was created initially for sdbm (a public-domain |
Bhaskar | 98634f0 | 2013-10-29 23:30:51 -0400 | [diff] [blame] | 3309 | reimplementation of ndbm) database library. It was found to do |
| 3310 | well in scrambling bits, causing better distribution of the keys |
| 3311 | and fewer splits. It also happens to be a good general hashing |
Bhaskar Maddala | b6c0ac9 | 2013-11-05 11:54:02 -0500 | [diff] [blame] | 3312 | function with good distribution, unless the total server weight |
| 3313 | is a multiple of 64, in which case applying the avalanche |
| 3314 | modifier may help. |
Bhaskar | 98634f0 | 2013-10-29 23:30:51 -0400 | [diff] [blame] | 3315 | |
| 3316 | djb2 this function was first proposed by Dan Bernstein many years ago |
| 3317 | on comp.lang.c. Studies have shown that for certain workload this |
Bhaskar Maddala | b6c0ac9 | 2013-11-05 11:54:02 -0500 | [diff] [blame] | 3318 | function provides a better distribution than sdbm. It generally |
| 3319 | works well with text-based inputs though it can perform extremely |
| 3320 | poorly with numeric-only input or when the total server weight is |
| 3321 | a multiple of 33, unless the avalanche modifier is also used. |
| 3322 | |
Willy Tarreau | a0f4271 | 2013-11-14 14:30:35 +0100 | [diff] [blame] | 3323 | wt6 this function was designed for haproxy while testing other |
| 3324 | functions in the past. It is not as smooth as the other ones, but |
| 3325 | is much less sensible to the input data set or to the number of |
| 3326 | servers. It can make sense as an alternative to sdbm+avalanche or |
| 3327 | djb2+avalanche for consistent hashing or when hashing on numeric |
| 3328 | data such as a source IP address or a visitor identifier in a URL |
| 3329 | parameter. |
| 3330 | |
Willy Tarreau | 324f07f | 2015-01-20 19:44:50 +0100 | [diff] [blame] | 3331 | crc32 this is the most common CRC32 implementation as used in Ethernet, |
| 3332 | gzip, PNG, etc. It is slower than the other ones but may provide |
| 3333 | a better distribution or less predictable results especially when |
| 3334 | used on strings. |
| 3335 | |
Bhaskar Maddala | b6c0ac9 | 2013-11-05 11:54:02 -0500 | [diff] [blame] | 3336 | <modifier> indicates an optional method applied after hashing the key : |
| 3337 | |
| 3338 | avalanche This directive indicates that the result from the hash |
| 3339 | function above should not be used in its raw form but that |
| 3340 | a 4-byte full avalanche hash must be applied first. The |
| 3341 | purpose of this step is to mix the resulting bits from the |
| 3342 | previous hash in order to avoid any undesired effect when |
| 3343 | the input contains some limited values or when the number of |
| 3344 | servers is a multiple of one of the hash's components (64 |
| 3345 | for SDBM, 33 for DJB2). Enabling avalanche tends to make the |
| 3346 | result less predictable, but it's also not as smooth as when |
| 3347 | using the original function. Some testing might be needed |
| 3348 | with some workloads. This hash is one of the many proposed |
| 3349 | by Bob Jenkins. |
Willy Tarreau | 6b2e11b | 2009-10-01 07:52:15 +0200 | [diff] [blame] | 3350 | |
Bhaskar | 98634f0 | 2013-10-29 23:30:51 -0400 | [diff] [blame] | 3351 | The default hash type is "map-based" and is recommended for most usages. The |
| 3352 | default function is "sdbm", the selection of a function should be based on |
| 3353 | the range of the values being hashed. |
Willy Tarreau | 6b2e11b | 2009-10-01 07:52:15 +0200 | [diff] [blame] | 3354 | |
| 3355 | See also : "balance", "server" |
| 3356 | |
| 3357 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 3358 | http-check disable-on-404 |
| 3359 | Enable a maintenance mode upon HTTP/404 response to health-checks |
| 3360 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 3361 | yes | no | yes | yes |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 3362 | Arguments : none |
| 3363 | |
| 3364 | When this option is set, a server which returns an HTTP code 404 will be |
| 3365 | excluded from further load-balancing, but will still receive persistent |
| 3366 | connections. This provides a very convenient method for Web administrators |
| 3367 | to perform a graceful shutdown of their servers. It is also important to note |
| 3368 | that a server which is detected as failed while it was in this mode will not |
| 3369 | generate an alert, just a notice. If the server responds 2xx or 3xx again, it |
| 3370 | will immediately be reinserted into the farm. The status on the stats page |
| 3371 | reports "NOLB" for a server in this mode. It is important to note that this |
Willy Tarreau | bd74154 | 2010-03-16 18:46:54 +0100 | [diff] [blame] | 3372 | option only works in conjunction with the "httpchk" option. If this option |
| 3373 | is used with "http-check expect", then it has precedence over it so that 404 |
| 3374 | responses will still be considered as soft-stop. |
| 3375 | |
| 3376 | See also : "option httpchk", "http-check expect" |
| 3377 | |
| 3378 | |
| 3379 | http-check expect [!] <match> <pattern> |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 3380 | Make HTTP health checks consider response contents or specific status codes |
Willy Tarreau | bd74154 | 2010-03-16 18:46:54 +0100 | [diff] [blame] | 3381 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | 1ee51a6 | 2011-08-19 20:04:17 +0200 | [diff] [blame] | 3382 | yes | no | yes | yes |
Willy Tarreau | bd74154 | 2010-03-16 18:46:54 +0100 | [diff] [blame] | 3383 | Arguments : |
| 3384 | <match> is a keyword indicating how to look for a specific pattern in the |
| 3385 | response. The keyword may be one of "status", "rstatus", |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 3386 | "string", or "rstring". The keyword may be preceded by an |
Willy Tarreau | bd74154 | 2010-03-16 18:46:54 +0100 | [diff] [blame] | 3387 | exclamation mark ("!") to negate the match. Spaces are allowed |
| 3388 | between the exclamation mark and the keyword. See below for more |
| 3389 | details on the supported keywords. |
| 3390 | |
| 3391 | <pattern> is the pattern to look for. It may be a string or a regular |
| 3392 | expression. If the pattern contains spaces, they must be escaped |
| 3393 | with the usual backslash ('\'). |
| 3394 | |
| 3395 | By default, "option httpchk" considers that response statuses 2xx and 3xx |
| 3396 | are valid, and that others are invalid. When "http-check expect" is used, |
| 3397 | it defines what is considered valid or invalid. Only one "http-check" |
| 3398 | statement is supported in a backend. If a server fails to respond or times |
| 3399 | out, the check obviously fails. The available matches are : |
| 3400 | |
| 3401 | status <string> : test the exact string match for the HTTP status code. |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 3402 | A health check response will be considered valid if the |
Willy Tarreau | bd74154 | 2010-03-16 18:46:54 +0100 | [diff] [blame] | 3403 | response's status code is exactly this string. If the |
| 3404 | "status" keyword is prefixed with "!", then the response |
| 3405 | will be considered invalid if the status code matches. |
| 3406 | |
| 3407 | rstatus <regex> : test a regular expression for the HTTP status code. |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 3408 | A health check response will be considered valid if the |
Willy Tarreau | bd74154 | 2010-03-16 18:46:54 +0100 | [diff] [blame] | 3409 | response's status code matches the expression. If the |
| 3410 | "rstatus" keyword is prefixed with "!", then the response |
| 3411 | will be considered invalid if the status code matches. |
| 3412 | This is mostly used to check for multiple codes. |
| 3413 | |
| 3414 | string <string> : test the exact string match in the HTTP response body. |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 3415 | A health check response will be considered valid if the |
Willy Tarreau | bd74154 | 2010-03-16 18:46:54 +0100 | [diff] [blame] | 3416 | response's body contains this exact string. If the |
| 3417 | "string" keyword is prefixed with "!", then the response |
| 3418 | will be considered invalid if the body contains this |
| 3419 | string. This can be used to look for a mandatory word at |
| 3420 | the end of a dynamic page, or to detect a failure when a |
| 3421 | specific error appears on the check page (eg: a stack |
| 3422 | trace). |
| 3423 | |
| 3424 | rstring <regex> : test a regular expression on the HTTP response body. |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 3425 | A health check response will be considered valid if the |
Willy Tarreau | bd74154 | 2010-03-16 18:46:54 +0100 | [diff] [blame] | 3426 | response's body matches this expression. If the "rstring" |
| 3427 | keyword is prefixed with "!", then the response will be |
| 3428 | considered invalid if the body matches the expression. |
| 3429 | This can be used to look for a mandatory word at the end |
| 3430 | of a dynamic page, or to detect a failure when a specific |
| 3431 | error appears on the check page (eg: a stack trace). |
| 3432 | |
| 3433 | It is important to note that the responses will be limited to a certain size |
| 3434 | defined by the global "tune.chksize" option, which defaults to 16384 bytes. |
| 3435 | Thus, too large responses may not contain the mandatory pattern when using |
| 3436 | "string" or "rstring". If a large response is absolutely required, it is |
| 3437 | possible to change the default max size by setting the global variable. |
| 3438 | However, it is worth keeping in mind that parsing very large responses can |
| 3439 | waste some CPU cycles, especially when regular expressions are used, and that |
| 3440 | it is always better to focus the checks on smaller resources. |
| 3441 | |
Cyril Bonté | 32602d2 | 2015-01-30 00:07:07 +0100 | [diff] [blame] | 3442 | Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it |
| 3443 | will automatically append a "Connection: close" header, meaning that this |
| 3444 | header should not be present in the request provided by "option httpchk". |
| 3445 | |
Willy Tarreau | bd74154 | 2010-03-16 18:46:54 +0100 | [diff] [blame] | 3446 | Last, if "http-check expect" is combined with "http-check disable-on-404", |
| 3447 | then this last one has precedence when the server responds with 404. |
| 3448 | |
| 3449 | Examples : |
| 3450 | # only accept status 200 as valid |
Willy Tarreau | 8f2a1e7 | 2011-01-06 16:36:10 +0100 | [diff] [blame] | 3451 | http-check expect status 200 |
Willy Tarreau | bd74154 | 2010-03-16 18:46:54 +0100 | [diff] [blame] | 3452 | |
| 3453 | # consider SQL errors as errors |
Willy Tarreau | 8f2a1e7 | 2011-01-06 16:36:10 +0100 | [diff] [blame] | 3454 | http-check expect ! string SQL\ Error |
Willy Tarreau | bd74154 | 2010-03-16 18:46:54 +0100 | [diff] [blame] | 3455 | |
| 3456 | # consider status 5xx only as errors |
Willy Tarreau | 8f2a1e7 | 2011-01-06 16:36:10 +0100 | [diff] [blame] | 3457 | http-check expect ! rstatus ^5 |
Willy Tarreau | bd74154 | 2010-03-16 18:46:54 +0100 | [diff] [blame] | 3458 | |
| 3459 | # check that we have a correct hexadecimal tag before /html |
Willy Tarreau | 8f2a1e7 | 2011-01-06 16:36:10 +0100 | [diff] [blame] | 3460 | http-check expect rstring <!--tag:[0-9a-f]*</html> |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 3461 | |
Willy Tarreau | bd74154 | 2010-03-16 18:46:54 +0100 | [diff] [blame] | 3462 | See also : "option httpchk", "http-check disable-on-404" |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 3463 | |
| 3464 | |
Willy Tarreau | ef78104 | 2010-01-27 11:53:01 +0100 | [diff] [blame] | 3465 | http-check send-state |
| 3466 | Enable emission of a state header with HTTP health checks |
| 3467 | May be used in sections : defaults | frontend | listen | backend |
| 3468 | yes | no | yes | yes |
| 3469 | Arguments : none |
| 3470 | |
| 3471 | When this option is set, haproxy will systematically send a special header |
| 3472 | "X-Haproxy-Server-State" with a list of parameters indicating to each server |
| 3473 | how they are seen by haproxy. This can be used for instance when a server is |
| 3474 | manipulated without access to haproxy and the operator needs to know whether |
| 3475 | haproxy still sees it up or not, or if the server is the last one in a farm. |
| 3476 | |
| 3477 | The header is composed of fields delimited by semi-colons, the first of which |
| 3478 | is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid |
| 3479 | checks on the total number before transition, just as appears in the stats |
| 3480 | interface. Next headers are in the form "<variable>=<value>", indicating in |
| 3481 | no specific order some values available in the stats interface : |
Joseph Lynch | 514061c | 2015-01-15 17:52:59 -0800 | [diff] [blame] | 3482 | - a variable "address", containing the address of the backend server. |
| 3483 | This corresponds to the <address> field in the server declaration. For |
| 3484 | unix domain sockets, it will read "unix". |
| 3485 | |
| 3486 | - a variable "port", containing the port of the backend server. This |
| 3487 | corresponds to the <port> field in the server declaration. For unix |
| 3488 | domain sockets, it will read "unix". |
| 3489 | |
Willy Tarreau | ef78104 | 2010-01-27 11:53:01 +0100 | [diff] [blame] | 3490 | - a variable "name", containing the name of the backend followed by a slash |
| 3491 | ("/") then the name of the server. This can be used when a server is |
| 3492 | checked in multiple backends. |
| 3493 | |
| 3494 | - a variable "node" containing the name of the haproxy node, as set in the |
| 3495 | global "node" variable, otherwise the system's hostname if unspecified. |
| 3496 | |
| 3497 | - a variable "weight" indicating the weight of the server, a slash ("/") |
| 3498 | and the total weight of the farm (just counting usable servers). This |
| 3499 | helps to know if other servers are available to handle the load when this |
| 3500 | one fails. |
| 3501 | |
| 3502 | - a variable "scur" indicating the current number of concurrent connections |
| 3503 | on the server, followed by a slash ("/") then the total number of |
| 3504 | connections on all servers of the same backend. |
| 3505 | |
| 3506 | - a variable "qcur" indicating the current number of requests in the |
| 3507 | server's queue. |
| 3508 | |
| 3509 | Example of a header received by the application server : |
| 3510 | >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \ |
| 3511 | scur=13/22; qcur=0 |
| 3512 | |
| 3513 | See also : "option httpchk", "http-check disable-on-404" |
| 3514 | |
Willy Tarreau | ccbcc37 | 2012-12-27 12:37:57 +0100 | [diff] [blame] | 3515 | http-request { allow | deny | tarpit | auth [realm <realm>] | redirect <rule> | |
Willy Tarreau | f4c43c1 | 2013-06-11 17:01:13 +0200 | [diff] [blame] | 3516 | add-header <name> <fmt> | set-header <name> <fmt> | |
Thierry FOURNIER | 82bf70d | 2015-05-26 17:58:29 +0200 | [diff] [blame] | 3517 | capture <sample> [ len <length> | id <id> ] | |
Thierry FOURNIER | dad3d1d | 2014-04-22 18:07:25 +0200 | [diff] [blame] | 3518 | del-header <name> | set-nice <nice> | set-log-level <level> | |
Sasha Pachev | 218f064 | 2014-06-16 12:05:59 -0600 | [diff] [blame] | 3519 | replace-header <name> <match-regex> <replace-fmt> | |
| 3520 | replace-value <name> <match-regex> <replace-fmt> | |
Willy Tarreau | a0dc23f | 2015-01-22 20:46:11 +0100 | [diff] [blame] | 3521 | set-method <fmt> | set-path <fmt> | set-query <fmt> | |
| 3522 | set-uri <fmt> | set-tos <tos> | set-mark <mark> | |
Baptiste Assmann | fabcbe0 | 2014-04-24 22:16:59 +0200 | [diff] [blame] | 3523 | add-acl(<file name>) <key fmt> | |
| 3524 | del-acl(<file name>) <key fmt> | |
| 3525 | del-map(<file name>) <key fmt> | |
Baptiste Assmann | bb7e86a | 2014-09-03 18:29:47 +0200 | [diff] [blame] | 3526 | set-map(<file name>) <key fmt> <value fmt> | |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 3527 | set-var(<var name>) <expr> | |
Thierry FOURNIER | 90da191 | 2015-03-05 11:17:06 +0100 | [diff] [blame] | 3528 | { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] | |
Thierry FOURNIER | e0627bd | 2015-08-04 08:20:33 +0200 | [diff] [blame] | 3529 | sc-inc-gpc0(<sc-id>) | |
Thierry FOURNIER | 236657b | 2015-08-19 08:25:14 +0200 | [diff] [blame] | 3530 | sc-set-gpt0(<sc-id>) <int> | |
Willy Tarreau | 2d392c2 | 2015-08-24 01:43:45 +0200 | [diff] [blame] | 3531 | silent-drop | |
Baptiste Assmann | fabcbe0 | 2014-04-24 22:16:59 +0200 | [diff] [blame] | 3532 | } |
Cyril Bonté | f0c6061 | 2010-02-06 14:44:47 +0100 | [diff] [blame] | 3533 | [ { if | unless } <condition> ] |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 3534 | Access control for Layer 7 requests |
| 3535 | |
| 3536 | May be used in sections: defaults | frontend | listen | backend |
| 3537 | no | yes | yes | yes |
| 3538 | |
Willy Tarreau | 20b0de5 | 2012-12-24 15:45:22 +0100 | [diff] [blame] | 3539 | The http-request statement defines a set of rules which apply to layer 7 |
| 3540 | processing. The rules are evaluated in their declaration order when they are |
| 3541 | met in a frontend, listen or backend section. Any rule may optionally be |
| 3542 | followed by an ACL-based condition, in which case it will only be evaluated |
| 3543 | if the condition is true. |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 3544 | |
Willy Tarreau | 20b0de5 | 2012-12-24 15:45:22 +0100 | [diff] [blame] | 3545 | The first keyword is the rule's action. Currently supported actions include : |
| 3546 | - "allow" : this stops the evaluation of the rules and lets the request |
| 3547 | pass the check. No further "http-request" rules are evaluated. |
| 3548 | |
| 3549 | - "deny" : this stops the evaluation of the rules and immediately rejects |
| 3550 | the request and emits an HTTP 403 error. No further "http-request" rules |
| 3551 | are evaluated. |
| 3552 | |
Willy Tarreau | ccbcc37 | 2012-12-27 12:37:57 +0100 | [diff] [blame] | 3553 | - "tarpit" : this stops the evaluation of the rules and immediately blocks |
| 3554 | the request without responding for a delay specified by "timeout tarpit" |
| 3555 | or "timeout connect" if the former is not set. After that delay, if the |
| 3556 | client is still connected, an HTTP error 500 is returned so that the |
| 3557 | client does not suspect it has been tarpitted. Logs will report the flags |
| 3558 | "PT". The goal of the tarpit rule is to slow down robots during an attack |
| 3559 | when they're limited on the number of concurrent requests. It can be very |
| 3560 | efficient against very dumb robots, and will significantly reduce the |
| 3561 | load on firewalls compared to a "deny" rule. But when facing "correctly" |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 3562 | developed robots, it can make things worse by forcing haproxy and the |
Willy Tarreau | 2d392c2 | 2015-08-24 01:43:45 +0200 | [diff] [blame] | 3563 | front firewall to support insane number of concurrent connections. See |
| 3564 | also the "silent-drop" action below. |
Willy Tarreau | ccbcc37 | 2012-12-27 12:37:57 +0100 | [diff] [blame] | 3565 | |
Willy Tarreau | 20b0de5 | 2012-12-24 15:45:22 +0100 | [diff] [blame] | 3566 | - "auth" : this stops the evaluation of the rules and immediately responds |
| 3567 | with an HTTP 401 or 407 error code to invite the user to present a valid |
| 3568 | user name and password. No further "http-request" rules are evaluated. An |
| 3569 | optional "realm" parameter is supported, it sets the authentication realm |
| 3570 | that is returned with the response (typically the application's name). |
| 3571 | |
Willy Tarreau | 81499eb | 2012-12-27 12:19:02 +0100 | [diff] [blame] | 3572 | - "redirect" : this performs an HTTP redirection based on a redirect rule. |
| 3573 | This is exactly the same as the "redirect" statement except that it |
| 3574 | inserts a redirect rule which can be processed in the middle of other |
Thierry FOURNIER | d18cd0f | 2013-11-29 12:15:45 +0100 | [diff] [blame] | 3575 | "http-request" rules and that these rules use the "log-format" strings. |
| 3576 | See the "redirect" keyword for the rule's syntax. |
Willy Tarreau | 81499eb | 2012-12-27 12:19:02 +0100 | [diff] [blame] | 3577 | |
Willy Tarreau | 20b0de5 | 2012-12-24 15:45:22 +0100 | [diff] [blame] | 3578 | - "add-header" appends an HTTP header field whose name is specified in |
| 3579 | <name> and whose value is defined by <fmt> which follows the log-format |
| 3580 | rules (see Custom Log Format in section 8.2.4). This is particularly |
| 3581 | useful to pass connection-specific information to the server (eg: the |
| 3582 | client's SSL certificate), or to combine several headers into one. This |
| 3583 | rule is not final, so it is possible to add other similar rules. Note |
| 3584 | that header addition is performed immediately, so one rule might reuse |
| 3585 | the resulting header from a previous rule. |
| 3586 | |
| 3587 | - "set-header" does the same as "add-header" except that the header name |
| 3588 | is first removed if it existed. This is useful when passing security |
| 3589 | information to the server, where the header must not be manipulated by |
Willy Tarreau | 8560328 | 2015-01-21 20:39:27 +0100 | [diff] [blame] | 3590 | external users. Note that the new value is computed before the removal so |
| 3591 | it is possible to concatenate a value to an existing header. |
Willy Tarreau | 20b0de5 | 2012-12-24 15:45:22 +0100 | [diff] [blame] | 3592 | |
Thierry FOURNIER | dad3d1d | 2014-04-22 18:07:25 +0200 | [diff] [blame] | 3593 | - "del-header" removes all HTTP header fields whose name is specified in |
| 3594 | <name>. |
| 3595 | |
Sasha Pachev | 218f064 | 2014-06-16 12:05:59 -0600 | [diff] [blame] | 3596 | - "replace-header" matches the regular expression in all occurrences of |
| 3597 | header field <name> according to <match-regex>, and replaces them with |
| 3598 | the <replace-fmt> argument. Format characters are allowed in replace-fmt |
| 3599 | and work like in <fmt> arguments in "add-header". The match is only |
| 3600 | case-sensitive. It is important to understand that this action only |
| 3601 | considers whole header lines, regardless of the number of values they |
| 3602 | may contain. This usage is suited to headers naturally containing commas |
| 3603 | in their value, such as If-Modified-Since and so on. |
| 3604 | |
| 3605 | Example: |
| 3606 | |
| 3607 | http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2 |
| 3608 | |
| 3609 | applied to: |
| 3610 | |
| 3611 | Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT; |
| 3612 | |
| 3613 | outputs: |
| 3614 | |
| 3615 | Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT; |
| 3616 | |
| 3617 | assuming the backend IP is 192.168.1.20 |
| 3618 | |
| 3619 | - "replace-value" works like "replace-header" except that it matches the |
| 3620 | regex against every comma-delimited value of the header field <name> |
| 3621 | instead of the entire header. This is suited for all headers which are |
| 3622 | allowed to carry more than one value. An example could be the Accept |
| 3623 | header. |
| 3624 | |
| 3625 | Example: |
| 3626 | |
| 3627 | http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1 |
| 3628 | |
| 3629 | applied to: |
| 3630 | |
| 3631 | X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37 |
| 3632 | |
| 3633 | outputs: |
| 3634 | |
| 3635 | X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37 |
| 3636 | |
Willy Tarreau | a0dc23f | 2015-01-22 20:46:11 +0100 | [diff] [blame] | 3637 | - "set-method" rewrites the request method with the result of the |
| 3638 | evaluation of format string <fmt>. There should be very few valid reasons |
| 3639 | for having to do so as this is more likely to break something than to fix |
| 3640 | it. |
| 3641 | |
| 3642 | - "set-path" rewrites the request path with the result of the evaluation of |
| 3643 | format string <fmt>. The query string, if any, is left intact. If a |
| 3644 | scheme and authority is found before the path, they are left intact as |
| 3645 | well. If the request doesn't have a path ("*"), this one is replaced with |
| 3646 | the format. This can be used to prepend a directory component in front of |
| 3647 | a path for example. See also "set-query" and "set-uri". |
| 3648 | |
| 3649 | Example : |
| 3650 | # prepend the host name before the path |
| 3651 | http-request set-path /%[hdr(host)]%[path] |
| 3652 | |
| 3653 | - "set-query" rewrites the request's query string which appears after the |
| 3654 | first question mark ("?") with the result of the evaluation of format |
| 3655 | string <fmt>. The part prior to the question mark is left intact. If the |
| 3656 | request doesn't contain a question mark and the new value is not empty, |
| 3657 | then one is added at the end of the URI, followed by the new value. If |
| 3658 | a question mark was present, it will never be removed even if the value |
| 3659 | is empty. This can be used to add or remove parameters from the query |
| 3660 | string. See also "set-query" and "set-uri". |
| 3661 | |
| 3662 | Example : |
| 3663 | # replace "%3D" with "=" in the query string |
| 3664 | http-request set-query %[query,regsub(%3D,=,g)] |
| 3665 | |
| 3666 | - "set-uri" rewrites the request URI with the result of the evaluation of |
| 3667 | format string <fmt>. The scheme, authority, path and query string are all |
| 3668 | replaced at once. This can be used to rewrite hosts in front of proxies, |
| 3669 | or to perform complex modifications to the URI such as moving parts |
| 3670 | between the path and the query string. See also "set-path" and |
| 3671 | "set-query". |
| 3672 | |
Willy Tarreau | f4c43c1 | 2013-06-11 17:01:13 +0200 | [diff] [blame] | 3673 | - "set-nice" sets the "nice" factor of the current request being processed. |
| 3674 | It only has effect against the other requests being processed at the same |
| 3675 | time. The default value is 0, unless altered by the "nice" setting on the |
| 3676 | "bind" line. The accepted range is -1024..1024. The higher the value, the |
| 3677 | nicest the request will be. Lower values will make the request more |
| 3678 | important than other ones. This can be useful to improve the speed of |
| 3679 | some requests, or lower the priority of non-important requests. Using |
| 3680 | this setting without prior experimentation can cause some major slowdown. |
| 3681 | |
Willy Tarreau | 9a355ec | 2013-06-11 17:45:46 +0200 | [diff] [blame] | 3682 | - "set-log-level" is used to change the log level of the current request |
| 3683 | when a certain condition is met. Valid levels are the 8 syslog levels |
| 3684 | (see the "log" keyword) plus the special level "silent" which disables |
| 3685 | logging for this request. This rule is not final so the last matching |
| 3686 | rule wins. This rule can be useful to disable health checks coming from |
| 3687 | another equipment. |
| 3688 | |
Willy Tarreau | 42cf39e | 2013-06-11 18:51:32 +0200 | [diff] [blame] | 3689 | - "set-tos" is used to set the TOS or DSCP field value of packets sent to |
| 3690 | the client to the value passed in <tos> on platforms which support this. |
| 3691 | This value represents the whole 8 bits of the IP TOS field, and can be |
| 3692 | expressed both in decimal or hexadecimal format (prefixed by "0x"). Note |
| 3693 | that only the 6 higher bits are used in DSCP or TOS, and the two lower |
| 3694 | bits are always 0. This can be used to adjust some routing behaviour on |
| 3695 | border routers based on some information from the request. See RFC 2474, |
| 3696 | 2597, 3260 and 4594 for more information. |
| 3697 | |
Willy Tarreau | 51347ed | 2013-06-11 19:34:13 +0200 | [diff] [blame] | 3698 | - "set-mark" is used to set the Netfilter MARK on all packets sent to the |
| 3699 | client to the value passed in <mark> on platforms which support it. This |
| 3700 | value is an unsigned 32 bit value which can be matched by netfilter and |
| 3701 | by the routing table. It can be expressed both in decimal or hexadecimal |
| 3702 | format (prefixed by "0x"). This can be useful to force certain packets to |
| 3703 | take a different route (for example a cheaper network path for bulk |
| 3704 | downloads). This works on Linux kernels 2.6.32 and above and requires |
| 3705 | admin privileges. |
| 3706 | |
Baptiste Assmann | fabcbe0 | 2014-04-24 22:16:59 +0200 | [diff] [blame] | 3707 | - "add-acl" is used to add a new entry into an ACL. The ACL must be loaded |
| 3708 | from a file (even a dummy empty file). The file name of the ACL to be |
| 3709 | updated is passed between parentheses. It takes one argument: <key fmt>, |
| 3710 | which follows log-format rules, to collect content of the new entry. It |
| 3711 | performs a lookup in the ACL before insertion, to avoid duplicated (or |
| 3712 | more) values. This lookup is done by a linear search and can be expensive |
| 3713 | with large lists! It is the equivalent of the "add acl" command from the |
| 3714 | stats socket, but can be triggered by an HTTP request. |
| 3715 | |
| 3716 | - "del-acl" is used to delete an entry from an ACL. The ACL must be loaded |
| 3717 | from a file (even a dummy empty file). The file name of the ACL to be |
| 3718 | updated is passed between parentheses. It takes one argument: <key fmt>, |
| 3719 | which follows log-format rules, to collect content of the entry to delete. |
| 3720 | It is the equivalent of the "del acl" command from the stats socket, but |
| 3721 | can be triggered by an HTTP request. |
| 3722 | |
| 3723 | - "del-map" is used to delete an entry from a MAP. The MAP must be loaded |
| 3724 | from a file (even a dummy empty file). The file name of the MAP to be |
| 3725 | updated is passed between parentheses. It takes one argument: <key fmt>, |
| 3726 | which follows log-format rules, to collect content of the entry to delete. |
| 3727 | It takes one argument: "file name" It is the equivalent of the "del map" |
| 3728 | command from the stats socket, but can be triggered by an HTTP request. |
| 3729 | |
| 3730 | - "set-map" is used to add a new entry into a MAP. The MAP must be loaded |
| 3731 | from a file (even a dummy empty file). The file name of the MAP to be |
| 3732 | updated is passed between parentheses. It takes 2 arguments: <key fmt>, |
| 3733 | which follows log-format rules, used to collect MAP key, and <value fmt>, |
| 3734 | which follows log-format rules, used to collect content for the new entry. |
| 3735 | It performs a lookup in the MAP before insertion, to avoid duplicated (or |
| 3736 | more) values. This lookup is done by a linear search and can be expensive |
| 3737 | with large lists! It is the equivalent of the "set map" command from the |
| 3738 | stats socket, but can be triggered by an HTTP request. |
| 3739 | |
Thierry FOURNIER | 82bf70d | 2015-05-26 17:58:29 +0200 | [diff] [blame] | 3740 | - capture <sample> [ len <length> | id <id> ] : |
Willy Tarreau | a9083d0 | 2015-05-08 15:27:59 +0200 | [diff] [blame] | 3741 | captures sample expression <sample> from the request buffer, and converts |
| 3742 | it to a string of at most <len> characters. The resulting string is |
| 3743 | stored into the next request "capture" slot, so it will possibly appear |
| 3744 | next to some captured HTTP headers. It will then automatically appear in |
| 3745 | the logs, and it will be possible to extract it using sample fetch rules |
| 3746 | to feed it into headers or anything. The length should be limited given |
| 3747 | that this size will be allocated for each capture during the whole |
| 3748 | session life. Please check section 7.3 (Fetching samples) and "capture |
| 3749 | request header" for more information. |
| 3750 | |
Thierry FOURNIER | 82bf70d | 2015-05-26 17:58:29 +0200 | [diff] [blame] | 3751 | If the keyword "id" is used instead of "len", the action tries to store |
| 3752 | the captured string in a previously declared capture slot. This is useful |
| 3753 | to run captures in backends. The slot id can be declared by a previous |
| 3754 | directive "http-request capture" or with the "declare capture" keyword. |
Baptiste Assmann | e954493 | 2015-11-03 23:31:35 +0100 | [diff] [blame] | 3755 | If the slot <id> doesn't exist, then HAProxy fails parsing the |
| 3756 | configuration to prevent unexpected behavior at run time. |
Thierry FOURNIER | 82bf70d | 2015-05-26 17:58:29 +0200 | [diff] [blame] | 3757 | |
Willy Tarreau | 09448f7 | 2014-06-25 18:12:15 +0200 | [diff] [blame] | 3758 | - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] : |
| 3759 | enables tracking of sticky counters from current request. These rules |
| 3760 | do not stop evaluation and do not change default action. Three sets of |
| 3761 | counters may be simultaneously tracked by the same connection. The first |
| 3762 | "track-sc0" rule executed enables tracking of the counters of the |
| 3763 | specified table as the first set. The first "track-sc1" rule executed |
| 3764 | enables tracking of the counters of the specified table as the second |
| 3765 | set. The first "track-sc2" rule executed enables tracking of the |
| 3766 | counters of the specified table as the third set. It is a recommended |
| 3767 | practice to use the first set of counters for the per-frontend counters |
| 3768 | and the second set for the per-backend ones. But this is just a |
| 3769 | guideline, all may be used everywhere. |
| 3770 | |
| 3771 | These actions take one or two arguments : |
| 3772 | <key> is mandatory, and is a sample expression rule as described |
| 3773 | in section 7.3. It describes what elements of the incoming |
| 3774 | request or connection will be analysed, extracted, combined, |
| 3775 | and used to select which table entry to update the counters. |
| 3776 | |
| 3777 | <table> is an optional table to be used instead of the default one, |
| 3778 | which is the stick-table declared in the current proxy. All |
| 3779 | the counters for the matches and updates for the key will |
| 3780 | then be performed in that table until the session ends. |
| 3781 | |
| 3782 | Once a "track-sc*" rule is executed, the key is looked up in the table |
| 3783 | and if it is not found, an entry is allocated for it. Then a pointer to |
| 3784 | that entry is kept during all the session's life, and this entry's |
| 3785 | counters are updated as often as possible, every time the session's |
| 3786 | counters are updated, and also systematically when the session ends. |
| 3787 | Counters are only updated for events that happen after the tracking has |
| 3788 | been started. As an exception, connection counters and request counters |
| 3789 | are systematically updated so that they reflect useful information. |
| 3790 | |
| 3791 | If the entry tracks concurrent connection counters, one connection is |
| 3792 | counted for as long as the entry is tracked, and the entry will not |
| 3793 | expire during that time. Tracking counters also provides a performance |
| 3794 | advantage over just checking the keys, because only one table lookup is |
| 3795 | performed for all ACL checks that make use of it. |
| 3796 | |
Thierry FOURNIER | 236657b | 2015-08-19 08:25:14 +0200 | [diff] [blame] | 3797 | - sc-set-gpt0(<sc-id>) <int> : |
| 3798 | This action sets the GPT0 tag according to the sticky counter designated |
| 3799 | by <sc-id> and the value of <int>. The expected result is a boolean. If |
| 3800 | an error occurs, this action silently fails and the actions evaluation |
| 3801 | continues. |
| 3802 | |
Thierry FOURNIER | e0627bd | 2015-08-04 08:20:33 +0200 | [diff] [blame] | 3803 | - sc-inc-gpc0(<sc-id>): |
| 3804 | This action increments the GPC0 counter according with the sticky counter |
| 3805 | designated by <sc-id>. If an error occurs, this action silently fails and |
| 3806 | the actions evaluation continues. |
| 3807 | |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 3808 | - set-var(<var-name>) <expr> : |
| 3809 | Is used to set the contents of a variable. The variable is declared |
| 3810 | inline. |
| 3811 | |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 3812 | <var-name> The name of the variable starts with an indication about |
| 3813 | its scope. The scopes allowed are: |
| 3814 | "sess" : the variable is shared with the whole session |
| 3815 | "txn" : the variable is shared with the transaction |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 3816 | (request and response) |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 3817 | "req" : the variable is shared only during request |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 3818 | processing |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 3819 | "res" : the variable is shared only during response |
| 3820 | processing |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 3821 | This prefix is followed by a name. The separator is a '.'. |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 3822 | The name may only contain characters 'a-z', 'A-Z', '0-9' |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 3823 | and '_'. |
| 3824 | |
| 3825 | <expr> Is a standard HAProxy expression formed by a sample-fetch |
| 3826 | followed by some converters. |
| 3827 | |
| 3828 | Example: |
| 3829 | |
| 3830 | http-request set-var(req.my_var) req.fhdr(user-agent),lower |
| 3831 | |
Adis Nezirovic | 2fbcafc | 2015-07-06 15:44:30 +0200 | [diff] [blame] | 3832 | - set-src <expr> : |
| 3833 | Is used to set the source IP address to the value of specified |
| 3834 | expression. Useful when a proxy in front of HAProxy rewrites source IP, |
| 3835 | but provides the correct IP in a HTTP header; or you want to mask |
| 3836 | source IP for privacy. |
| 3837 | |
| 3838 | <expr> Is a standard HAProxy expression formed by a sample-fetch |
| 3839 | followed by some converters. |
| 3840 | |
| 3841 | Example: |
| 3842 | |
| 3843 | http-request set-src hdr(x-forwarded-for) |
| 3844 | http-request set-src src,ipmask(24) |
| 3845 | |
| 3846 | When set-src is successful, the source port is set to 0. |
| 3847 | |
Willy Tarreau | 2d392c2 | 2015-08-24 01:43:45 +0200 | [diff] [blame] | 3848 | - "silent-drop" : this stops the evaluation of the rules and makes the |
| 3849 | client-facing connection suddenly disappear using a system-dependant way |
| 3850 | that tries to prevent the client from being notified. The effect it then |
| 3851 | that the client still sees an established connection while there's none |
| 3852 | on HAProxy. The purpose is to achieve a comparable effect to "tarpit" |
| 3853 | except that it doesn't use any local resource at all on the machine |
| 3854 | running HAProxy. It can resist much higher loads than "tarpit", and slow |
| 3855 | down stronger attackers. It is important to undestand the impact of using |
| 3856 | this mechanism. All stateful equipments placed between the client and |
| 3857 | HAProxy (firewalls, proxies, load balancers) will also keep the |
| 3858 | established connection for a long time and may suffer from this action. |
| 3859 | On modern Linux systems running with enough privileges, the TCP_REPAIR |
| 3860 | socket option is used to block the emission of a TCP reset. On other |
| 3861 | systems, the socket's TTL is reduced to 1 so that the TCP reset doesn't |
| 3862 | pass the first router, though it's still delivered to local networks. Do |
| 3863 | not use it unless you fully understand how it works. |
| 3864 | |
Willy Tarreau | 20b0de5 | 2012-12-24 15:45:22 +0100 | [diff] [blame] | 3865 | There is no limit to the number of http-request statements per instance. |
| 3866 | |
| 3867 | It is important to know that http-request rules are processed very early in |
| 3868 | the HTTP processing, just after "block" rules and before "reqdel" or "reqrep" |
Ruoshan Huang | eb5a363 | 2015-12-08 21:00:23 +0800 | [diff] [blame] | 3869 | or "reqadd" rules. That way, headers added by "add-header"/"set-header" are |
| 3870 | visible by almost all further ACL rules. |
| 3871 | |
| 3872 | Using "reqadd"/"reqdel"/"reqrep" to manipulate request headers is discouraged |
| 3873 | in newer versions (>= 1.5). But if you need to use regular expression to |
| 3874 | delete headers, you can still use "reqdel". Also please use |
| 3875 | "http-request deny/allow/tarpit" instead of "reqdeny"/"reqpass"/"reqtarpit". |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 3876 | |
| 3877 | Example: |
Cyril Bonté | 78caf84 | 2010-03-10 22:41:43 +0100 | [diff] [blame] | 3878 | acl nagios src 192.168.129.3 |
| 3879 | acl local_net src 192.168.0.0/16 |
| 3880 | acl auth_ok http_auth(L1) |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 3881 | |
Cyril Bonté | 78caf84 | 2010-03-10 22:41:43 +0100 | [diff] [blame] | 3882 | http-request allow if nagios |
| 3883 | http-request allow if local_net auth_ok |
| 3884 | http-request auth realm Gimme if local_net auth_ok |
| 3885 | http-request deny |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 3886 | |
Cyril Bonté | 78caf84 | 2010-03-10 22:41:43 +0100 | [diff] [blame] | 3887 | Example: |
| 3888 | acl auth_ok http_auth_group(L1) G1 |
Cyril Bonté | 78caf84 | 2010-03-10 22:41:43 +0100 | [diff] [blame] | 3889 | http-request auth unless auth_ok |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 3890 | |
Willy Tarreau | 20b0de5 | 2012-12-24 15:45:22 +0100 | [diff] [blame] | 3891 | Example: |
| 3892 | http-request set-header X-Haproxy-Current-Date %T |
| 3893 | http-request set-header X-SSL %[ssl_fc] |
Willy Tarreau | fca4261 | 2015-08-27 17:15:05 +0200 | [diff] [blame] | 3894 | http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex] |
Willy Tarreau | 20b0de5 | 2012-12-24 15:45:22 +0100 | [diff] [blame] | 3895 | http-request set-header X-SSL-Client-Verify %[ssl_c_verify] |
| 3896 | http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn] |
| 3897 | http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)] |
| 3898 | http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn] |
| 3899 | http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore] |
| 3900 | http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter] |
| 3901 | |
Baptiste Assmann | fabcbe0 | 2014-04-24 22:16:59 +0200 | [diff] [blame] | 3902 | Example: |
| 3903 | acl key req.hdr(X-Add-Acl-Key) -m found |
| 3904 | acl add path /addacl |
| 3905 | acl del path /delacl |
| 3906 | |
| 3907 | acl myhost hdr(Host) -f myhost.lst |
| 3908 | |
| 3909 | http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add |
| 3910 | http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del |
| 3911 | |
| 3912 | Example: |
| 3913 | acl value req.hdr(X-Value) -m found |
| 3914 | acl setmap path /setmap |
| 3915 | acl delmap path /delmap |
| 3916 | |
| 3917 | use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found } |
| 3918 | |
| 3919 | http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value |
| 3920 | http-request del-map(map.lst) %[src] if delmap |
| 3921 | |
Cyril Bonté | 2be1b3f | 2010-09-30 23:46:30 +0200 | [diff] [blame] | 3922 | See also : "stats http-request", section 3.4 about userlists and section 7 |
| 3923 | about ACL usage. |
Willy Tarreau | ef78104 | 2010-01-27 11:53:01 +0100 | [diff] [blame] | 3924 | |
Willy Tarreau | f4c43c1 | 2013-06-11 17:01:13 +0200 | [diff] [blame] | 3925 | http-response { allow | deny | add-header <name> <fmt> | set-nice <nice> | |
Willy Tarreau | 51d861a | 2015-05-22 17:30:48 +0200 | [diff] [blame] | 3926 | capture <sample> id <id> | redirect <rule> | |
Thierry FOURNIER | dad3d1d | 2014-04-22 18:07:25 +0200 | [diff] [blame] | 3927 | set-header <name> <fmt> | del-header <name> | |
Sasha Pachev | 218f064 | 2014-06-16 12:05:59 -0600 | [diff] [blame] | 3928 | replace-header <name> <regex-match> <replace-fmt> | |
| 3929 | replace-value <name> <regex-match> <replace-fmt> | |
Thierry FOURNIER | 35d70ef | 2015-08-26 16:21:56 +0200 | [diff] [blame] | 3930 | set-status <status> | |
Baptiste Assmann | fabcbe0 | 2014-04-24 22:16:59 +0200 | [diff] [blame] | 3931 | set-log-level <level> | set-mark <mark> | set-tos <tos> | |
| 3932 | add-acl(<file name>) <key fmt> | |
| 3933 | del-acl(<file name>) <key fmt> | |
| 3934 | del-map(<file name>) <key fmt> | |
Thierry FOURNIER | 90da191 | 2015-03-05 11:17:06 +0100 | [diff] [blame] | 3935 | set-map(<file name>) <key fmt> <value fmt> | |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 3936 | set-var(<var-name>) <expr> | |
Thierry FOURNIER | e0627bd | 2015-08-04 08:20:33 +0200 | [diff] [blame] | 3937 | sc-inc-gpc0(<sc-id>) | |
Thierry FOURNIER | 236657b | 2015-08-19 08:25:14 +0200 | [diff] [blame] | 3938 | sc-set-gpt0(<sc-id>) <int> | |
Willy Tarreau | 2d392c2 | 2015-08-24 01:43:45 +0200 | [diff] [blame] | 3939 | silent-drop | |
Baptiste Assmann | fabcbe0 | 2014-04-24 22:16:59 +0200 | [diff] [blame] | 3940 | } |
Lukas Tribus | 2dd1d1a | 2013-06-19 23:34:41 +0200 | [diff] [blame] | 3941 | [ { if | unless } <condition> ] |
Willy Tarreau | e365c0b | 2013-06-11 16:06:12 +0200 | [diff] [blame] | 3942 | Access control for Layer 7 responses |
| 3943 | |
| 3944 | May be used in sections: defaults | frontend | listen | backend |
| 3945 | no | yes | yes | yes |
| 3946 | |
| 3947 | The http-response statement defines a set of rules which apply to layer 7 |
| 3948 | processing. The rules are evaluated in their declaration order when they are |
| 3949 | met in a frontend, listen or backend section. Any rule may optionally be |
| 3950 | followed by an ACL-based condition, in which case it will only be evaluated |
| 3951 | if the condition is true. Since these rules apply on responses, the backend |
| 3952 | rules are applied first, followed by the frontend's rules. |
| 3953 | |
| 3954 | The first keyword is the rule's action. Currently supported actions include : |
| 3955 | - "allow" : this stops the evaluation of the rules and lets the response |
| 3956 | pass the check. No further "http-response" rules are evaluated for the |
| 3957 | current section. |
| 3958 | |
| 3959 | - "deny" : this stops the evaluation of the rules and immediately rejects |
| 3960 | the response and emits an HTTP 502 error. No further "http-response" |
| 3961 | rules are evaluated. |
| 3962 | |
| 3963 | - "add-header" appends an HTTP header field whose name is specified in |
| 3964 | <name> and whose value is defined by <fmt> which follows the log-format |
| 3965 | rules (see Custom Log Format in section 8.2.4). This may be used to send |
| 3966 | a cookie to a client for example, or to pass some internal information. |
| 3967 | This rule is not final, so it is possible to add other similar rules. |
| 3968 | Note that header addition is performed immediately, so one rule might |
| 3969 | reuse the resulting header from a previous rule. |
| 3970 | |
| 3971 | - "set-header" does the same as "add-header" except that the header name |
| 3972 | is first removed if it existed. This is useful when passing security |
| 3973 | information to the server, where the header must not be manipulated by |
| 3974 | external users. |
| 3975 | |
Thierry FOURNIER | dad3d1d | 2014-04-22 18:07:25 +0200 | [diff] [blame] | 3976 | - "del-header" removes all HTTP header fields whose name is specified in |
| 3977 | <name>. |
| 3978 | |
Sasha Pachev | 218f064 | 2014-06-16 12:05:59 -0600 | [diff] [blame] | 3979 | - "replace-header" matches the regular expression in all occurrences of |
| 3980 | header field <name> according to <match-regex>, and replaces them with |
| 3981 | the <replace-fmt> argument. Format characters are allowed in replace-fmt |
| 3982 | and work like in <fmt> arguments in "add-header". The match is only |
| 3983 | case-sensitive. It is important to understand that this action only |
| 3984 | considers whole header lines, regardless of the number of values they |
| 3985 | may contain. This usage is suited to headers naturally containing commas |
| 3986 | in their value, such as Set-Cookie, Expires and so on. |
| 3987 | |
| 3988 | Example: |
| 3989 | |
| 3990 | http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2 |
| 3991 | |
| 3992 | applied to: |
| 3993 | |
| 3994 | Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT |
| 3995 | |
| 3996 | outputs: |
| 3997 | |
| 3998 | Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT |
| 3999 | |
| 4000 | assuming the backend IP is 192.168.1.20. |
| 4001 | |
| 4002 | - "replace-value" works like "replace-header" except that it matches the |
| 4003 | regex against every comma-delimited value of the header field <name> |
| 4004 | instead of the entire header. This is suited for all headers which are |
| 4005 | allowed to carry more than one value. An example could be the Accept |
| 4006 | header. |
| 4007 | |
| 4008 | Example: |
| 4009 | |
| 4010 | http-response replace-value Cache-control ^public$ private |
| 4011 | |
| 4012 | applied to: |
| 4013 | |
| 4014 | Cache-Control: max-age=3600, public |
| 4015 | |
| 4016 | outputs: |
| 4017 | |
| 4018 | Cache-Control: max-age=3600, private |
| 4019 | |
Thierry FOURNIER | 35d70ef | 2015-08-26 16:21:56 +0200 | [diff] [blame] | 4020 | - "set-status" replaces the response status code with <status> which must |
| 4021 | be an integer between 100 and 999. Note that the reason is automatically |
| 4022 | adapted to the new code. |
| 4023 | |
| 4024 | Example: |
| 4025 | |
| 4026 | # return "431 Request Header Fields Too Large" |
| 4027 | http-response set-status 431 |
| 4028 | |
Willy Tarreau | f4c43c1 | 2013-06-11 17:01:13 +0200 | [diff] [blame] | 4029 | - "set-nice" sets the "nice" factor of the current request being processed. |
| 4030 | It only has effect against the other requests being processed at the same |
| 4031 | time. The default value is 0, unless altered by the "nice" setting on the |
| 4032 | "bind" line. The accepted range is -1024..1024. The higher the value, the |
| 4033 | nicest the request will be. Lower values will make the request more |
| 4034 | important than other ones. This can be useful to improve the speed of |
| 4035 | some requests, or lower the priority of non-important requests. Using |
| 4036 | this setting without prior experimentation can cause some major slowdown. |
| 4037 | |
Willy Tarreau | 9a355ec | 2013-06-11 17:45:46 +0200 | [diff] [blame] | 4038 | - "set-log-level" is used to change the log level of the current request |
| 4039 | when a certain condition is met. Valid levels are the 8 syslog levels |
| 4040 | (see the "log" keyword) plus the special level "silent" which disables |
| 4041 | logging for this request. This rule is not final so the last matching |
| 4042 | rule wins. This rule can be useful to disable health checks coming from |
| 4043 | another equipment. |
| 4044 | |
Willy Tarreau | 42cf39e | 2013-06-11 18:51:32 +0200 | [diff] [blame] | 4045 | - "set-tos" is used to set the TOS or DSCP field value of packets sent to |
| 4046 | the client to the value passed in <tos> on platforms which support this. |
| 4047 | This value represents the whole 8 bits of the IP TOS field, and can be |
| 4048 | expressed both in decimal or hexadecimal format (prefixed by "0x"). Note |
| 4049 | that only the 6 higher bits are used in DSCP or TOS, and the two lower |
| 4050 | bits are always 0. This can be used to adjust some routing behaviour on |
| 4051 | border routers based on some information from the request. See RFC 2474, |
| 4052 | 2597, 3260 and 4594 for more information. |
| 4053 | |
Willy Tarreau | 51347ed | 2013-06-11 19:34:13 +0200 | [diff] [blame] | 4054 | - "set-mark" is used to set the Netfilter MARK on all packets sent to the |
| 4055 | client to the value passed in <mark> on platforms which support it. This |
| 4056 | value is an unsigned 32 bit value which can be matched by netfilter and |
| 4057 | by the routing table. It can be expressed both in decimal or hexadecimal |
| 4058 | format (prefixed by "0x"). This can be useful to force certain packets to |
| 4059 | take a different route (for example a cheaper network path for bulk |
| 4060 | downloads). This works on Linux kernels 2.6.32 and above and requires |
| 4061 | admin privileges. |
| 4062 | |
Baptiste Assmann | fabcbe0 | 2014-04-24 22:16:59 +0200 | [diff] [blame] | 4063 | - "add-acl" is used to add a new entry into an ACL. The ACL must be loaded |
| 4064 | from a file (even a dummy empty file). The file name of the ACL to be |
| 4065 | updated is passed between parentheses. It takes one argument: <key fmt>, |
| 4066 | which follows log-format rules, to collect content of the new entry. It |
| 4067 | performs a lookup in the ACL before insertion, to avoid duplicated (or |
| 4068 | more) values. This lookup is done by a linear search and can be expensive |
| 4069 | with large lists! It is the equivalent of the "add acl" command from the |
| 4070 | stats socket, but can be triggered by an HTTP response. |
| 4071 | |
| 4072 | - "del-acl" is used to delete an entry from an ACL. The ACL must be loaded |
| 4073 | from a file (even a dummy empty file). The file name of the ACL to be |
| 4074 | updated is passed between parentheses. It takes one argument: <key fmt>, |
| 4075 | which follows log-format rules, to collect content of the entry to delete. |
| 4076 | It is the equivalent of the "del acl" command from the stats socket, but |
| 4077 | can be triggered by an HTTP response. |
| 4078 | |
| 4079 | - "del-map" is used to delete an entry from a MAP. The MAP must be loaded |
| 4080 | from a file (even a dummy empty file). The file name of the MAP to be |
| 4081 | updated is passed between parentheses. It takes one argument: <key fmt>, |
| 4082 | which follows log-format rules, to collect content of the entry to delete. |
| 4083 | It takes one argument: "file name" It is the equivalent of the "del map" |
| 4084 | command from the stats socket, but can be triggered by an HTTP response. |
| 4085 | |
| 4086 | - "set-map" is used to add a new entry into a MAP. The MAP must be loaded |
| 4087 | from a file (even a dummy empty file). The file name of the MAP to be |
| 4088 | updated is passed between parentheses. It takes 2 arguments: <key fmt>, |
| 4089 | which follows log-format rules, used to collect MAP key, and <value fmt>, |
| 4090 | which follows log-format rules, used to collect content for the new entry. |
| 4091 | It performs a lookup in the MAP before insertion, to avoid duplicated (or |
| 4092 | more) values. This lookup is done by a linear search and can be expensive |
| 4093 | with large lists! It is the equivalent of the "set map" command from the |
| 4094 | stats socket, but can be triggered by an HTTP response. |
| 4095 | |
Thierry FOURNIER | e80fada | 2015-05-26 18:06:31 +0200 | [diff] [blame] | 4096 | - capture <sample> id <id> : |
| 4097 | captures sample expression <sample> from the response buffer, and converts |
| 4098 | it to a string. The resulting string is stored into the next request |
| 4099 | "capture" slot, so it will possibly appear next to some captured HTTP |
| 4100 | headers. It will then automatically appear in the logs, and it will be |
| 4101 | possible to extract it using sample fetch rules to feed it into headers or |
| 4102 | anything. Please check section 7.3 (Fetching samples) and "capture |
| 4103 | response header" for more information. |
| 4104 | |
| 4105 | The keyword "id" is the id of the capture slot which is used for storing |
| 4106 | the string. The capture slot must be defined in an associated frontend. |
| 4107 | This is useful to run captures in backends. The slot id can be declared by |
| 4108 | a previous directive "http-response capture" or with the "declare capture" |
| 4109 | keyword. |
Baptiste Assmann | e954493 | 2015-11-03 23:31:35 +0100 | [diff] [blame] | 4110 | If the slot <id> doesn't exist, then HAProxy fails parsing the |
| 4111 | configuration to prevent unexpected behavior at run time. |
Thierry FOURNIER | e80fada | 2015-05-26 18:06:31 +0200 | [diff] [blame] | 4112 | |
Willy Tarreau | 51d861a | 2015-05-22 17:30:48 +0200 | [diff] [blame] | 4113 | - "redirect" : this performs an HTTP redirection based on a redirect rule. |
| 4114 | This supports a format string similarly to "http-request redirect" rules, |
| 4115 | with the exception that only the "location" type of redirect is possible |
| 4116 | on the response. See the "redirect" keyword for the rule's syntax. When |
| 4117 | a redirect rule is applied during a response, connections to the server |
| 4118 | are closed so that no data can be forwarded from the server to the client. |
| 4119 | |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 4120 | - set-var(<var-name>) expr: |
| 4121 | Is used to set the contents of a variable. The variable is declared |
| 4122 | inline. |
| 4123 | |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 4124 | <var-name> The name of the variable starts with an indication about |
| 4125 | its scope. The scopes allowed are: |
| 4126 | "sess" : the variable is shared with the whole session |
| 4127 | "txn" : the variable is shared with the transaction |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 4128 | (request and response) |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 4129 | "req" : the variable is shared only during request |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 4130 | processing |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 4131 | "res" : the variable is shared only during response |
| 4132 | processing |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 4133 | This prefix is followed by a name. The separator is a '.'. |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 4134 | The name may only contain characters 'a-z', 'A-Z', '0-9' |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 4135 | and '_'. |
| 4136 | |
| 4137 | <expr> Is a standard HAProxy expression formed by a sample-fetch |
| 4138 | followed by some converters. |
| 4139 | |
| 4140 | Example: |
| 4141 | |
| 4142 | http-response set-var(sess.last_redir) res.hdr(location) |
| 4143 | |
Thierry FOURNIER | 236657b | 2015-08-19 08:25:14 +0200 | [diff] [blame] | 4144 | - sc-set-gpt0(<sc-id>) <int> : |
| 4145 | This action sets the GPT0 tag according to the sticky counter designated |
| 4146 | by <sc-id> and the value of <int>. The expected result is a boolean. If |
| 4147 | an error occurs, this action silently fails and the actions evaluation |
| 4148 | continues. |
| 4149 | |
Thierry FOURNIER | e0627bd | 2015-08-04 08:20:33 +0200 | [diff] [blame] | 4150 | - sc-inc-gpc0(<sc-id>): |
| 4151 | This action increments the GPC0 counter according with the sticky counter |
| 4152 | designated by <sc-id>. If an error occurs, this action silently fails and |
| 4153 | the actions evaluation continues. |
| 4154 | |
Willy Tarreau | 2d392c2 | 2015-08-24 01:43:45 +0200 | [diff] [blame] | 4155 | - "silent-drop" : this stops the evaluation of the rules and makes the |
| 4156 | client-facing connection suddenly disappear using a system-dependant way |
| 4157 | that tries to prevent the client from being notified. The effect it then |
| 4158 | that the client still sees an established connection while there's none |
| 4159 | on HAProxy. The purpose is to achieve a comparable effect to "tarpit" |
| 4160 | except that it doesn't use any local resource at all on the machine |
| 4161 | running HAProxy. It can resist much higher loads than "tarpit", and slow |
| 4162 | down stronger attackers. It is important to undestand the impact of using |
| 4163 | this mechanism. All stateful equipments placed between the client and |
| 4164 | HAProxy (firewalls, proxies, load balancers) will also keep the |
| 4165 | established connection for a long time and may suffer from this action. |
| 4166 | On modern Linux systems running with enough privileges, the TCP_REPAIR |
| 4167 | socket option is used to block the emission of a TCP reset. On other |
| 4168 | systems, the socket's TTL is reduced to 1 so that the TCP reset doesn't |
| 4169 | pass the first router, though it's still delivered to local networks. Do |
| 4170 | not use it unless you fully understand how it works. |
| 4171 | |
Willy Tarreau | e365c0b | 2013-06-11 16:06:12 +0200 | [diff] [blame] | 4172 | There is no limit to the number of http-response statements per instance. |
| 4173 | |
Godbach | 0925026 | 2013-07-02 01:19:15 +0800 | [diff] [blame] | 4174 | It is important to know that http-response rules are processed very early in |
Ruoshan Huang | eb5a363 | 2015-12-08 21:00:23 +0800 | [diff] [blame] | 4175 | the HTTP processing, before "rspdel" or "rsprep" or "rspadd" rules. That way, |
| 4176 | headers added by "add-header"/"set-header" are visible by almost all further ACL |
Willy Tarreau | e365c0b | 2013-06-11 16:06:12 +0200 | [diff] [blame] | 4177 | rules. |
| 4178 | |
Ruoshan Huang | eb5a363 | 2015-12-08 21:00:23 +0800 | [diff] [blame] | 4179 | Using "rspadd"/"rspdel"/"rsprep" to manipulate request headers is discouraged |
| 4180 | in newer versions (>= 1.5). But if you need to use regular expression to |
| 4181 | delete headers, you can still use "rspdel". Also please use |
| 4182 | "http-response deny" instead of "rspdeny". |
| 4183 | |
Baptiste Assmann | fabcbe0 | 2014-04-24 22:16:59 +0200 | [diff] [blame] | 4184 | Example: |
| 4185 | acl key_acl res.hdr(X-Acl-Key) -m found |
| 4186 | |
| 4187 | acl myhost hdr(Host) -f myhost.lst |
| 4188 | |
| 4189 | http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl |
| 4190 | http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl |
| 4191 | |
| 4192 | Example: |
| 4193 | acl value res.hdr(X-Value) -m found |
| 4194 | |
| 4195 | use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found } |
| 4196 | |
| 4197 | http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value |
| 4198 | http-response del-map(map.lst) %[src] if ! value |
| 4199 | |
Willy Tarreau | e365c0b | 2013-06-11 16:06:12 +0200 | [diff] [blame] | 4200 | See also : "http-request", section 3.4 about userlists and section 7 about |
| 4201 | ACL usage. |
| 4202 | |
Baptiste Assmann | 5ecb77f | 2013-10-06 23:24:13 +0200 | [diff] [blame] | 4203 | |
Willy Tarreau | 3063195 | 2015-08-06 15:05:24 +0200 | [diff] [blame] | 4204 | http-reuse { never | safe | aggressive | always } |
| 4205 | Declare how idle HTTP connections may be shared between requests |
| 4206 | |
| 4207 | May be used in sections: defaults | frontend | listen | backend |
| 4208 | yes | no | yes | yes |
| 4209 | |
| 4210 | By default, a connection established between haproxy and the backend server |
| 4211 | belongs to the session that initiated it. The downside is that between the |
| 4212 | response and the next request, the connection remains idle and is not used. |
| 4213 | In many cases for performance reasons it is desirable to make it possible to |
| 4214 | reuse these idle connections to serve other requests from different sessions. |
| 4215 | This directive allows to tune this behaviour. |
| 4216 | |
| 4217 | The argument indicates the desired connection reuse strategy : |
| 4218 | |
| 4219 | - "never" : idle connections are never shared between sessions. This is |
| 4220 | the default choice. It may be enforced to cancel a different |
| 4221 | strategy inherited from a defaults section or for |
| 4222 | troubleshooting. For example, if an old bogus application |
| 4223 | considers that multiple requests over the same connection come |
| 4224 | from the same client and it is not possible to fix the |
| 4225 | application, it may be desirable to disable connection sharing |
| 4226 | in a single backend. An example of such an application could |
| 4227 | be an old haproxy using cookie insertion in tunnel mode and |
| 4228 | not checking any request past the first one. |
| 4229 | |
| 4230 | - "safe" : this is the recommended strategy. The first request of a |
| 4231 | session is always sent over its own connection, and only |
| 4232 | subsequent requests may be dispatched over other existing |
| 4233 | connections. This ensures that in case the server closes the |
| 4234 | connection when the request is being sent, the browser can |
| 4235 | decide to silently retry it. Since it is exactly equivalent to |
| 4236 | regular keep-alive, there should be no side effects. |
| 4237 | |
| 4238 | - "aggressive" : this mode may be useful in webservices environments where |
| 4239 | all servers are not necessarily known and where it would be |
| 4240 | appreciable to deliver most first requests over existing |
| 4241 | connections. In this case, first requests are only delivered |
| 4242 | over existing connections that have been reused at least once, |
| 4243 | proving that the server correctly supports connection reuse. |
| 4244 | It should only be used when it's sure that the client can |
| 4245 | retry a failed request once in a while and where the benefit |
| 4246 | of aggressive connection reuse significantly outweights the |
| 4247 | downsides of rare connection failures. |
| 4248 | |
| 4249 | - "always" : this mode is only recommended when the path to the server is |
| 4250 | known for never breaking existing connections quickly after |
| 4251 | releasing them. It allows the first request of a session to be |
| 4252 | sent to an existing connection. This can provide a significant |
| 4253 | performance increase over the "safe" strategy when the backend |
| 4254 | is a cache farm, since such components tend to show a |
| 4255 | consistent behaviour and will benefit from the connection |
| 4256 | sharing. It is recommended that the "http-keep-alive" timeout |
| 4257 | remains low in this mode so that no dead connections remain |
| 4258 | usable. In most cases, this will lead to the same performance |
| 4259 | gains as "aggressive" but with more risks. It should only be |
| 4260 | used when it improves the situation over "aggressive". |
| 4261 | |
| 4262 | When http connection sharing is enabled, a great care is taken to respect the |
| 4263 | connection properties and compatiblities. Specifically : |
| 4264 | - connections made with "usesrc" followed by a client-dependant value |
| 4265 | ("client", "clientip", "hdr_ip") are marked private and never shared ; |
| 4266 | |
| 4267 | - connections sent to a server with a TLS SNI extension are marked private |
| 4268 | and are never shared ; |
| 4269 | |
| 4270 | - connections receiving a status code 401 or 407 expect some authentication |
| 4271 | to be sent in return. Due to certain bogus authentication schemes (such |
| 4272 | as NTLM) relying on the connection, these connections are marked private |
| 4273 | and are never shared ; |
| 4274 | |
| 4275 | No connection pool is involved, once a session dies, the last idle connection |
| 4276 | it was attached to is deleted at the same time. This ensures that connections |
| 4277 | may not last after all sessions are closed. |
| 4278 | |
| 4279 | Note: connection reuse improves the accuracy of the "server maxconn" setting, |
| 4280 | because almost no new connection will be established while idle connections |
| 4281 | remain available. This is particularly true with the "always" strategy. |
| 4282 | |
| 4283 | See also : "option http-keep-alive", "server maxconn" |
| 4284 | |
| 4285 | |
Mark Lamourine | c2247f0 | 2012-01-04 13:02:01 -0500 | [diff] [blame] | 4286 | http-send-name-header [<header>] |
| 4287 | Add the server name to a request. Use the header string given by <header> |
| 4288 | |
| 4289 | May be used in sections: defaults | frontend | listen | backend |
| 4290 | yes | no | yes | yes |
| 4291 | |
| 4292 | Arguments : |
| 4293 | |
| 4294 | <header> The header string to use to send the server name |
| 4295 | |
| 4296 | The "http-send-name-header" statement causes the name of the target |
| 4297 | server to be added to the headers of an HTTP request. The name |
| 4298 | is added with the header string proved. |
| 4299 | |
| 4300 | See also : "server" |
| 4301 | |
Krzysztof Piotr Oledzki | f58a962 | 2008-02-23 01:19:10 +0100 | [diff] [blame] | 4302 | id <value> |
Willy Tarreau | 53fb4ae | 2009-10-04 23:04:08 +0200 | [diff] [blame] | 4303 | Set a persistent ID to a proxy. |
| 4304 | May be used in sections : defaults | frontend | listen | backend |
| 4305 | no | yes | yes | yes |
| 4306 | Arguments : none |
| 4307 | |
| 4308 | Set a persistent ID for the proxy. This ID must be unique and positive. |
| 4309 | An unused ID will automatically be assigned if unset. The first assigned |
| 4310 | value will be 1. This ID is currently only returned in statistics. |
Krzysztof Piotr Oledzki | f58a962 | 2008-02-23 01:19:10 +0100 | [diff] [blame] | 4311 | |
| 4312 | |
Cyril Bonté | 0d4bf01 | 2010-04-25 23:21:46 +0200 | [diff] [blame] | 4313 | ignore-persist { if | unless } <condition> |
| 4314 | Declare a condition to ignore persistence |
| 4315 | May be used in sections: defaults | frontend | listen | backend |
| 4316 | no | yes | yes | yes |
| 4317 | |
| 4318 | By default, when cookie persistence is enabled, every requests containing |
| 4319 | the cookie are unconditionally persistent (assuming the target server is up |
| 4320 | and running). |
| 4321 | |
| 4322 | The "ignore-persist" statement allows one to declare various ACL-based |
| 4323 | conditions which, when met, will cause a request to ignore persistence. |
| 4324 | This is sometimes useful to load balance requests for static files, which |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 4325 | often don't require persistence. This can also be used to fully disable |
Cyril Bonté | 0d4bf01 | 2010-04-25 23:21:46 +0200 | [diff] [blame] | 4326 | persistence for a specific User-Agent (for example, some web crawler bots). |
| 4327 | |
Cyril Bonté | 0d4bf01 | 2010-04-25 23:21:46 +0200 | [diff] [blame] | 4328 | The persistence is ignored when an "if" condition is met, or unless an |
| 4329 | "unless" condition is met. |
| 4330 | |
| 4331 | See also : "force-persist", "cookie", and section 7 about ACL usage. |
| 4332 | |
Baptiste Assmann | 01c6cc3 | 2015-08-23 11:45:29 +0200 | [diff] [blame] | 4333 | load-server-state-from-file { global | local | none } |
| 4334 | Allow seamless reload of HAProxy |
| 4335 | May be used in sections: defaults | frontend | listen | backend |
| 4336 | yes | no | yes | yes |
| 4337 | |
| 4338 | This directive points HAProxy to a file where server state from previous |
| 4339 | running process has been saved. That way, when starting up, before handling |
| 4340 | traffic, the new process can apply old states to servers exactly has if no |
| 4341 | reload occured. The purpose of the "load-server-state-from-file" directive is |
| 4342 | to tell haproxy which file to use. For now, only 2 arguments to either prevent |
| 4343 | loading state or load states from a file containing all backends and servers. |
| 4344 | The state file can be generated by running the command "show servers state" |
| 4345 | over the stats socket and redirect output. |
| 4346 | |
| 4347 | The format of the file is versionned and is very specific. To understand it, |
| 4348 | please read the documentation of the "show servers state" command (chapter |
Kevin Decherf | 949c720 | 2015-10-13 23:26:44 +0200 | [diff] [blame] | 4349 | 9.2 of Management Guide). |
Baptiste Assmann | 01c6cc3 | 2015-08-23 11:45:29 +0200 | [diff] [blame] | 4350 | |
| 4351 | Arguments: |
| 4352 | global load the content of the file pointed by the global directive |
| 4353 | named "server-state-file". |
| 4354 | |
| 4355 | local load the content of the file pointed by the directive |
| 4356 | "server-state-file-name" if set. If not set, then the backend |
| 4357 | name is used as a file name. |
| 4358 | |
| 4359 | none don't load any stat for this backend |
| 4360 | |
| 4361 | Notes: |
| 4362 | - server's IP address is not updated unless DNS resolution is enabled on |
| 4363 | the server. It means that if a server IP address has been changed using |
| 4364 | the stat socket, this information won't be re-applied after reloading. |
| 4365 | |
| 4366 | - server's weight is applied from previous running process unless it has |
| 4367 | has changed between previous and new configuration files. |
| 4368 | |
| 4369 | Example 1: |
| 4370 | |
| 4371 | Minimal configuration: |
| 4372 | |
| 4373 | global |
| 4374 | stats socket /tmp/socket |
| 4375 | server-state-file /tmp/server_state |
| 4376 | |
| 4377 | defaults |
| 4378 | load-server-state-from-file global |
| 4379 | |
| 4380 | backend bk |
| 4381 | server s1 127.0.0.1:22 check weight 11 |
| 4382 | server s2 127.0.0.1:22 check weight 12 |
| 4383 | |
| 4384 | Then one can run : |
| 4385 | |
| 4386 | socat /tmp/socket - <<< "show servers state" > /tmp/server_state |
| 4387 | |
| 4388 | Content of the file /tmp/server_state would be like this: |
| 4389 | |
| 4390 | 1 |
| 4391 | # <field names skipped for the doc example> |
| 4392 | 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0 |
| 4393 | 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0 |
| 4394 | |
| 4395 | Example 2: |
| 4396 | |
| 4397 | Minimal configuration: |
| 4398 | |
| 4399 | global |
| 4400 | stats socket /tmp/socket |
| 4401 | server-state-base /etc/haproxy/states |
| 4402 | |
| 4403 | defaults |
| 4404 | load-server-state-from-file local |
| 4405 | |
| 4406 | backend bk |
| 4407 | server s1 127.0.0.1:22 check weight 11 |
| 4408 | server s2 127.0.0.1:22 check weight 12 |
| 4409 | |
| 4410 | Then one can run : |
| 4411 | |
| 4412 | socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk |
| 4413 | |
| 4414 | Content of the file /etc/haproxy/states/bk would be like this: |
| 4415 | |
| 4416 | 1 |
| 4417 | # <field names skipped for the doc example> |
| 4418 | 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0 |
| 4419 | 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0 |
| 4420 | |
| 4421 | See also: "server-state-file", "server-state-file-name", and |
| 4422 | "show servers state" |
| 4423 | |
Cyril Bonté | 0d4bf01 | 2010-04-25 23:21:46 +0200 | [diff] [blame] | 4424 | |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4425 | log global |
Willy Tarreau | 18324f5 | 2014-06-27 18:10:07 +0200 | [diff] [blame] | 4426 | log <address> [len <length>] <facility> [<level> [<minlevel>]] |
William Lallemand | 0f99e34 | 2011-10-12 17:50:54 +0200 | [diff] [blame] | 4427 | no log |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4428 | Enable per-instance logging of events and traffic. |
| 4429 | May be used in sections : defaults | frontend | listen | backend |
| 4430 | yes | yes | yes | yes |
William Lallemand | 0f99e34 | 2011-10-12 17:50:54 +0200 | [diff] [blame] | 4431 | |
| 4432 | Prefix : |
| 4433 | no should be used when the logger list must be flushed. For example, |
| 4434 | if you don't want to inherit from the default logger list. This |
| 4435 | prefix does not allow arguments. |
| 4436 | |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4437 | Arguments : |
| 4438 | global should be used when the instance's logging parameters are the |
| 4439 | same as the global ones. This is the most common usage. "global" |
| 4440 | replaces <address>, <facility> and <level> with those of the log |
| 4441 | entries found in the "global" section. Only one "log global" |
| 4442 | statement may be used per instance, and this form takes no other |
| 4443 | parameter. |
| 4444 | |
| 4445 | <address> indicates where to send the logs. It takes the same format as |
| 4446 | for the "global" section's logs, and can be one of : |
| 4447 | |
| 4448 | - An IPv4 address optionally followed by a colon (':') and a UDP |
| 4449 | port. If no port is specified, 514 is used by default (the |
| 4450 | standard syslog port). |
| 4451 | |
David du Colombier | 24bb5f5 | 2011-03-17 10:40:23 +0100 | [diff] [blame] | 4452 | - An IPv6 address followed by a colon (':') and optionally a UDP |
| 4453 | port. If no port is specified, 514 is used by default (the |
| 4454 | standard syslog port). |
| 4455 | |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4456 | - A filesystem path to a UNIX domain socket, keeping in mind |
| 4457 | considerations for chroot (be sure the path is accessible |
| 4458 | inside the chroot) and uid/gid (be sure the path is |
| 4459 | appropriately writeable). |
| 4460 | |
William Lallemand | b2f0745 | 2015-05-12 14:27:13 +0200 | [diff] [blame] | 4461 | You may want to reference some environment variables in the |
| 4462 | address parameter, see section 2.3 about environment variables. |
Willy Tarreau | dad36a3 | 2013-03-11 01:20:04 +0100 | [diff] [blame] | 4463 | |
Willy Tarreau | 18324f5 | 2014-06-27 18:10:07 +0200 | [diff] [blame] | 4464 | <length> is an optional maximum line length. Log lines larger than this |
| 4465 | value will be truncated before being sent. The reason is that |
| 4466 | syslog servers act differently on log line length. All servers |
| 4467 | support the default value of 1024, but some servers simply drop |
| 4468 | larger lines while others do log them. If a server supports long |
| 4469 | lines, it may make sense to set this value here in order to avoid |
| 4470 | truncating long lines. Similarly, if a server drops long lines, |
| 4471 | it is preferable to truncate them before sending them. Accepted |
| 4472 | values are 80 to 65535 inclusive. The default value of 1024 is |
| 4473 | generally fine for all standard usages. Some specific cases of |
| 4474 | long captures or JSON-formated logs may require larger values. |
| 4475 | |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4476 | <facility> must be one of the 24 standard syslog facilities : |
| 4477 | |
| 4478 | kern user mail daemon auth syslog lpr news |
| 4479 | uucp cron auth2 ftp ntp audit alert cron2 |
| 4480 | local0 local1 local2 local3 local4 local5 local6 local7 |
| 4481 | |
| 4482 | <level> is optional and can be specified to filter outgoing messages. By |
| 4483 | default, all messages are sent. If a level is specified, only |
| 4484 | messages with a severity at least as important as this level |
Willy Tarreau | f7edefa | 2009-05-10 17:20:05 +0200 | [diff] [blame] | 4485 | will be sent. An optional minimum level can be specified. If it |
| 4486 | is set, logs emitted with a more severe level than this one will |
| 4487 | be capped to this level. This is used to avoid sending "emerg" |
| 4488 | messages on all terminals on some default syslog configurations. |
| 4489 | Eight levels are known : |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4490 | |
| 4491 | emerg alert crit err warning notice info debug |
| 4492 | |
William Lallemand | 0f99e34 | 2011-10-12 17:50:54 +0200 | [diff] [blame] | 4493 | It is important to keep in mind that it is the frontend which decides what to |
| 4494 | log from a connection, and that in case of content switching, the log entries |
| 4495 | from the backend will be ignored. Connections are logged at level "info". |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 4496 | |
| 4497 | However, backend log declaration define how and where servers status changes |
| 4498 | will be logged. Level "notice" will be used to indicate a server going up, |
| 4499 | "warning" will be used for termination signals and definitive service |
| 4500 | termination, and "alert" will be used for when a server goes down. |
| 4501 | |
| 4502 | Note : According to RFC3164, messages are truncated to 1024 bytes before |
| 4503 | being emitted. |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4504 | |
| 4505 | Example : |
| 4506 | log global |
Willy Tarreau | f7edefa | 2009-05-10 17:20:05 +0200 | [diff] [blame] | 4507 | log 127.0.0.1:514 local0 notice # only send important events |
| 4508 | log 127.0.0.1:514 local0 notice notice # same but limit output level |
William Lallemand | b2f0745 | 2015-05-12 14:27:13 +0200 | [diff] [blame] | 4509 | log "${LOCAL_SYSLOG}:514" local0 notice # send to local server |
Willy Tarreau | dad36a3 | 2013-03-11 01:20:04 +0100 | [diff] [blame] | 4510 | |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4511 | |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 4512 | log-format <string> |
Willy Tarreau | fb4e7ea | 2015-01-07 14:55:17 +0100 | [diff] [blame] | 4513 | Specifies the log format string to use for traffic logs |
| 4514 | May be used in sections: defaults | frontend | listen | backend |
| 4515 | yes | yes | yes | no |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 4516 | |
Willy Tarreau | fb4e7ea | 2015-01-07 14:55:17 +0100 | [diff] [blame] | 4517 | This directive specifies the log format string that will be used for all logs |
| 4518 | resulting from traffic passing through the frontend using this line. If the |
| 4519 | directive is used in a defaults section, all subsequent frontends will use |
| 4520 | the same log format. Please see section 8.2.4 which covers the log format |
| 4521 | string in depth. |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 4522 | |
Dragan Dosen | 7ad3154 | 2015-09-28 17:16:47 +0200 | [diff] [blame] | 4523 | log-format-sd <string> |
| 4524 | Specifies the RFC5424 structured-data log format string |
| 4525 | May be used in sections: defaults | frontend | listen | backend |
| 4526 | yes | yes | yes | no |
| 4527 | |
| 4528 | This directive specifies the RFC5424 structured-data log format string that |
| 4529 | will be used for all logs resulting from traffic passing through the frontend |
| 4530 | using this line. If the directive is used in a defaults section, all |
| 4531 | subsequent frontends will use the same log format. Please see section 8.2.4 |
| 4532 | which covers the log format string in depth. |
| 4533 | |
| 4534 | See https://tools.ietf.org/html/rfc5424#section-6.3 for more information |
| 4535 | about the RFC5424 structured-data part. |
| 4536 | |
| 4537 | Note : This log format string will be used only for loggers that have set |
| 4538 | log format to "rfc5424". |
| 4539 | |
| 4540 | Example : |
| 4541 | log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"] |
| 4542 | |
| 4543 | |
Willy Tarreau | 094af4e | 2015-01-07 15:03:42 +0100 | [diff] [blame] | 4544 | log-tag <string> |
| 4545 | Specifies the log tag to use for all outgoing logs |
| 4546 | May be used in sections: defaults | frontend | listen | backend |
| 4547 | yes | yes | yes | yes |
| 4548 | |
| 4549 | Sets the tag field in the syslog header to this string. It defaults to the |
| 4550 | log-tag set in the global section, otherwise the program name as launched |
| 4551 | from the command line, which usually is "haproxy". Sometimes it can be useful |
| 4552 | to differentiate between multiple processes running on the same host, or to |
| 4553 | differentiate customer instances running in the same process. In the backend, |
| 4554 | logs about servers up/down will use this tag. As a hint, it can be convenient |
| 4555 | to set a log-tag related to a hosted customer in a defaults section then put |
| 4556 | all the frontends and backends for that customer, then start another customer |
| 4557 | in a new defaults section. See also the global "log-tag" directive. |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4558 | |
Willy Tarreau | c35362a | 2014-04-25 13:58:37 +0200 | [diff] [blame] | 4559 | max-keep-alive-queue <value> |
| 4560 | Set the maximum server queue size for maintaining keep-alive connections |
| 4561 | May be used in sections: defaults | frontend | listen | backend |
| 4562 | yes | no | yes | yes |
| 4563 | |
| 4564 | HTTP keep-alive tries to reuse the same server connection whenever possible, |
| 4565 | but sometimes it can be counter-productive, for example if a server has a lot |
| 4566 | of connections while other ones are idle. This is especially true for static |
| 4567 | servers. |
| 4568 | |
| 4569 | The purpose of this setting is to set a threshold on the number of queued |
| 4570 | connections at which haproxy stops trying to reuse the same server and prefers |
| 4571 | to find another one. The default value, -1, means there is no limit. A value |
| 4572 | of zero means that keep-alive requests will never be queued. For very close |
| 4573 | servers which can be reached with a low latency and which are not sensible to |
| 4574 | breaking keep-alive, a low value is recommended (eg: local static server can |
| 4575 | use a value of 10 or less). For remote servers suffering from a high latency, |
| 4576 | higher values might be needed to cover for the latency and/or the cost of |
| 4577 | picking a different server. |
| 4578 | |
| 4579 | Note that this has no impact on responses which are maintained to the same |
| 4580 | server consecutively to a 401 response. They will still go to the same server |
| 4581 | even if they have to be queued. |
| 4582 | |
| 4583 | See also : "option http-server-close", "option prefer-last-server", server |
| 4584 | "maxconn" and cookie persistence. |
| 4585 | |
| 4586 | |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4587 | maxconn <conns> |
| 4588 | Fix the maximum number of concurrent connections on a frontend |
| 4589 | May be used in sections : defaults | frontend | listen | backend |
| 4590 | yes | yes | yes | no |
| 4591 | Arguments : |
| 4592 | <conns> is the maximum number of concurrent connections the frontend will |
| 4593 | accept to serve. Excess connections will be queued by the system |
| 4594 | in the socket's listen queue and will be served once a connection |
| 4595 | closes. |
| 4596 | |
| 4597 | If the system supports it, it can be useful on big sites to raise this limit |
| 4598 | very high so that haproxy manages connection queues, instead of leaving the |
| 4599 | clients with unanswered connection attempts. This value should not exceed the |
| 4600 | global maxconn. Also, keep in mind that a connection contains two buffers |
Baptiste Assmann | 79fb45d | 2016-03-06 23:34:31 +0100 | [diff] [blame] | 4601 | of tune.bufsize (16kB by default) each, as well as some other data resulting |
| 4602 | in about 33 kB of RAM being consumed per established connection. That means |
| 4603 | that a medium system equipped with 1GB of RAM can withstand around |
| 4604 | 20000-25000 concurrent connections if properly tuned. |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4605 | |
| 4606 | Also, when <conns> is set to large values, it is possible that the servers |
| 4607 | are not sized to accept such loads, and for this reason it is generally wise |
| 4608 | to assign them some reasonable connection limits. |
| 4609 | |
Vincent Bernat | 6341be5 | 2012-06-27 17:18:30 +0200 | [diff] [blame] | 4610 | By default, this value is set to 2000. |
| 4611 | |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4612 | See also : "server", global section's "maxconn", "fullconn" |
| 4613 | |
| 4614 | |
| 4615 | mode { tcp|http|health } |
| 4616 | Set the running mode or protocol of the instance |
| 4617 | May be used in sections : defaults | frontend | listen | backend |
| 4618 | yes | yes | yes | yes |
| 4619 | Arguments : |
| 4620 | tcp The instance will work in pure TCP mode. A full-duplex connection |
| 4621 | will be established between clients and servers, and no layer 7 |
| 4622 | examination will be performed. This is the default mode. It |
| 4623 | should be used for SSL, SSH, SMTP, ... |
| 4624 | |
| 4625 | http The instance will work in HTTP mode. The client request will be |
| 4626 | analyzed in depth before connecting to any server. Any request |
| 4627 | which is not RFC-compliant will be rejected. Layer 7 filtering, |
| 4628 | processing and switching will be possible. This is the mode which |
| 4629 | brings HAProxy most of its value. |
| 4630 | |
| 4631 | health The instance will work in "health" mode. It will just reply "OK" |
Willy Tarreau | 82569f9 | 2012-09-27 23:48:56 +0200 | [diff] [blame] | 4632 | to incoming connections and close the connection. Alternatively, |
| 4633 | If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent |
| 4634 | instead. Nothing will be logged in either case. This mode is used |
| 4635 | to reply to external components health checks. This mode is |
| 4636 | deprecated and should not be used anymore as it is possible to do |
| 4637 | the same and even better by combining TCP or HTTP modes with the |
| 4638 | "monitor" keyword. |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4639 | |
Cyril Bonté | 108cf6e | 2012-04-21 23:30:29 +0200 | [diff] [blame] | 4640 | When doing content switching, it is mandatory that the frontend and the |
| 4641 | backend are in the same mode (generally HTTP), otherwise the configuration |
| 4642 | will be refused. |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4643 | |
Cyril Bonté | 108cf6e | 2012-04-21 23:30:29 +0200 | [diff] [blame] | 4644 | Example : |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4645 | defaults http_instances |
| 4646 | mode http |
| 4647 | |
Cyril Bonté | 108cf6e | 2012-04-21 23:30:29 +0200 | [diff] [blame] | 4648 | See also : "monitor", "monitor-net" |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4649 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 4650 | |
Cyril Bonté | f0c6061 | 2010-02-06 14:44:47 +0100 | [diff] [blame] | 4651 | monitor fail { if | unless } <condition> |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4652 | Add a condition to report a failure to a monitor HTTP request. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 4653 | May be used in sections : defaults | frontend | listen | backend |
| 4654 | no | yes | yes | no |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 4655 | Arguments : |
| 4656 | if <cond> the monitor request will fail if the condition is satisfied, |
| 4657 | and will succeed otherwise. The condition should describe a |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 4658 | combined test which must induce a failure if all conditions |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 4659 | are met, for instance a low number of servers both in a |
| 4660 | backend and its backup. |
| 4661 | |
| 4662 | unless <cond> the monitor request will succeed only if the condition is |
| 4663 | satisfied, and will fail otherwise. Such a condition may be |
| 4664 | based on a test on the presence of a minimum number of active |
| 4665 | servers in a list of backends. |
| 4666 | |
| 4667 | This statement adds a condition which can force the response to a monitor |
| 4668 | request to report a failure. By default, when an external component queries |
| 4669 | the URI dedicated to monitoring, a 200 response is returned. When one of the |
| 4670 | conditions above is met, haproxy will return 503 instead of 200. This is |
| 4671 | very useful to report a site failure to an external component which may base |
| 4672 | routing advertisements between multiple sites on the availability reported by |
| 4673 | haproxy. In this case, one would rely on an ACL involving the "nbsrv" |
Willy Tarreau | ae94d4d | 2011-05-11 16:28:49 +0200 | [diff] [blame] | 4674 | criterion. Note that "monitor fail" only works in HTTP mode. Both status |
| 4675 | messages may be tweaked using "errorfile" or "errorloc" if needed. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 4676 | |
| 4677 | Example: |
| 4678 | frontend www |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4679 | mode http |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 4680 | acl site_dead nbsrv(dynamic) lt 2 |
| 4681 | acl site_dead nbsrv(static) lt 2 |
| 4682 | monitor-uri /site_alive |
| 4683 | monitor fail if site_dead |
| 4684 | |
Willy Tarreau | ae94d4d | 2011-05-11 16:28:49 +0200 | [diff] [blame] | 4685 | See also : "monitor-net", "monitor-uri", "errorfile", "errorloc" |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4686 | |
| 4687 | |
| 4688 | monitor-net <source> |
| 4689 | Declare a source network which is limited to monitor requests |
| 4690 | May be used in sections : defaults | frontend | listen | backend |
| 4691 | yes | yes | yes | no |
| 4692 | Arguments : |
| 4693 | <source> is the source IPv4 address or network which will only be able to |
| 4694 | get monitor responses to any request. It can be either an IPv4 |
| 4695 | address, a host name, or an address followed by a slash ('/') |
| 4696 | followed by a mask. |
| 4697 | |
| 4698 | In TCP mode, any connection coming from a source matching <source> will cause |
| 4699 | the connection to be immediately closed without any log. This allows another |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 4700 | equipment to probe the port and verify that it is still listening, without |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4701 | forwarding the connection to a remote server. |
| 4702 | |
| 4703 | In HTTP mode, a connection coming from a source matching <source> will be |
| 4704 | accepted, the following response will be sent without waiting for a request, |
| 4705 | then the connection will be closed : "HTTP/1.0 200 OK". This is normally |
| 4706 | enough for any front-end HTTP probe to detect that the service is UP and |
Willy Tarreau | 82569f9 | 2012-09-27 23:48:56 +0200 | [diff] [blame] | 4707 | running without forwarding the request to a backend server. Note that this |
| 4708 | response is sent in raw format, without any transformation. This is important |
| 4709 | as it means that it will not be SSL-encrypted on SSL listeners. |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4710 | |
Willy Tarreau | 82569f9 | 2012-09-27 23:48:56 +0200 | [diff] [blame] | 4711 | Monitor requests are processed very early, just after tcp-request connection |
| 4712 | ACLs which are the only ones able to block them. These connections are short |
| 4713 | lived and never wait for any data from the client. They cannot be logged, and |
| 4714 | it is the intended purpose. They are only used to report HAProxy's health to |
| 4715 | an upper component, nothing more. Please note that "monitor fail" rules do |
| 4716 | not apply to connections intercepted by "monitor-net". |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4717 | |
Willy Tarreau | 95cd283 | 2010-03-04 23:36:33 +0100 | [diff] [blame] | 4718 | Last, please note that only one "monitor-net" statement can be specified in |
| 4719 | a frontend. If more than one is found, only the last one will be considered. |
Cyril Bonté | 108cf6e | 2012-04-21 23:30:29 +0200 | [diff] [blame] | 4720 | |
Willy Tarreau | 2769aa0 | 2007-12-27 18:26:09 +0100 | [diff] [blame] | 4721 | Example : |
| 4722 | # addresses .252 and .253 are just probing us. |
| 4723 | frontend www |
| 4724 | monitor-net 192.168.0.252/31 |
| 4725 | |
| 4726 | See also : "monitor fail", "monitor-uri" |
| 4727 | |
| 4728 | |
| 4729 | monitor-uri <uri> |
| 4730 | Intercept a URI used by external components' monitor requests |
| 4731 | May be used in sections : defaults | frontend | listen | backend |
| 4732 | yes | yes | yes | no |
| 4733 | Arguments : |
| 4734 | <uri> is the exact URI which we want to intercept to return HAProxy's |
| 4735 | health status instead of forwarding the request. |
| 4736 | |
| 4737 | When an HTTP request referencing <uri> will be received on a frontend, |
| 4738 | HAProxy will not forward it nor log it, but instead will return either |
| 4739 | "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure |
| 4740 | conditions defined with "monitor fail". This is normally enough for any |
| 4741 | front-end HTTP probe to detect that the service is UP and running without |
| 4742 | forwarding the request to a backend server. Note that the HTTP method, the |
| 4743 | version and all headers are ignored, but the request must at least be valid |
| 4744 | at the HTTP level. This keyword may only be used with an HTTP-mode frontend. |
| 4745 | |
| 4746 | Monitor requests are processed very early. It is not possible to block nor |
| 4747 | divert them using ACLs. They cannot be logged either, and it is the intended |
| 4748 | purpose. They are only used to report HAProxy's health to an upper component, |
| 4749 | nothing more. However, it is possible to add any number of conditions using |
| 4750 | "monitor fail" and ACLs so that the result can be adjusted to whatever check |
| 4751 | can be imagined (most often the number of available servers in a backend). |
| 4752 | |
| 4753 | Example : |
| 4754 | # Use /haproxy_test to report haproxy's status |
| 4755 | frontend www |
| 4756 | mode http |
| 4757 | monitor-uri /haproxy_test |
| 4758 | |
| 4759 | See also : "monitor fail", "monitor-net" |
| 4760 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 4761 | |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 4762 | option abortonclose |
| 4763 | no option abortonclose |
| 4764 | Enable or disable early dropping of aborted requests pending in queues. |
| 4765 | May be used in sections : defaults | frontend | listen | backend |
| 4766 | yes | no | yes | yes |
| 4767 | Arguments : none |
| 4768 | |
| 4769 | In presence of very high loads, the servers will take some time to respond. |
| 4770 | The per-instance connection queue will inflate, and the response time will |
| 4771 | increase respective to the size of the queue times the average per-session |
| 4772 | response time. When clients will wait for more than a few seconds, they will |
Willy Tarreau | 198a744 | 2008-01-17 12:05:32 +0100 | [diff] [blame] | 4773 | often hit the "STOP" button on their browser, leaving a useless request in |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 4774 | the queue, and slowing down other users, and the servers as well, because the |
| 4775 | request will eventually be served, then aborted at the first error |
| 4776 | encountered while delivering the response. |
| 4777 | |
| 4778 | As there is no way to distinguish between a full STOP and a simple output |
| 4779 | close on the client side, HTTP agents should be conservative and consider |
| 4780 | that the client might only have closed its output channel while waiting for |
| 4781 | the response. However, this introduces risks of congestion when lots of users |
| 4782 | do the same, and is completely useless nowadays because probably no client at |
| 4783 | all will close the session while waiting for the response. Some HTTP agents |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 4784 | support this behaviour (Squid, Apache, HAProxy), and others do not (TUX, most |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 4785 | hardware-based load balancers). So the probability for a closed input channel |
Willy Tarreau | 198a744 | 2008-01-17 12:05:32 +0100 | [diff] [blame] | 4786 | to represent a user hitting the "STOP" button is close to 100%, and the risk |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 4787 | of being the single component to break rare but valid traffic is extremely |
| 4788 | low, which adds to the temptation to be able to abort a session early while |
| 4789 | still not served and not pollute the servers. |
| 4790 | |
| 4791 | In HAProxy, the user can choose the desired behaviour using the option |
| 4792 | "abortonclose". By default (without the option) the behaviour is HTTP |
| 4793 | compliant and aborted requests will be served. But when the option is |
| 4794 | specified, a session with an incoming channel closed will be aborted while |
| 4795 | it is still possible, either pending in the queue for a connection slot, or |
| 4796 | during the connection establishment if the server has not yet acknowledged |
| 4797 | the connection request. This considerably reduces the queue size and the load |
| 4798 | on saturated servers when users are tempted to click on STOP, which in turn |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 4799 | reduces the response time for other users. |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 4800 | |
| 4801 | If this option has been enabled in a "defaults" section, it can be disabled |
| 4802 | in a specific instance by prepending the "no" keyword before it. |
| 4803 | |
| 4804 | See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters |
| 4805 | |
| 4806 | |
Willy Tarreau | 4076a15 | 2009-04-02 15:18:36 +0200 | [diff] [blame] | 4807 | option accept-invalid-http-request |
| 4808 | no option accept-invalid-http-request |
| 4809 | Enable or disable relaxing of HTTP request parsing |
| 4810 | May be used in sections : defaults | frontend | listen | backend |
| 4811 | yes | yes | yes | no |
| 4812 | Arguments : none |
| 4813 | |
Willy Tarreau | 91852eb | 2015-05-01 13:26:00 +0200 | [diff] [blame] | 4814 | By default, HAProxy complies with RFC7230 in terms of message parsing. This |
Willy Tarreau | 4076a15 | 2009-04-02 15:18:36 +0200 | [diff] [blame] | 4815 | means that invalid characters in header names are not permitted and cause an |
| 4816 | error to be returned to the client. This is the desired behaviour as such |
| 4817 | forbidden characters are essentially used to build attacks exploiting server |
| 4818 | weaknesses, and bypass security filtering. Sometimes, a buggy browser or |
| 4819 | server will emit invalid header names for whatever reason (configuration, |
| 4820 | implementation) and the issue will not be immediately fixed. In such a case, |
| 4821 | it is possible to relax HAProxy's header name parser to accept any character |
Willy Tarreau | 422246e | 2012-01-07 23:54:13 +0100 | [diff] [blame] | 4822 | even if that does not make sense, by specifying this option. Similarly, the |
| 4823 | list of characters allowed to appear in a URI is well defined by RFC3986, and |
| 4824 | chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96 |
| 4825 | ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are |
| 4826 | not allowed at all. Haproxy always blocks a number of them (0..32, 127). The |
Willy Tarreau | 91852eb | 2015-05-01 13:26:00 +0200 | [diff] [blame] | 4827 | remaining ones are blocked by default unless this option is enabled. This |
Willy Tarreau | 1331766 | 2015-05-01 13:47:08 +0200 | [diff] [blame] | 4828 | option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests |
| 4829 | to pass through (no version specified) and multiple digits for both the major |
| 4830 | and the minor version. |
Willy Tarreau | 4076a15 | 2009-04-02 15:18:36 +0200 | [diff] [blame] | 4831 | |
| 4832 | This option should never be enabled by default as it hides application bugs |
| 4833 | and open security breaches. It should only be deployed after a problem has |
| 4834 | been confirmed. |
| 4835 | |
| 4836 | When this option is enabled, erroneous header names will still be accepted in |
| 4837 | requests, but the complete request will be captured in order to permit later |
Willy Tarreau | 422246e | 2012-01-07 23:54:13 +0100 | [diff] [blame] | 4838 | analysis using the "show errors" request on the UNIX stats socket. Similarly, |
| 4839 | requests containing invalid chars in the URI part will be logged. Doing this |
Willy Tarreau | 4076a15 | 2009-04-02 15:18:36 +0200 | [diff] [blame] | 4840 | also helps confirming that the issue has been solved. |
| 4841 | |
| 4842 | If this option has been enabled in a "defaults" section, it can be disabled |
| 4843 | in a specific instance by prepending the "no" keyword before it. |
| 4844 | |
| 4845 | See also : "option accept-invalid-http-response" and "show errors" on the |
| 4846 | stats socket. |
| 4847 | |
| 4848 | |
| 4849 | option accept-invalid-http-response |
| 4850 | no option accept-invalid-http-response |
| 4851 | Enable or disable relaxing of HTTP response parsing |
| 4852 | May be used in sections : defaults | frontend | listen | backend |
| 4853 | yes | no | yes | yes |
| 4854 | Arguments : none |
| 4855 | |
Willy Tarreau | 91852eb | 2015-05-01 13:26:00 +0200 | [diff] [blame] | 4856 | By default, HAProxy complies with RFC7230 in terms of message parsing. This |
Willy Tarreau | 4076a15 | 2009-04-02 15:18:36 +0200 | [diff] [blame] | 4857 | means that invalid characters in header names are not permitted and cause an |
| 4858 | error to be returned to the client. This is the desired behaviour as such |
| 4859 | forbidden characters are essentially used to build attacks exploiting server |
| 4860 | weaknesses, and bypass security filtering. Sometimes, a buggy browser or |
| 4861 | server will emit invalid header names for whatever reason (configuration, |
| 4862 | implementation) and the issue will not be immediately fixed. In such a case, |
| 4863 | it is possible to relax HAProxy's header name parser to accept any character |
Willy Tarreau | 91852eb | 2015-05-01 13:26:00 +0200 | [diff] [blame] | 4864 | even if that does not make sense, by specifying this option. This option also |
| 4865 | relaxes the test on the HTTP version format, it allows multiple digits for |
| 4866 | both the major and the minor version. |
Willy Tarreau | 4076a15 | 2009-04-02 15:18:36 +0200 | [diff] [blame] | 4867 | |
| 4868 | This option should never be enabled by default as it hides application bugs |
| 4869 | and open security breaches. It should only be deployed after a problem has |
| 4870 | been confirmed. |
| 4871 | |
| 4872 | When this option is enabled, erroneous header names will still be accepted in |
| 4873 | responses, but the complete response will be captured in order to permit |
| 4874 | later analysis using the "show errors" request on the UNIX stats socket. |
| 4875 | Doing this also helps confirming that the issue has been solved. |
| 4876 | |
| 4877 | If this option has been enabled in a "defaults" section, it can be disabled |
| 4878 | in a specific instance by prepending the "no" keyword before it. |
| 4879 | |
| 4880 | See also : "option accept-invalid-http-request" and "show errors" on the |
| 4881 | stats socket. |
| 4882 | |
| 4883 | |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 4884 | option allbackups |
| 4885 | no option allbackups |
| 4886 | Use either all backup servers at a time or only the first one |
| 4887 | May be used in sections : defaults | frontend | listen | backend |
| 4888 | yes | no | yes | yes |
| 4889 | Arguments : none |
| 4890 | |
| 4891 | By default, the first operational backup server gets all traffic when normal |
| 4892 | servers are all down. Sometimes, it may be preferred to use multiple backups |
| 4893 | at once, because one will not be enough. When "option allbackups" is enabled, |
| 4894 | the load balancing will be performed among all backup servers when all normal |
| 4895 | ones are unavailable. The same load balancing algorithm will be used and the |
| 4896 | servers' weights will be respected. Thus, there will not be any priority |
| 4897 | order between the backup servers anymore. |
| 4898 | |
| 4899 | This option is mostly used with static server farms dedicated to return a |
| 4900 | "sorry" page when an application is completely offline. |
| 4901 | |
| 4902 | If this option has been enabled in a "defaults" section, it can be disabled |
| 4903 | in a specific instance by prepending the "no" keyword before it. |
| 4904 | |
| 4905 | |
| 4906 | option checkcache |
| 4907 | no option checkcache |
Godbach | 7056a35 | 2013-12-11 20:01:07 +0800 | [diff] [blame] | 4908 | Analyze all server responses and block responses with cacheable cookies |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 4909 | May be used in sections : defaults | frontend | listen | backend |
| 4910 | yes | no | yes | yes |
| 4911 | Arguments : none |
| 4912 | |
| 4913 | Some high-level frameworks set application cookies everywhere and do not |
| 4914 | always let enough control to the developer to manage how the responses should |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 4915 | be cached. When a session cookie is returned on a cacheable object, there is a |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 4916 | high risk of session crossing or stealing between users traversing the same |
| 4917 | caches. In some situations, it is better to block the response than to let |
Willy Tarreau | 3c92c5f | 2011-08-28 09:45:47 +0200 | [diff] [blame] | 4918 | some sensitive session information go in the wild. |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 4919 | |
| 4920 | The option "checkcache" enables deep inspection of all server responses for |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 4921 | strict compliance with HTTP specification in terms of cacheability. It |
Willy Tarreau | 198a744 | 2008-01-17 12:05:32 +0100 | [diff] [blame] | 4922 | carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 4923 | response to check if there's a risk of caching a cookie on a client-side |
| 4924 | proxy. When this option is enabled, the only responses which can be delivered |
Willy Tarreau | 198a744 | 2008-01-17 12:05:32 +0100 | [diff] [blame] | 4925 | to the client are : |
| 4926 | - all those without "Set-Cookie" header ; |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 4927 | - all those with a return code other than 200, 203, 206, 300, 301, 410, |
Willy Tarreau | 198a744 | 2008-01-17 12:05:32 +0100 | [diff] [blame] | 4928 | provided that the server has not set a "Cache-control: public" header ; |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 4929 | - all those that come from a POST request, provided that the server has not |
| 4930 | set a 'Cache-Control: public' header ; |
| 4931 | - those with a 'Pragma: no-cache' header |
| 4932 | - those with a 'Cache-control: private' header |
| 4933 | - those with a 'Cache-control: no-store' header |
| 4934 | - those with a 'Cache-control: max-age=0' header |
| 4935 | - those with a 'Cache-control: s-maxage=0' header |
| 4936 | - those with a 'Cache-control: no-cache' header |
| 4937 | - those with a 'Cache-control: no-cache="set-cookie"' header |
| 4938 | - those with a 'Cache-control: no-cache="set-cookie,' header |
| 4939 | (allowing other fields after set-cookie) |
| 4940 | |
| 4941 | If a response doesn't respect these requirements, then it will be blocked |
Willy Tarreau | 198a744 | 2008-01-17 12:05:32 +0100 | [diff] [blame] | 4942 | just as if it was from an "rspdeny" filter, with an "HTTP 502 bad gateway". |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 4943 | The session state shows "PH--" meaning that the proxy blocked the response |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 4944 | during headers processing. Additionally, an alert will be sent in the logs so |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 4945 | that admins are informed that there's something to be fixed. |
| 4946 | |
| 4947 | Due to the high impact on the application, the application should be tested |
| 4948 | in depth with the option enabled before going to production. It is also a |
Willy Tarreau | d2a4aa2 | 2008-01-31 15:28:22 +0100 | [diff] [blame] | 4949 | good practice to always activate it during tests, even if it is not used in |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 4950 | production, as it will report potentially dangerous application behaviours. |
| 4951 | |
| 4952 | If this option has been enabled in a "defaults" section, it can be disabled |
| 4953 | in a specific instance by prepending the "no" keyword before it. |
| 4954 | |
| 4955 | |
| 4956 | option clitcpka |
| 4957 | no option clitcpka |
| 4958 | Enable or disable the sending of TCP keepalive packets on the client side |
| 4959 | May be used in sections : defaults | frontend | listen | backend |
| 4960 | yes | yes | yes | no |
| 4961 | Arguments : none |
| 4962 | |
| 4963 | When there is a firewall or any session-aware component between a client and |
| 4964 | a server, and when the protocol involves very long sessions with long idle |
| 4965 | periods (eg: remote desktops), there is a risk that one of the intermediate |
| 4966 | components decides to expire a session which has remained idle for too long. |
| 4967 | |
| 4968 | Enabling socket-level TCP keep-alives makes the system regularly send packets |
| 4969 | to the other end of the connection, leaving it active. The delay between |
| 4970 | keep-alive probes is controlled by the system only and depends both on the |
| 4971 | operating system and its tuning parameters. |
| 4972 | |
| 4973 | It is important to understand that keep-alive packets are neither emitted nor |
| 4974 | received at the application level. It is only the network stacks which sees |
| 4975 | them. For this reason, even if one side of the proxy already uses keep-alives |
| 4976 | to maintain its connection alive, those keep-alive packets will not be |
| 4977 | forwarded to the other side of the proxy. |
| 4978 | |
| 4979 | Please note that this has nothing to do with HTTP keep-alive. |
| 4980 | |
| 4981 | Using option "clitcpka" enables the emission of TCP keep-alive probes on the |
| 4982 | client side of a connection, which should help when session expirations are |
| 4983 | noticed between HAProxy and a client. |
| 4984 | |
| 4985 | If this option has been enabled in a "defaults" section, it can be disabled |
| 4986 | in a specific instance by prepending the "no" keyword before it. |
| 4987 | |
| 4988 | See also : "option srvtcpka", "option tcpka" |
| 4989 | |
| 4990 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 4991 | option contstats |
| 4992 | Enable continuous traffic statistics updates |
| 4993 | May be used in sections : defaults | frontend | listen | backend |
| 4994 | yes | yes | yes | no |
| 4995 | Arguments : none |
| 4996 | |
| 4997 | By default, counters used for statistics calculation are incremented |
| 4998 | only when a session finishes. It works quite well when serving small |
| 4999 | objects, but with big ones (for example large images or archives) or |
| 5000 | with A/V streaming, a graph generated from haproxy counters looks like |
| 5001 | a hedgehog. With this option enabled counters get incremented continuously, |
| 5002 | during a whole session. Recounting touches a hotpath directly so |
| 5003 | it is not enabled by default, as it has small performance impact (~0.5%). |
| 5004 | |
| 5005 | |
Willy Tarreau | c9bd0cc | 2009-05-10 11:57:02 +0200 | [diff] [blame] | 5006 | option dontlog-normal |
| 5007 | no option dontlog-normal |
| 5008 | Enable or disable logging of normal, successful connections |
| 5009 | May be used in sections : defaults | frontend | listen | backend |
| 5010 | yes | yes | yes | no |
| 5011 | Arguments : none |
| 5012 | |
| 5013 | There are large sites dealing with several thousand connections per second |
| 5014 | and for which logging is a major pain. Some of them are even forced to turn |
| 5015 | logs off and cannot debug production issues. Setting this option ensures that |
| 5016 | normal connections, those which experience no error, no timeout, no retry nor |
| 5017 | redispatch, will not be logged. This leaves disk space for anomalies. In HTTP |
| 5018 | mode, the response status code is checked and return codes 5xx will still be |
| 5019 | logged. |
| 5020 | |
| 5021 | It is strongly discouraged to use this option as most of the time, the key to |
| 5022 | complex issues is in the normal logs which will not be logged here. If you |
| 5023 | need to separate logs, see the "log-separate-errors" option instead. |
| 5024 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 5025 | See also : "log", "dontlognull", "log-separate-errors" and section 8 about |
Willy Tarreau | c9bd0cc | 2009-05-10 11:57:02 +0200 | [diff] [blame] | 5026 | logging. |
| 5027 | |
| 5028 | |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 5029 | option dontlognull |
| 5030 | no option dontlognull |
| 5031 | Enable or disable logging of null connections |
| 5032 | May be used in sections : defaults | frontend | listen | backend |
| 5033 | yes | yes | yes | no |
| 5034 | Arguments : none |
| 5035 | |
| 5036 | In certain environments, there are components which will regularly connect to |
| 5037 | various systems to ensure that they are still alive. It can be the case from |
| 5038 | another load balancer as well as from monitoring systems. By default, even a |
| 5039 | simple port probe or scan will produce a log. If those connections pollute |
| 5040 | the logs too much, it is possible to enable option "dontlognull" to indicate |
| 5041 | that a connection on which no data has been transferred will not be logged, |
Willy Tarreau | 0f228a0 | 2015-05-01 15:37:53 +0200 | [diff] [blame] | 5042 | which typically corresponds to those probes. Note that errors will still be |
| 5043 | returned to the client and accounted for in the stats. If this is not what is |
| 5044 | desired, option http-ignore-probes can be used instead. |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 5045 | |
| 5046 | It is generally recommended not to use this option in uncontrolled |
| 5047 | environments (eg: internet), otherwise scans and other malicious activities |
| 5048 | would not be logged. |
| 5049 | |
| 5050 | If this option has been enabled in a "defaults" section, it can be disabled |
| 5051 | in a specific instance by prepending the "no" keyword before it. |
| 5052 | |
Willy Tarreau | 0f228a0 | 2015-05-01 15:37:53 +0200 | [diff] [blame] | 5053 | See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and |
| 5054 | section 8 about logging. |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 5055 | |
| 5056 | |
| 5057 | option forceclose |
| 5058 | no option forceclose |
| 5059 | Enable or disable active connection closing after response is transferred. |
| 5060 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | a31e5df | 2009-12-30 01:10:35 +0100 | [diff] [blame] | 5061 | yes | yes | yes | yes |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 5062 | Arguments : none |
| 5063 | |
| 5064 | Some HTTP servers do not necessarily close the connections when they receive |
| 5065 | the "Connection: close" set by "option httpclose", and if the client does not |
| 5066 | close either, then the connection remains open till the timeout expires. This |
| 5067 | causes high number of simultaneous connections on the servers and shows high |
| 5068 | global session times in the logs. |
| 5069 | |
| 5070 | When this happens, it is possible to use "option forceclose". It will |
Willy Tarreau | 82eeaf2 | 2009-12-29 12:09:05 +0100 | [diff] [blame] | 5071 | actively close the outgoing server channel as soon as the server has finished |
Cyril Bonté | 653dcd6 | 2014-02-20 00:13:15 +0100 | [diff] [blame] | 5072 | to respond and release some resources earlier than with "option httpclose". |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 5073 | |
Willy Tarreau | 8a8e1d9 | 2010-04-05 16:15:16 +0200 | [diff] [blame] | 5074 | This option may also be combined with "option http-pretend-keepalive", which |
| 5075 | will disable sending of the "Connection: close" header, but will still cause |
| 5076 | the connection to be closed once the whole response is received. |
| 5077 | |
Cyril Bonté | 653dcd6 | 2014-02-20 00:13:15 +0100 | [diff] [blame] | 5078 | This option disables and replaces any previous "option httpclose", "option |
| 5079 | http-server-close", "option http-keep-alive", or "option http-tunnel". |
Willy Tarreau | 02bce8b | 2014-01-30 00:15:28 +0100 | [diff] [blame] | 5080 | |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 5081 | If this option has been enabled in a "defaults" section, it can be disabled |
| 5082 | in a specific instance by prepending the "no" keyword before it. |
| 5083 | |
Willy Tarreau | 8a8e1d9 | 2010-04-05 16:15:16 +0200 | [diff] [blame] | 5084 | See also : "option httpclose" and "option http-pretend-keepalive" |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 5085 | |
| 5086 | |
Willy Tarreau | 87cf514 | 2011-08-19 22:57:24 +0200 | [diff] [blame] | 5087 | option forwardfor [ except <network> ] [ header <name> ] [ if-none ] |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5088 | Enable insertion of the X-Forwarded-For header to requests sent to servers |
| 5089 | May be used in sections : defaults | frontend | listen | backend |
| 5090 | yes | yes | yes | yes |
| 5091 | Arguments : |
| 5092 | <network> is an optional argument used to disable this option for sources |
| 5093 | matching <network> |
Ross West | af72a1d | 2008-08-03 10:51:45 +0200 | [diff] [blame] | 5094 | <name> an optional argument to specify a different "X-Forwarded-For" |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 5095 | header name. |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5096 | |
| 5097 | Since HAProxy works in reverse-proxy mode, the servers see its IP address as |
| 5098 | their client address. This is sometimes annoying when the client's IP address |
| 5099 | is expected in server logs. To solve this problem, the well-known HTTP header |
| 5100 | "X-Forwarded-For" may be added by HAProxy to all requests sent to the server. |
| 5101 | This header contains a value representing the client's IP address. Since this |
| 5102 | header is always appended at the end of the existing header list, the server |
| 5103 | must be configured to always use the last occurrence of this header only. See |
Ross West | af72a1d | 2008-08-03 10:51:45 +0200 | [diff] [blame] | 5104 | the server's manual to find how to enable use of this standard header. Note |
| 5105 | that only the last occurrence of the header must be used, since it is really |
| 5106 | possible that the client has already brought one. |
| 5107 | |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 5108 | The keyword "header" may be used to supply a different header name to replace |
Ross West | af72a1d | 2008-08-03 10:51:45 +0200 | [diff] [blame] | 5109 | the default "X-Forwarded-For". This can be useful where you might already |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 5110 | have a "X-Forwarded-For" header from a different application (eg: stunnel), |
| 5111 | and you need preserve it. Also if your backend server doesn't use the |
Ross West | af72a1d | 2008-08-03 10:51:45 +0200 | [diff] [blame] | 5112 | "X-Forwarded-For" header and requires different one (eg: Zeus Web Servers |
| 5113 | require "X-Cluster-Client-IP"). |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5114 | |
| 5115 | Sometimes, a same HAProxy instance may be shared between a direct client |
| 5116 | access and a reverse-proxy access (for instance when an SSL reverse-proxy is |
| 5117 | used to decrypt HTTPS traffic). It is possible to disable the addition of the |
| 5118 | header for a known source address or network by adding the "except" keyword |
| 5119 | followed by the network address. In this case, any source IP matching the |
| 5120 | network will not cause an addition of this header. Most common uses are with |
| 5121 | private networks or 127.0.0.1. |
| 5122 | |
Willy Tarreau | 87cf514 | 2011-08-19 22:57:24 +0200 | [diff] [blame] | 5123 | Alternatively, the keyword "if-none" states that the header will only be |
| 5124 | added if it is not present. This should only be used in perfectly trusted |
| 5125 | environment, as this might cause a security issue if headers reaching haproxy |
| 5126 | are under the control of the end-user. |
| 5127 | |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5128 | This option may be specified either in the frontend or in the backend. If at |
Ross West | af72a1d | 2008-08-03 10:51:45 +0200 | [diff] [blame] | 5129 | least one of them uses it, the header will be added. Note that the backend's |
| 5130 | setting of the header subargument takes precedence over the frontend's if |
Willy Tarreau | 87cf514 | 2011-08-19 22:57:24 +0200 | [diff] [blame] | 5131 | both are defined. In the case of the "if-none" argument, if at least one of |
| 5132 | the frontend or the backend does not specify it, it wants the addition to be |
| 5133 | mandatory, so it wins. |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5134 | |
Ross West | af72a1d | 2008-08-03 10:51:45 +0200 | [diff] [blame] | 5135 | Examples : |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5136 | # Public HTTP address also used by stunnel on the same machine |
| 5137 | frontend www |
| 5138 | mode http |
| 5139 | option forwardfor except 127.0.0.1 # stunnel already adds the header |
| 5140 | |
Ross West | af72a1d | 2008-08-03 10:51:45 +0200 | [diff] [blame] | 5141 | # Those servers want the IP Address in X-Client |
| 5142 | backend www |
| 5143 | mode http |
| 5144 | option forwardfor header X-Client |
| 5145 | |
Willy Tarreau | 87cf514 | 2011-08-19 22:57:24 +0200 | [diff] [blame] | 5146 | See also : "option httpclose", "option http-server-close", |
Willy Tarreau | 70dffda | 2014-01-30 03:07:23 +0100 | [diff] [blame] | 5147 | "option forceclose", "option http-keep-alive" |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5148 | |
Willy Tarreau | 8a8e1d9 | 2010-04-05 16:15:16 +0200 | [diff] [blame] | 5149 | |
Willy Tarreau | 9fbe18e | 2015-05-01 22:42:08 +0200 | [diff] [blame] | 5150 | option http-buffer-request |
| 5151 | no option http-buffer-request |
| 5152 | Enable or disable waiting for whole HTTP request body before proceeding |
| 5153 | May be used in sections : defaults | frontend | listen | backend |
| 5154 | yes | yes | yes | yes |
| 5155 | Arguments : none |
| 5156 | |
| 5157 | It is sometimes desirable to wait for the body of an HTTP request before |
| 5158 | taking a decision. This is what is being done by "balance url_param" for |
| 5159 | example. The first use case is to buffer requests from slow clients before |
| 5160 | connecting to the server. Another use case consists in taking the routing |
| 5161 | decision based on the request body's contents. This option placed in a |
| 5162 | frontend or backend forces the HTTP processing to wait until either the whole |
| 5163 | body is received, or the request buffer is full, or the first chunk is |
| 5164 | complete in case of chunked encoding. It can have undesired side effects with |
| 5165 | some applications abusing HTTP by expecting unbufferred transmissions between |
| 5166 | the frontend and the backend, so this should definitely not be used by |
| 5167 | default. |
| 5168 | |
Baptiste Assmann | eccdf43 | 2015-10-28 13:49:01 +0100 | [diff] [blame] | 5169 | See also : "option http-no-delay", "timeout http-request" |
Willy Tarreau | 9fbe18e | 2015-05-01 22:42:08 +0200 | [diff] [blame] | 5170 | |
| 5171 | |
Willy Tarreau | 0f228a0 | 2015-05-01 15:37:53 +0200 | [diff] [blame] | 5172 | option http-ignore-probes |
| 5173 | no option http-ignore-probes |
| 5174 | Enable or disable logging of null connections and request timeouts |
| 5175 | May be used in sections : defaults | frontend | listen | backend |
| 5176 | yes | yes | yes | no |
| 5177 | Arguments : none |
| 5178 | |
| 5179 | Recently some browsers started to implement a "pre-connect" feature |
| 5180 | consisting in speculatively connecting to some recently visited web sites |
| 5181 | just in case the user would like to visit them. This results in many |
| 5182 | connections being established to web sites, which end up in 408 Request |
| 5183 | Timeout if the timeout strikes first, or 400 Bad Request when the browser |
| 5184 | decides to close them first. These ones pollute the log and feed the error |
| 5185 | counters. There was already "option dontlognull" but it's insufficient in |
| 5186 | this case. Instead, this option does the following things : |
| 5187 | - prevent any 400/408 message from being sent to the client if nothing |
| 5188 | was received over a connection before it was closed ; |
| 5189 | - prevent any log from being emitted in this situation ; |
| 5190 | - prevent any error counter from being incremented |
| 5191 | |
| 5192 | That way the empty connection is silently ignored. Note that it is better |
| 5193 | not to use this unless it is clear that it is needed, because it will hide |
| 5194 | real problems. The most common reason for not receiving a request and seeing |
| 5195 | a 408 is due to an MTU inconsistency between the client and an intermediary |
| 5196 | element such as a VPN, which blocks too large packets. These issues are |
| 5197 | generally seen with POST requests as well as GET with large cookies. The logs |
| 5198 | are often the only way to detect them. |
| 5199 | |
| 5200 | If this option has been enabled in a "defaults" section, it can be disabled |
| 5201 | in a specific instance by prepending the "no" keyword before it. |
| 5202 | |
| 5203 | See also : "log", "dontlognull", "errorfile", and section 8 about logging. |
| 5204 | |
| 5205 | |
Willy Tarreau | 16bfb02 | 2010-01-16 19:48:41 +0100 | [diff] [blame] | 5206 | option http-keep-alive |
| 5207 | no option http-keep-alive |
| 5208 | Enable or disable HTTP keep-alive from client to server |
| 5209 | May be used in sections : defaults | frontend | listen | backend |
| 5210 | yes | yes | yes | yes |
| 5211 | Arguments : none |
| 5212 | |
Willy Tarreau | 70dffda | 2014-01-30 03:07:23 +0100 | [diff] [blame] | 5213 | By default HAProxy operates in keep-alive mode with regards to persistent |
| 5214 | connections: for each connection it processes each request and response, and |
| 5215 | leaves the connection idle on both sides between the end of a response and the |
| 5216 | start of a new request. This mode may be changed by several options such as |
| 5217 | "option http-server-close", "option forceclose", "option httpclose" or |
| 5218 | "option http-tunnel". This option allows to set back the keep-alive mode, |
| 5219 | which can be useful when another mode was used in a defaults section. |
| 5220 | |
| 5221 | Setting "option http-keep-alive" enables HTTP keep-alive mode on the client- |
| 5222 | and server- sides. This provides the lowest latency on the client side (slow |
Willy Tarreau | 16bfb02 | 2010-01-16 19:48:41 +0100 | [diff] [blame] | 5223 | network) and the fastest session reuse on the server side at the expense |
| 5224 | of maintaining idle connections to the servers. In general, it is possible |
| 5225 | with this option to achieve approximately twice the request rate that the |
| 5226 | "http-server-close" option achieves on small objects. There are mainly two |
| 5227 | situations where this option may be useful : |
| 5228 | |
| 5229 | - when the server is non-HTTP compliant and authenticates the connection |
| 5230 | instead of requests (eg: NTLM authentication) |
| 5231 | |
| 5232 | - when the cost of establishing the connection to the server is significant |
| 5233 | compared to the cost of retrieving the associated object from the server. |
| 5234 | |
| 5235 | This last case can happen when the server is a fast static server of cache. |
| 5236 | In this case, the server will need to be properly tuned to support high enough |
| 5237 | connection counts because connections will last until the client sends another |
| 5238 | request. |
| 5239 | |
| 5240 | If the client request has to go to another backend or another server due to |
| 5241 | content switching or the load balancing algorithm, the idle connection will |
Willy Tarreau | 9420b12 | 2013-12-15 18:58:25 +0100 | [diff] [blame] | 5242 | immediately be closed and a new one re-opened. Option "prefer-last-server" is |
| 5243 | available to try optimize server selection so that if the server currently |
| 5244 | attached to an idle connection is usable, it will be used. |
Willy Tarreau | 16bfb02 | 2010-01-16 19:48:41 +0100 | [diff] [blame] | 5245 | |
| 5246 | In general it is preferred to use "option http-server-close" with application |
| 5247 | servers, and some static servers might benefit from "option http-keep-alive". |
| 5248 | |
| 5249 | At the moment, logs will not indicate whether requests came from the same |
| 5250 | session or not. The accept date reported in the logs corresponds to the end |
| 5251 | of the previous request, and the request time corresponds to the time spent |
| 5252 | waiting for a new request. The keep-alive request time is still bound to the |
| 5253 | timeout defined by "timeout http-keep-alive" or "timeout http-request" if |
| 5254 | not set. |
| 5255 | |
Cyril Bonté | 653dcd6 | 2014-02-20 00:13:15 +0100 | [diff] [blame] | 5256 | This option disables and replaces any previous "option httpclose", "option |
| 5257 | http-server-close", "option forceclose" or "option http-tunnel". When backend |
Willy Tarreau | 70dffda | 2014-01-30 03:07:23 +0100 | [diff] [blame] | 5258 | and frontend options differ, all of these 4 options have precedence over |
Cyril Bonté | 653dcd6 | 2014-02-20 00:13:15 +0100 | [diff] [blame] | 5259 | "option http-keep-alive". |
Willy Tarreau | 16bfb02 | 2010-01-16 19:48:41 +0100 | [diff] [blame] | 5260 | |
| 5261 | See also : "option forceclose", "option http-server-close", |
Willy Tarreau | 9420b12 | 2013-12-15 18:58:25 +0100 | [diff] [blame] | 5262 | "option prefer-last-server", "option http-pretend-keepalive", |
| 5263 | "option httpclose", and "1.1. The HTTP transaction model". |
Willy Tarreau | 16bfb02 | 2010-01-16 19:48:41 +0100 | [diff] [blame] | 5264 | |
| 5265 | |
Willy Tarreau | 96e3121 | 2011-05-30 18:10:30 +0200 | [diff] [blame] | 5266 | option http-no-delay |
| 5267 | no option http-no-delay |
| 5268 | Instruct the system to favor low interactive delays over performance in HTTP |
| 5269 | May be used in sections : defaults | frontend | listen | backend |
| 5270 | yes | yes | yes | yes |
| 5271 | Arguments : none |
| 5272 | |
| 5273 | In HTTP, each payload is unidirectional and has no notion of interactivity. |
| 5274 | Any agent is expected to queue data somewhat for a reasonably low delay. |
| 5275 | There are some very rare server-to-server applications that abuse the HTTP |
| 5276 | protocol and expect the payload phase to be highly interactive, with many |
| 5277 | interleaved data chunks in both directions within a single request. This is |
| 5278 | absolutely not supported by the HTTP specification and will not work across |
| 5279 | most proxies or servers. When such applications attempt to do this through |
| 5280 | haproxy, it works but they will experience high delays due to the network |
| 5281 | optimizations which favor performance by instructing the system to wait for |
| 5282 | enough data to be available in order to only send full packets. Typical |
| 5283 | delays are around 200 ms per round trip. Note that this only happens with |
| 5284 | abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not |
| 5285 | affected. |
| 5286 | |
| 5287 | When "option http-no-delay" is present in either the frontend or the backend |
| 5288 | used by a connection, all such optimizations will be disabled in order to |
| 5289 | make the exchanges as fast as possible. Of course this offers no guarantee on |
| 5290 | the functionality, as it may break at any other place. But if it works via |
| 5291 | HAProxy, it will work as fast as possible. This option should never be used |
| 5292 | by default, and should never be used at all unless such a buggy application |
| 5293 | is discovered. The impact of using this option is an increase of bandwidth |
| 5294 | usage and CPU usage, which may significantly lower performance in high |
| 5295 | latency environments. |
| 5296 | |
Willy Tarreau | 9fbe18e | 2015-05-01 22:42:08 +0200 | [diff] [blame] | 5297 | See also : "option http-buffer-request" |
| 5298 | |
Willy Tarreau | 96e3121 | 2011-05-30 18:10:30 +0200 | [diff] [blame] | 5299 | |
Willy Tarreau | 8a8e1d9 | 2010-04-05 16:15:16 +0200 | [diff] [blame] | 5300 | option http-pretend-keepalive |
| 5301 | no option http-pretend-keepalive |
| 5302 | Define whether haproxy will announce keepalive to the server or not |
| 5303 | May be used in sections : defaults | frontend | listen | backend |
| 5304 | yes | yes | yes | yes |
| 5305 | Arguments : none |
| 5306 | |
| 5307 | When running with "option http-server-close" or "option forceclose", haproxy |
| 5308 | adds a "Connection: close" header to the request forwarded to the server. |
| 5309 | Unfortunately, when some servers see this header, they automatically refrain |
| 5310 | from using the chunked encoding for responses of unknown length, while this |
| 5311 | is totally unrelated. The immediate effect is that this prevents haproxy from |
| 5312 | maintaining the client connection alive. A second effect is that a client or |
| 5313 | a cache could receive an incomplete response without being aware of it, and |
| 5314 | consider the response complete. |
| 5315 | |
| 5316 | By setting "option http-pretend-keepalive", haproxy will make the server |
| 5317 | believe it will keep the connection alive. The server will then not fall back |
| 5318 | to the abnormal undesired above. When haproxy gets the whole response, it |
| 5319 | will close the connection with the server just as it would do with the |
| 5320 | "forceclose" option. That way the client gets a normal response and the |
| 5321 | connection is correctly closed on the server side. |
| 5322 | |
| 5323 | It is recommended not to enable this option by default, because most servers |
| 5324 | will more efficiently close the connection themselves after the last packet, |
| 5325 | and release its buffers slightly earlier. Also, the added packet on the |
| 5326 | network could slightly reduce the overall peak performance. However it is |
| 5327 | worth noting that when this option is enabled, haproxy will have slightly |
| 5328 | less work to do. So if haproxy is the bottleneck on the whole architecture, |
| 5329 | enabling this option might save a few CPU cycles. |
| 5330 | |
| 5331 | This option may be set both in a frontend and in a backend. It is enabled if |
| 5332 | at least one of the frontend or backend holding a connection has it enabled. |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 5333 | This option may be combined with "option httpclose", which will cause |
Willy Tarreau | 22a9534 | 2010-09-29 14:31:41 +0200 | [diff] [blame] | 5334 | keepalive to be announced to the server and close to be announced to the |
| 5335 | client. This practice is discouraged though. |
Willy Tarreau | 8a8e1d9 | 2010-04-05 16:15:16 +0200 | [diff] [blame] | 5336 | |
| 5337 | If this option has been enabled in a "defaults" section, it can be disabled |
| 5338 | in a specific instance by prepending the "no" keyword before it. |
| 5339 | |
Willy Tarreau | 16bfb02 | 2010-01-16 19:48:41 +0100 | [diff] [blame] | 5340 | See also : "option forceclose", "option http-server-close", and |
| 5341 | "option http-keep-alive" |
Willy Tarreau | 8a8e1d9 | 2010-04-05 16:15:16 +0200 | [diff] [blame] | 5342 | |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5343 | |
Willy Tarreau | b608feb | 2010-01-02 22:47:18 +0100 | [diff] [blame] | 5344 | option http-server-close |
| 5345 | no option http-server-close |
| 5346 | Enable or disable HTTP connection closing on the server side |
| 5347 | May be used in sections : defaults | frontend | listen | backend |
| 5348 | yes | yes | yes | yes |
| 5349 | Arguments : none |
| 5350 | |
Willy Tarreau | 70dffda | 2014-01-30 03:07:23 +0100 | [diff] [blame] | 5351 | By default HAProxy operates in keep-alive mode with regards to persistent |
| 5352 | connections: for each connection it processes each request and response, and |
| 5353 | leaves the connection idle on both sides between the end of a response and |
| 5354 | the start of a new request. This mode may be changed by several options such |
| 5355 | as "option http-server-close", "option forceclose", "option httpclose" or |
| 5356 | "option http-tunnel". Setting "option http-server-close" enables HTTP |
| 5357 | connection-close mode on the server side while keeping the ability to support |
| 5358 | HTTP keep-alive and pipelining on the client side. This provides the lowest |
| 5359 | latency on the client side (slow network) and the fastest session reuse on |
| 5360 | the server side to save server resources, similarly to "option forceclose". |
| 5361 | It also permits non-keepalive capable servers to be served in keep-alive mode |
| 5362 | to the clients if they conform to the requirements of RFC2616. Please note |
| 5363 | that some servers do not always conform to those requirements when they see |
| 5364 | "Connection: close" in the request. The effect will be that keep-alive will |
| 5365 | never be used. A workaround consists in enabling "option |
| 5366 | http-pretend-keepalive". |
Willy Tarreau | b608feb | 2010-01-02 22:47:18 +0100 | [diff] [blame] | 5367 | |
| 5368 | At the moment, logs will not indicate whether requests came from the same |
| 5369 | session or not. The accept date reported in the logs corresponds to the end |
| 5370 | of the previous request, and the request time corresponds to the time spent |
| 5371 | waiting for a new request. The keep-alive request time is still bound to the |
Willy Tarreau | b16a574 | 2010-01-10 14:46:16 +0100 | [diff] [blame] | 5372 | timeout defined by "timeout http-keep-alive" or "timeout http-request" if |
| 5373 | not set. |
Willy Tarreau | b608feb | 2010-01-02 22:47:18 +0100 | [diff] [blame] | 5374 | |
| 5375 | This option may be set both in a frontend and in a backend. It is enabled if |
| 5376 | at least one of the frontend or backend holding a connection has it enabled. |
Cyril Bonté | 653dcd6 | 2014-02-20 00:13:15 +0100 | [diff] [blame] | 5377 | It disables and replaces any previous "option httpclose", "option forceclose", |
| 5378 | "option http-tunnel" or "option http-keep-alive". Please check section 4 |
Willy Tarreau | 70dffda | 2014-01-30 03:07:23 +0100 | [diff] [blame] | 5379 | ("Proxies") to see how this option combines with others when frontend and |
| 5380 | backend options differ. |
Willy Tarreau | b608feb | 2010-01-02 22:47:18 +0100 | [diff] [blame] | 5381 | |
| 5382 | If this option has been enabled in a "defaults" section, it can be disabled |
| 5383 | in a specific instance by prepending the "no" keyword before it. |
| 5384 | |
Patrick Mezard | 9ec2ec4 | 2010-06-12 17:02:45 +0200 | [diff] [blame] | 5385 | See also : "option forceclose", "option http-pretend-keepalive", |
Willy Tarreau | 16bfb02 | 2010-01-16 19:48:41 +0100 | [diff] [blame] | 5386 | "option httpclose", "option http-keep-alive", and |
| 5387 | "1.1. The HTTP transaction model". |
Willy Tarreau | b608feb | 2010-01-02 22:47:18 +0100 | [diff] [blame] | 5388 | |
| 5389 | |
Willy Tarreau | 02bce8b | 2014-01-30 00:15:28 +0100 | [diff] [blame] | 5390 | option http-tunnel |
| 5391 | no option http-tunnel |
| 5392 | Disable or enable HTTP connection processing after first transaction |
| 5393 | May be used in sections : defaults | frontend | listen | backend |
| 5394 | yes | yes | yes | yes |
| 5395 | Arguments : none |
| 5396 | |
Willy Tarreau | 70dffda | 2014-01-30 03:07:23 +0100 | [diff] [blame] | 5397 | By default HAProxy operates in keep-alive mode with regards to persistent |
| 5398 | connections: for each connection it processes each request and response, and |
| 5399 | leaves the connection idle on both sides between the end of a response and |
| 5400 | the start of a new request. This mode may be changed by several options such |
| 5401 | as "option http-server-close", "option forceclose", "option httpclose" or |
| 5402 | "option http-tunnel". |
| 5403 | |
| 5404 | Option "http-tunnel" disables any HTTP processing past the first request and |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 5405 | the first response. This is the mode which was used by default in versions |
Willy Tarreau | 70dffda | 2014-01-30 03:07:23 +0100 | [diff] [blame] | 5406 | 1.0 to 1.5-dev21. It is the mode with the lowest processing overhead, which |
| 5407 | is normally not needed anymore unless in very specific cases such as when |
| 5408 | using an in-house protocol that looks like HTTP but is not compatible, or |
| 5409 | just to log one request per client in order to reduce log size. Note that |
| 5410 | everything which works at the HTTP level, including header parsing/addition, |
| 5411 | cookie processing or content switching will only work for the first request |
| 5412 | and will be ignored after the first response. |
Willy Tarreau | 02bce8b | 2014-01-30 00:15:28 +0100 | [diff] [blame] | 5413 | |
| 5414 | If this option has been enabled in a "defaults" section, it can be disabled |
| 5415 | in a specific instance by prepending the "no" keyword before it. |
| 5416 | |
| 5417 | See also : "option forceclose", "option http-server-close", |
| 5418 | "option httpclose", "option http-keep-alive", and |
| 5419 | "1.1. The HTTP transaction model". |
| 5420 | |
| 5421 | |
Willy Tarreau | 88d349d | 2010-01-25 12:15:43 +0100 | [diff] [blame] | 5422 | option http-use-proxy-header |
Cyril Bonté | f0c6061 | 2010-02-06 14:44:47 +0100 | [diff] [blame] | 5423 | no option http-use-proxy-header |
Willy Tarreau | 88d349d | 2010-01-25 12:15:43 +0100 | [diff] [blame] | 5424 | Make use of non-standard Proxy-Connection header instead of Connection |
| 5425 | May be used in sections : defaults | frontend | listen | backend |
| 5426 | yes | yes | yes | no |
| 5427 | Arguments : none |
| 5428 | |
| 5429 | While RFC2616 explicitly states that HTTP/1.1 agents must use the |
| 5430 | Connection header to indicate their wish of persistent or non-persistent |
| 5431 | connections, both browsers and proxies ignore this header for proxied |
| 5432 | connections and make use of the undocumented, non-standard Proxy-Connection |
| 5433 | header instead. The issue begins when trying to put a load balancer between |
| 5434 | browsers and such proxies, because there will be a difference between what |
| 5435 | haproxy understands and what the client and the proxy agree on. |
| 5436 | |
| 5437 | By setting this option in a frontend, haproxy can automatically switch to use |
| 5438 | that non-standard header if it sees proxied requests. A proxied request is |
Lukas Tribus | f01a9cd | 2016-02-03 18:09:37 +0100 | [diff] [blame] | 5439 | defined here as one where the URI begins with neither a '/' nor a '*'. This |
| 5440 | is incompatible with the HTTP tunnel mode. Note that this option can only be |
| 5441 | specified in a frontend and will affect the request along its whole life. |
Willy Tarreau | 88d349d | 2010-01-25 12:15:43 +0100 | [diff] [blame] | 5442 | |
Willy Tarreau | 844a7e7 | 2010-01-31 21:46:18 +0100 | [diff] [blame] | 5443 | Also, when this option is set, a request which requires authentication will |
| 5444 | automatically switch to use proxy authentication headers if it is itself a |
| 5445 | proxied request. That makes it possible to check or enforce authentication in |
| 5446 | front of an existing proxy. |
| 5447 | |
Willy Tarreau | 88d349d | 2010-01-25 12:15:43 +0100 | [diff] [blame] | 5448 | This option should normally never be used, except in front of a proxy. |
| 5449 | |
| 5450 | See also : "option httpclose", "option forceclose" and "option |
| 5451 | http-server-close". |
| 5452 | |
| 5453 | |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 5454 | option httpchk |
| 5455 | option httpchk <uri> |
| 5456 | option httpchk <method> <uri> |
| 5457 | option httpchk <method> <uri> <version> |
| 5458 | Enable HTTP protocol to check on the servers health |
| 5459 | May be used in sections : defaults | frontend | listen | backend |
| 5460 | yes | no | yes | yes |
| 5461 | Arguments : |
| 5462 | <method> is the optional HTTP method used with the requests. When not set, |
| 5463 | the "OPTIONS" method is used, as it generally requires low server |
| 5464 | processing and is easy to filter out from the logs. Any method |
| 5465 | may be used, though it is not recommended to invent non-standard |
| 5466 | ones. |
| 5467 | |
| 5468 | <uri> is the URI referenced in the HTTP requests. It defaults to " / " |
| 5469 | which is accessible by default on almost any server, but may be |
| 5470 | changed to any other URI. Query strings are permitted. |
| 5471 | |
| 5472 | <version> is the optional HTTP version string. It defaults to "HTTP/1.0" |
| 5473 | but some servers might behave incorrectly in HTTP 1.0, so turning |
| 5474 | it to HTTP/1.1 may sometimes help. Note that the Host field is |
| 5475 | mandatory in HTTP/1.1, and as a trick, it is possible to pass it |
| 5476 | after "\r\n" following the version string. |
| 5477 | |
| 5478 | By default, server health checks only consist in trying to establish a TCP |
| 5479 | connection. When "option httpchk" is specified, a complete HTTP request is |
| 5480 | sent once the TCP connection is established, and responses 2xx and 3xx are |
| 5481 | considered valid, while all other ones indicate a server failure, including |
| 5482 | the lack of any response. |
| 5483 | |
| 5484 | The port and interval are specified in the server configuration. |
| 5485 | |
| 5486 | This option does not necessarily require an HTTP backend, it also works with |
| 5487 | plain TCP backends. This is particularly useful to check simple scripts bound |
| 5488 | to some dedicated ports using the inetd daemon. |
| 5489 | |
| 5490 | Examples : |
| 5491 | # Relay HTTPS traffic to Apache instance and check service availability |
| 5492 | # using HTTP request "OPTIONS * HTTP/1.1" on port 80. |
| 5493 | backend https_relay |
| 5494 | mode tcp |
| 5495 | option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www |
| 5496 | server apache1 192.168.1.1:443 check port 80 |
| 5497 | |
Simon Horman | afc47ee | 2013-11-25 10:46:35 +0900 | [diff] [blame] | 5498 | See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check", |
| 5499 | "option pgsql-check", "http-check" and the "check", "port" and |
| 5500 | "inter" server options. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 5501 | |
| 5502 | |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5503 | option httpclose |
| 5504 | no option httpclose |
| 5505 | Enable or disable passive HTTP connection closing |
| 5506 | May be used in sections : defaults | frontend | listen | backend |
| 5507 | yes | yes | yes | yes |
| 5508 | Arguments : none |
| 5509 | |
Willy Tarreau | 70dffda | 2014-01-30 03:07:23 +0100 | [diff] [blame] | 5510 | By default HAProxy operates in keep-alive mode with regards to persistent |
| 5511 | connections: for each connection it processes each request and response, and |
| 5512 | leaves the connection idle on both sides between the end of a response and |
| 5513 | the start of a new request. This mode may be changed by several options such |
Cyril Bonté | 653dcd6 | 2014-02-20 00:13:15 +0100 | [diff] [blame] | 5514 | as "option http-server-close", "option forceclose", "option httpclose" or |
Willy Tarreau | 70dffda | 2014-01-30 03:07:23 +0100 | [diff] [blame] | 5515 | "option http-tunnel". |
| 5516 | |
| 5517 | If "option httpclose" is set, HAProxy will work in HTTP tunnel mode and check |
| 5518 | if a "Connection: close" header is already set in each direction, and will |
| 5519 | add one if missing. Each end should react to this by actively closing the TCP |
| 5520 | connection after each transfer, thus resulting in a switch to the HTTP close |
| 5521 | mode. Any "Connection" header different from "close" will also be removed. |
| 5522 | Note that this option is deprecated since what it does is very cheap but not |
| 5523 | reliable. Using "option http-server-close" or "option forceclose" is strongly |
| 5524 | recommended instead. |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5525 | |
| 5526 | It seldom happens that some servers incorrectly ignore this header and do not |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 5527 | close the connection even though they reply "Connection: close". For this |
Willy Tarreau | 0dfdf19 | 2010-01-05 11:33:11 +0100 | [diff] [blame] | 5528 | reason, they are not compatible with older HTTP 1.0 browsers. If this happens |
| 5529 | it is possible to use the "option forceclose" which actively closes the |
| 5530 | request connection once the server responds. Option "forceclose" also |
| 5531 | releases the server connection earlier because it does not have to wait for |
| 5532 | the client to acknowledge it. |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5533 | |
| 5534 | This option may be set both in a frontend and in a backend. It is enabled if |
| 5535 | at least one of the frontend or backend holding a connection has it enabled. |
Cyril Bonté | 653dcd6 | 2014-02-20 00:13:15 +0100 | [diff] [blame] | 5536 | It disables and replaces any previous "option http-server-close", |
| 5537 | "option forceclose", "option http-keep-alive" or "option http-tunnel". Please |
Willy Tarreau | 70dffda | 2014-01-30 03:07:23 +0100 | [diff] [blame] | 5538 | check section 4 ("Proxies") to see how this option combines with others when |
| 5539 | frontend and backend options differ. |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5540 | |
| 5541 | If this option has been enabled in a "defaults" section, it can be disabled |
| 5542 | in a specific instance by prepending the "no" keyword before it. |
| 5543 | |
Patrick Mezard | 9ec2ec4 | 2010-06-12 17:02:45 +0200 | [diff] [blame] | 5544 | See also : "option forceclose", "option http-server-close" and |
| 5545 | "1.1. The HTTP transaction model". |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5546 | |
| 5547 | |
Emeric Brun | 3a058f3 | 2009-06-30 18:26:00 +0200 | [diff] [blame] | 5548 | option httplog [ clf ] |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5549 | Enable logging of HTTP request, session state and timers |
| 5550 | May be used in sections : defaults | frontend | listen | backend |
| 5551 | yes | yes | yes | yes |
Emeric Brun | 3a058f3 | 2009-06-30 18:26:00 +0200 | [diff] [blame] | 5552 | Arguments : |
| 5553 | clf if the "clf" argument is added, then the output format will be |
| 5554 | the CLF format instead of HAProxy's default HTTP format. You can |
| 5555 | use this when you need to feed HAProxy's logs through a specific |
| 5556 | log analyser which only support the CLF format and which is not |
| 5557 | extensible. |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5558 | |
| 5559 | By default, the log output format is very poor, as it only contains the |
| 5560 | source and destination addresses, and the instance name. By specifying |
| 5561 | "option httplog", each log line turns into a much richer format including, |
| 5562 | but not limited to, the HTTP request, the connection timers, the session |
| 5563 | status, the connections numbers, the captured headers and cookies, the |
| 5564 | frontend, backend and server name, and of course the source address and |
| 5565 | ports. |
| 5566 | |
| 5567 | This option may be set either in the frontend or the backend. |
| 5568 | |
PiBa-NL | bd556bf | 2014-12-11 21:31:54 +0100 | [diff] [blame] | 5569 | Specifying only "option httplog" will automatically clear the 'clf' mode |
| 5570 | if it was set by default. |
Emeric Brun | 3a058f3 | 2009-06-30 18:26:00 +0200 | [diff] [blame] | 5571 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 5572 | See also : section 8 about logging. |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5573 | |
Willy Tarreau | 55165fe | 2009-05-10 12:02:55 +0200 | [diff] [blame] | 5574 | |
| 5575 | option http_proxy |
| 5576 | no option http_proxy |
| 5577 | Enable or disable plain HTTP proxy mode |
| 5578 | May be used in sections : defaults | frontend | listen | backend |
| 5579 | yes | yes | yes | yes |
| 5580 | Arguments : none |
| 5581 | |
| 5582 | It sometimes happens that people need a pure HTTP proxy which understands |
| 5583 | basic proxy requests without caching nor any fancy feature. In this case, |
| 5584 | it may be worth setting up an HAProxy instance with the "option http_proxy" |
| 5585 | set. In this mode, no server is declared, and the connection is forwarded to |
| 5586 | the IP address and port found in the URL after the "http://" scheme. |
| 5587 | |
| 5588 | No host address resolution is performed, so this only works when pure IP |
| 5589 | addresses are passed. Since this option's usage perimeter is rather limited, |
Lukas Tribus | f01a9cd | 2016-02-03 18:09:37 +0100 | [diff] [blame] | 5590 | it will probably be used only by experts who know they need exactly it. This |
| 5591 | is incompatible with the HTTP tunnel mode. |
Willy Tarreau | 55165fe | 2009-05-10 12:02:55 +0200 | [diff] [blame] | 5592 | |
| 5593 | If this option has been enabled in a "defaults" section, it can be disabled |
| 5594 | in a specific instance by prepending the "no" keyword before it. |
| 5595 | |
| 5596 | Example : |
| 5597 | # this backend understands HTTP proxy requests and forwards them directly. |
| 5598 | backend direct_forward |
| 5599 | option httpclose |
| 5600 | option http_proxy |
| 5601 | |
| 5602 | See also : "option httpclose" |
| 5603 | |
Willy Tarreau | 211ad24 | 2009-10-03 21:45:07 +0200 | [diff] [blame] | 5604 | |
Jamie Gloudon | 801a0a3 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 5605 | option independent-streams |
| 5606 | no option independent-streams |
| 5607 | Enable or disable independent timeout processing for both directions |
Willy Tarreau | f27b5ea | 2009-10-03 22:01:18 +0200 | [diff] [blame] | 5608 | May be used in sections : defaults | frontend | listen | backend |
| 5609 | yes | yes | yes | yes |
| 5610 | Arguments : none |
| 5611 | |
| 5612 | By default, when data is sent over a socket, both the write timeout and the |
| 5613 | read timeout for that socket are refreshed, because we consider that there is |
| 5614 | activity on that socket, and we have no other means of guessing if we should |
| 5615 | receive data or not. |
| 5616 | |
| 5617 | While this default behaviour is desirable for almost all applications, there |
| 5618 | exists a situation where it is desirable to disable it, and only refresh the |
| 5619 | read timeout if there are incoming data. This happens on sessions with large |
| 5620 | timeouts and low amounts of exchanged data such as telnet session. If the |
| 5621 | server suddenly disappears, the output data accumulates in the system's |
| 5622 | socket buffers, both timeouts are correctly refreshed, and there is no way |
| 5623 | to know the server does not receive them, so we don't timeout. However, when |
| 5624 | the underlying protocol always echoes sent data, it would be enough by itself |
| 5625 | to detect the issue using the read timeout. Note that this problem does not |
| 5626 | happen with more verbose protocols because data won't accumulate long in the |
| 5627 | socket buffers. |
| 5628 | |
| 5629 | When this option is set on the frontend, it will disable read timeout updates |
| 5630 | on data sent to the client. There probably is little use of this case. When |
| 5631 | the option is set on the backend, it will disable read timeout updates on |
| 5632 | data sent to the server. Doing so will typically break large HTTP posts from |
| 5633 | slow lines, so use it with caution. |
| 5634 | |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 5635 | Note: older versions used to call this setting "option independent-streams" |
Jamie Gloudon | 801a0a3 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 5636 | with a spelling mistake. This spelling is still supported but |
| 5637 | deprecated. |
| 5638 | |
Willy Tarreau | ce887fd | 2012-05-12 12:50:00 +0200 | [diff] [blame] | 5639 | See also : "timeout client", "timeout server" and "timeout tunnel" |
Willy Tarreau | f27b5ea | 2009-10-03 22:01:18 +0200 | [diff] [blame] | 5640 | |
| 5641 | |
Gabor Lekeny | b4c81e4 | 2010-09-29 18:17:05 +0200 | [diff] [blame] | 5642 | option ldap-check |
| 5643 | Use LDAPv3 health checks for server testing |
| 5644 | May be used in sections : defaults | frontend | listen | backend |
| 5645 | yes | no | yes | yes |
| 5646 | Arguments : none |
| 5647 | |
| 5648 | It is possible to test that the server correctly talks LDAPv3 instead of just |
| 5649 | testing that it accepts the TCP connection. When this option is set, an |
| 5650 | LDAPv3 anonymous simple bind message is sent to the server, and the response |
| 5651 | is analyzed to find an LDAPv3 bind response message. |
| 5652 | |
| 5653 | The server is considered valid only when the LDAP response contains success |
| 5654 | resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9). |
| 5655 | |
| 5656 | Logging of bind requests is server dependent see your documentation how to |
| 5657 | configure it. |
| 5658 | |
| 5659 | Example : |
| 5660 | option ldap-check |
| 5661 | |
| 5662 | See also : "option httpchk" |
| 5663 | |
| 5664 | |
Simon Horman | 98637e5 | 2014-06-20 12:30:16 +0900 | [diff] [blame] | 5665 | option external-check |
| 5666 | Use external processes for server health checks |
| 5667 | May be used in sections : defaults | frontend | listen | backend |
| 5668 | yes | no | yes | yes |
| 5669 | |
| 5670 | It is possible to test the health of a server using an external command. |
| 5671 | This is achieved by running the executable set using "external-check |
| 5672 | command". |
| 5673 | |
| 5674 | Requires the "external-check" global to be set. |
| 5675 | |
| 5676 | See also : "external-check", "external-check command", "external-check path" |
| 5677 | |
| 5678 | |
Willy Tarreau | 211ad24 | 2009-10-03 21:45:07 +0200 | [diff] [blame] | 5679 | option log-health-checks |
| 5680 | no option log-health-checks |
Willy Tarreau | bef1b32 | 2014-05-13 21:01:39 +0200 | [diff] [blame] | 5681 | Enable or disable logging of health checks status updates |
Willy Tarreau | 211ad24 | 2009-10-03 21:45:07 +0200 | [diff] [blame] | 5682 | May be used in sections : defaults | frontend | listen | backend |
| 5683 | yes | no | yes | yes |
| 5684 | Arguments : none |
| 5685 | |
Willy Tarreau | bef1b32 | 2014-05-13 21:01:39 +0200 | [diff] [blame] | 5686 | By default, failed health check are logged if server is UP and successful |
| 5687 | health checks are logged if server is DOWN, so the amount of additional |
| 5688 | information is limited. |
Willy Tarreau | 211ad24 | 2009-10-03 21:45:07 +0200 | [diff] [blame] | 5689 | |
Willy Tarreau | bef1b32 | 2014-05-13 21:01:39 +0200 | [diff] [blame] | 5690 | When this option is enabled, any change of the health check status or to |
| 5691 | the server's health will be logged, so that it becomes possible to know |
| 5692 | that a server was failing occasional checks before crashing, or exactly when |
| 5693 | it failed to respond a valid HTTP status, then when the port started to |
| 5694 | reject connections, then when the server stopped responding at all. |
| 5695 | |
| 5696 | Note that status changes not caused by health checks (eg: enable/disable on |
| 5697 | the CLI) are intentionally not logged by this option. |
Willy Tarreau | 211ad24 | 2009-10-03 21:45:07 +0200 | [diff] [blame] | 5698 | |
Willy Tarreau | bef1b32 | 2014-05-13 21:01:39 +0200 | [diff] [blame] | 5699 | See also: "option httpchk", "option ldap-check", "option mysql-check", |
| 5700 | "option pgsql-check", "option redis-check", "option smtpchk", |
| 5701 | "option tcp-check", "log" and section 8 about logging. |
Willy Tarreau | 211ad24 | 2009-10-03 21:45:07 +0200 | [diff] [blame] | 5702 | |
Willy Tarreau | c9bd0cc | 2009-05-10 11:57:02 +0200 | [diff] [blame] | 5703 | |
| 5704 | option log-separate-errors |
| 5705 | no option log-separate-errors |
| 5706 | Change log level for non-completely successful connections |
| 5707 | May be used in sections : defaults | frontend | listen | backend |
| 5708 | yes | yes | yes | no |
| 5709 | Arguments : none |
| 5710 | |
| 5711 | Sometimes looking for errors in logs is not easy. This option makes haproxy |
| 5712 | raise the level of logs containing potentially interesting information such |
| 5713 | as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The |
| 5714 | level changes from "info" to "err". This makes it possible to log them |
| 5715 | separately to a different file with most syslog daemons. Be careful not to |
| 5716 | remove them from the original file, otherwise you would lose ordering which |
| 5717 | provides very important information. |
| 5718 | |
| 5719 | Using this option, large sites dealing with several thousand connections per |
| 5720 | second may log normal traffic to a rotating buffer and only archive smaller |
| 5721 | error logs. |
| 5722 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 5723 | See also : "log", "dontlognull", "dontlog-normal" and section 8 about |
Willy Tarreau | c9bd0cc | 2009-05-10 11:57:02 +0200 | [diff] [blame] | 5724 | logging. |
| 5725 | |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5726 | |
| 5727 | option logasap |
| 5728 | no option logasap |
| 5729 | Enable or disable early logging of HTTP requests |
| 5730 | May be used in sections : defaults | frontend | listen | backend |
| 5731 | yes | yes | yes | no |
| 5732 | Arguments : none |
| 5733 | |
| 5734 | By default, HTTP requests are logged upon termination so that the total |
| 5735 | transfer time and the number of bytes appear in the logs. When large objects |
| 5736 | are being transferred, it may take a while before the request appears in the |
| 5737 | logs. Using "option logasap", the request gets logged as soon as the server |
| 5738 | sends the complete headers. The only missing information in the logs will be |
| 5739 | the total number of bytes which will indicate everything except the amount |
| 5740 | of data transferred, and the total time which will not take the transfer |
Willy Tarreau | d2a4aa2 | 2008-01-31 15:28:22 +0100 | [diff] [blame] | 5741 | time into account. In such a situation, it's a good practice to capture the |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5742 | "Content-Length" response header so that the logs at least indicate how many |
| 5743 | bytes are expected to be transferred. |
| 5744 | |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 5745 | Examples : |
| 5746 | listen http_proxy 0.0.0.0:80 |
| 5747 | mode http |
| 5748 | option httplog |
| 5749 | option logasap |
| 5750 | log 192.168.2.200 local3 |
| 5751 | |
| 5752 | >>> Feb 6 12:14:14 localhost \ |
| 5753 | haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \ |
| 5754 | static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \ |
| 5755 | "GET /image.iso HTTP/1.0" |
| 5756 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 5757 | See also : "option httplog", "capture response header", and section 8 about |
Willy Tarreau | c27debf | 2008-01-06 08:57:02 +0100 | [diff] [blame] | 5758 | logging. |
| 5759 | |
| 5760 | |
Nenad Merdanovic | 6639a7c | 2014-05-30 14:26:32 +0200 | [diff] [blame] | 5761 | option mysql-check [ user <username> [ post-41 ] ] |
Hervé COMMOWICK | 8776f1b | 2010-10-18 15:58:36 +0200 | [diff] [blame] | 5762 | Use MySQL health checks for server testing |
Hervé COMMOWICK | 698ae00 | 2010-01-12 09:25:13 +0100 | [diff] [blame] | 5763 | May be used in sections : defaults | frontend | listen | backend |
| 5764 | yes | no | yes | yes |
Hervé COMMOWICK | 8776f1b | 2010-10-18 15:58:36 +0200 | [diff] [blame] | 5765 | Arguments : |
Cyril Bonté | 108cf6e | 2012-04-21 23:30:29 +0200 | [diff] [blame] | 5766 | <username> This is the username which will be used when connecting to MySQL |
| 5767 | server. |
Nenad Merdanovic | 6639a7c | 2014-05-30 14:26:32 +0200 | [diff] [blame] | 5768 | post-41 Send post v4.1 client compatible checks |
Hervé COMMOWICK | 8776f1b | 2010-10-18 15:58:36 +0200 | [diff] [blame] | 5769 | |
| 5770 | If you specify a username, the check consists of sending two MySQL packet, |
| 5771 | one Client Authentication packet, and one QUIT packet, to correctly close |
| 5772 | MySQL session. We then parse the MySQL Handshake Initialisation packet and/or |
| 5773 | Error packet. It is a basic but useful test which does not produce error nor |
| 5774 | aborted connect on the server. However, it requires adding an authorization |
| 5775 | in the MySQL table, like this : |
| 5776 | |
| 5777 | USE mysql; |
| 5778 | INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>'); |
| 5779 | FLUSH PRIVILEGES; |
| 5780 | |
| 5781 | If you don't specify a username (it is deprecated and not recommended), the |
| 5782 | check only consists in parsing the Mysql Handshake Initialisation packet or |
| 5783 | Error packet, we don't send anything in this mode. It was reported that it |
| 5784 | can generate lockout if check is too frequent and/or if there is not enough |
| 5785 | traffic. In fact, you need in this case to check MySQL "max_connect_errors" |
| 5786 | value as if a connection is established successfully within fewer than MySQL |
| 5787 | "max_connect_errors" attempts after a previous connection was interrupted, |
| 5788 | the error count for the host is cleared to zero. If HAProxy's server get |
| 5789 | blocked, the "FLUSH HOSTS" statement is the only way to unblock it. |
| 5790 | |
| 5791 | Remember that this does not check database presence nor database consistency. |
| 5792 | To do this, you can use an external check with xinetd for example. |
Hervé COMMOWICK | 698ae00 | 2010-01-12 09:25:13 +0100 | [diff] [blame] | 5793 | |
Hervé COMMOWICK | 212f778 | 2011-06-10 14:05:59 +0200 | [diff] [blame] | 5794 | The check requires MySQL >=3.22, for older version, please use TCP check. |
Hervé COMMOWICK | 698ae00 | 2010-01-12 09:25:13 +0100 | [diff] [blame] | 5795 | |
| 5796 | Most often, an incoming MySQL server needs to see the client's IP address for |
| 5797 | various purposes, including IP privilege matching and connection logging. |
| 5798 | When possible, it is often wise to masquerade the client's IP address when |
| 5799 | connecting to the server using the "usesrc" argument of the "source" keyword, |
Willy Tarreau | 29fbe51 | 2015-08-20 19:35:14 +0200 | [diff] [blame] | 5800 | which requires the transparent proxy feature to be compiled in, and the MySQL |
| 5801 | server to route the client via the machine hosting haproxy. |
Hervé COMMOWICK | 698ae00 | 2010-01-12 09:25:13 +0100 | [diff] [blame] | 5802 | |
| 5803 | See also: "option httpchk" |
| 5804 | |
| 5805 | |
Willy Tarreau | a453bdd | 2008-01-08 19:50:52 +0100 | [diff] [blame] | 5806 | option nolinger |
| 5807 | no option nolinger |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 5808 | Enable or disable immediate session resource cleaning after close |
Willy Tarreau | a453bdd | 2008-01-08 19:50:52 +0100 | [diff] [blame] | 5809 | May be used in sections: defaults | frontend | listen | backend |
| 5810 | yes | yes | yes | yes |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 5811 | Arguments : none |
Willy Tarreau | a453bdd | 2008-01-08 19:50:52 +0100 | [diff] [blame] | 5812 | |
| 5813 | When clients or servers abort connections in a dirty way (eg: they are |
| 5814 | physically disconnected), the session timeouts triggers and the session is |
| 5815 | closed. But it will remain in FIN_WAIT1 state for some time in the system, |
| 5816 | using some resources and possibly limiting the ability to establish newer |
| 5817 | connections. |
| 5818 | |
| 5819 | When this happens, it is possible to activate "option nolinger" which forces |
| 5820 | the system to immediately remove any socket's pending data on close. Thus, |
| 5821 | the session is instantly purged from the system's tables. This usually has |
| 5822 | side effects such as increased number of TCP resets due to old retransmits |
| 5823 | getting immediately rejected. Some firewalls may sometimes complain about |
| 5824 | this too. |
| 5825 | |
| 5826 | For this reason, it is not recommended to use this option when not absolutely |
| 5827 | needed. You know that you need it when you have thousands of FIN_WAIT1 |
| 5828 | sessions on your system (TIME_WAIT ones do not count). |
| 5829 | |
| 5830 | This option may be used both on frontends and backends, depending on the side |
| 5831 | where it is required. Use it on the frontend for clients, and on the backend |
| 5832 | for servers. |
| 5833 | |
| 5834 | If this option has been enabled in a "defaults" section, it can be disabled |
| 5835 | in a specific instance by prepending the "no" keyword before it. |
| 5836 | |
| 5837 | |
Willy Tarreau | 55165fe | 2009-05-10 12:02:55 +0200 | [diff] [blame] | 5838 | option originalto [ except <network> ] [ header <name> ] |
| 5839 | Enable insertion of the X-Original-To header to requests sent to servers |
| 5840 | May be used in sections : defaults | frontend | listen | backend |
| 5841 | yes | yes | yes | yes |
| 5842 | Arguments : |
| 5843 | <network> is an optional argument used to disable this option for sources |
| 5844 | matching <network> |
| 5845 | <name> an optional argument to specify a different "X-Original-To" |
| 5846 | header name. |
| 5847 | |
| 5848 | Since HAProxy can work in transparent mode, every request from a client can |
| 5849 | be redirected to the proxy and HAProxy itself can proxy every request to a |
| 5850 | complex SQUID environment and the destination host from SO_ORIGINAL_DST will |
| 5851 | be lost. This is annoying when you want access rules based on destination ip |
| 5852 | addresses. To solve this problem, a new HTTP header "X-Original-To" may be |
| 5853 | added by HAProxy to all requests sent to the server. This header contains a |
| 5854 | value representing the original destination IP address. Since this must be |
| 5855 | configured to always use the last occurrence of this header only. Note that |
| 5856 | only the last occurrence of the header must be used, since it is really |
| 5857 | possible that the client has already brought one. |
| 5858 | |
| 5859 | The keyword "header" may be used to supply a different header name to replace |
| 5860 | the default "X-Original-To". This can be useful where you might already |
| 5861 | have a "X-Original-To" header from a different application, and you need |
| 5862 | preserve it. Also if your backend server doesn't use the "X-Original-To" |
| 5863 | header and requires different one. |
| 5864 | |
| 5865 | Sometimes, a same HAProxy instance may be shared between a direct client |
| 5866 | access and a reverse-proxy access (for instance when an SSL reverse-proxy is |
| 5867 | used to decrypt HTTPS traffic). It is possible to disable the addition of the |
| 5868 | header for a known source address or network by adding the "except" keyword |
| 5869 | followed by the network address. In this case, any source IP matching the |
| 5870 | network will not cause an addition of this header. Most common uses are with |
| 5871 | private networks or 127.0.0.1. |
| 5872 | |
| 5873 | This option may be specified either in the frontend or in the backend. If at |
| 5874 | least one of them uses it, the header will be added. Note that the backend's |
| 5875 | setting of the header subargument takes precedence over the frontend's if |
| 5876 | both are defined. |
| 5877 | |
Willy Tarreau | 55165fe | 2009-05-10 12:02:55 +0200 | [diff] [blame] | 5878 | Examples : |
| 5879 | # Original Destination address |
| 5880 | frontend www |
| 5881 | mode http |
| 5882 | option originalto except 127.0.0.1 |
| 5883 | |
| 5884 | # Those servers want the IP Address in X-Client-Dst |
| 5885 | backend www |
| 5886 | mode http |
| 5887 | option originalto header X-Client-Dst |
| 5888 | |
Willy Tarreau | 87cf514 | 2011-08-19 22:57:24 +0200 | [diff] [blame] | 5889 | See also : "option httpclose", "option http-server-close", |
| 5890 | "option forceclose" |
Willy Tarreau | 55165fe | 2009-05-10 12:02:55 +0200 | [diff] [blame] | 5891 | |
| 5892 | |
Willy Tarreau | a453bdd | 2008-01-08 19:50:52 +0100 | [diff] [blame] | 5893 | option persist |
| 5894 | no option persist |
| 5895 | Enable or disable forced persistence on down servers |
| 5896 | May be used in sections: defaults | frontend | listen | backend |
| 5897 | yes | no | yes | yes |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 5898 | Arguments : none |
Willy Tarreau | a453bdd | 2008-01-08 19:50:52 +0100 | [diff] [blame] | 5899 | |
| 5900 | When an HTTP request reaches a backend with a cookie which references a dead |
| 5901 | server, by default it is redispatched to another server. It is possible to |
| 5902 | force the request to be sent to the dead server first using "option persist" |
| 5903 | if absolutely needed. A common use case is when servers are under extreme |
| 5904 | load and spend their time flapping. In this case, the users would still be |
| 5905 | directed to the server they opened the session on, in the hope they would be |
| 5906 | correctly served. It is recommended to use "option redispatch" in conjunction |
| 5907 | with this option so that in the event it would not be possible to connect to |
| 5908 | the server at all (server definitely dead), the client would finally be |
| 5909 | redirected to another valid server. |
| 5910 | |
| 5911 | If this option has been enabled in a "defaults" section, it can be disabled |
| 5912 | in a specific instance by prepending the "no" keyword before it. |
| 5913 | |
Willy Tarreau | 4de9149 | 2010-01-22 19:10:05 +0100 | [diff] [blame] | 5914 | See also : "option redispatch", "retries", "force-persist" |
Willy Tarreau | a453bdd | 2008-01-08 19:50:52 +0100 | [diff] [blame] | 5915 | |
| 5916 | |
Willy Tarreau | 0c12282 | 2013-12-15 18:49:01 +0100 | [diff] [blame] | 5917 | option pgsql-check [ user <username> ] |
| 5918 | Use PostgreSQL health checks for server testing |
| 5919 | May be used in sections : defaults | frontend | listen | backend |
| 5920 | yes | no | yes | yes |
| 5921 | Arguments : |
| 5922 | <username> This is the username which will be used when connecting to |
| 5923 | PostgreSQL server. |
| 5924 | |
| 5925 | The check sends a PostgreSQL StartupMessage and waits for either |
| 5926 | Authentication request or ErrorResponse message. It is a basic but useful |
| 5927 | test which does not produce error nor aborted connect on the server. |
| 5928 | This check is identical with the "mysql-check". |
| 5929 | |
| 5930 | See also: "option httpchk" |
| 5931 | |
| 5932 | |
Willy Tarreau | 9420b12 | 2013-12-15 18:58:25 +0100 | [diff] [blame] | 5933 | option prefer-last-server |
| 5934 | no option prefer-last-server |
| 5935 | Allow multiple load balanced requests to remain on the same server |
| 5936 | May be used in sections: defaults | frontend | listen | backend |
| 5937 | yes | no | yes | yes |
| 5938 | Arguments : none |
| 5939 | |
| 5940 | When the load balancing algorithm in use is not deterministic, and a previous |
| 5941 | request was sent to a server to which haproxy still holds a connection, it is |
| 5942 | sometimes desirable that subsequent requests on a same session go to the same |
| 5943 | server as much as possible. Note that this is different from persistence, as |
| 5944 | we only indicate a preference which haproxy tries to apply without any form |
| 5945 | of warranty. The real use is for keep-alive connections sent to servers. When |
| 5946 | this option is used, haproxy will try to reuse the same connection that is |
| 5947 | attached to the server instead of rebalancing to another server, causing a |
| 5948 | close of the connection. This can make sense for static file servers. It does |
Willy Tarreau | 068621e | 2013-12-23 15:11:25 +0100 | [diff] [blame] | 5949 | not make much sense to use this in combination with hashing algorithms. Note, |
| 5950 | haproxy already automatically tries to stick to a server which sends a 401 or |
| 5951 | to a proxy which sends a 407 (authentication required). This is mandatory for |
| 5952 | use with the broken NTLM authentication challenge, and significantly helps in |
| 5953 | troubleshooting some faulty applications. Option prefer-last-server might be |
| 5954 | desirable in these environments as well, to avoid redistributing the traffic |
| 5955 | after every other response. |
Willy Tarreau | 9420b12 | 2013-12-15 18:58:25 +0100 | [diff] [blame] | 5956 | |
| 5957 | If this option has been enabled in a "defaults" section, it can be disabled |
| 5958 | in a specific instance by prepending the "no" keyword before it. |
| 5959 | |
| 5960 | See also: "option http-keep-alive" |
| 5961 | |
| 5962 | |
Krzysztof Piotr Oledzki | 25b501a | 2008-01-06 16:36:16 +0100 | [diff] [blame] | 5963 | option redispatch |
Joseph Lynch | 726ab71 | 2015-05-11 23:25:34 -0700 | [diff] [blame] | 5964 | option redispatch <interval> |
Krzysztof Piotr Oledzki | 25b501a | 2008-01-06 16:36:16 +0100 | [diff] [blame] | 5965 | no option redispatch |
| 5966 | Enable or disable session redistribution in case of connection failure |
| 5967 | May be used in sections: defaults | frontend | listen | backend |
| 5968 | yes | no | yes | yes |
Joseph Lynch | 726ab71 | 2015-05-11 23:25:34 -0700 | [diff] [blame] | 5969 | Arguments : |
| 5970 | <interval> The optional integer value that controls how often redispatches |
| 5971 | occur when retrying connections. Positive value P indicates a |
| 5972 | redispatch is desired on every Pth retry, and negative value |
| 5973 | N indicate a redispath is desired on the Nth retry prior to the |
| 5974 | last retry. For example, the default of -1 preserves the |
| 5975 | historical behaviour of redispatching on the last retry, a |
| 5976 | positive value of 1 would indicate a redispatch on every retry, |
| 5977 | and a positive value of 3 would indicate a redispatch on every |
| 5978 | third retry. You can disable redispatches with a value of 0. |
| 5979 | |
Krzysztof Piotr Oledzki | 25b501a | 2008-01-06 16:36:16 +0100 | [diff] [blame] | 5980 | |
| 5981 | In HTTP mode, if a server designated by a cookie is down, clients may |
| 5982 | definitely stick to it because they cannot flush the cookie, so they will not |
| 5983 | be able to access the service anymore. |
| 5984 | |
| 5985 | Specifying "option redispatch" will allow the proxy to break their |
| 5986 | persistence and redistribute them to a working server. |
| 5987 | |
Joseph Lynch | 726ab71 | 2015-05-11 23:25:34 -0700 | [diff] [blame] | 5988 | It also allows to retry connections to another server in case of multiple |
Krzysztof Piotr Oledzki | 25b501a | 2008-01-06 16:36:16 +0100 | [diff] [blame] | 5989 | connection failures. Of course, it requires having "retries" set to a nonzero |
| 5990 | value. |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 5991 | |
Krzysztof Piotr Oledzki | 25b501a | 2008-01-06 16:36:16 +0100 | [diff] [blame] | 5992 | This form is the preferred form, which replaces both the "redispatch" and |
| 5993 | "redisp" keywords. |
| 5994 | |
| 5995 | If this option has been enabled in a "defaults" section, it can be disabled |
| 5996 | in a specific instance by prepending the "no" keyword before it. |
| 5997 | |
Willy Tarreau | 4de9149 | 2010-01-22 19:10:05 +0100 | [diff] [blame] | 5998 | See also : "redispatch", "retries", "force-persist" |
Krzysztof Piotr Oledzki | 25b501a | 2008-01-06 16:36:16 +0100 | [diff] [blame] | 5999 | |
Willy Tarreau | a453bdd | 2008-01-08 19:50:52 +0100 | [diff] [blame] | 6000 | |
Hervé COMMOWICK | ec032d6 | 2011-08-05 16:23:48 +0200 | [diff] [blame] | 6001 | option redis-check |
| 6002 | Use redis health checks for server testing |
| 6003 | May be used in sections : defaults | frontend | listen | backend |
| 6004 | yes | no | yes | yes |
| 6005 | Arguments : none |
| 6006 | |
| 6007 | It is possible to test that the server correctly talks REDIS protocol instead |
| 6008 | of just testing that it accepts the TCP connection. When this option is set, |
| 6009 | a PING redis command is sent to the server, and the response is analyzed to |
| 6010 | find the "+PONG" response message. |
| 6011 | |
| 6012 | Example : |
| 6013 | option redis-check |
| 6014 | |
| 6015 | See also : "option httpchk" |
| 6016 | |
| 6017 | |
Willy Tarreau | a453bdd | 2008-01-08 19:50:52 +0100 | [diff] [blame] | 6018 | option smtpchk |
| 6019 | option smtpchk <hello> <domain> |
| 6020 | Use SMTP health checks for server testing |
| 6021 | May be used in sections : defaults | frontend | listen | backend |
| 6022 | yes | no | yes | yes |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 6023 | Arguments : |
Willy Tarreau | a453bdd | 2008-01-08 19:50:52 +0100 | [diff] [blame] | 6024 | <hello> is an optional argument. It is the "hello" command to use. It can |
| 6025 | be either "HELO" (for SMTP) or "EHLO" (for ESTMP). All other |
| 6026 | values will be turned into the default command ("HELO"). |
| 6027 | |
| 6028 | <domain> is the domain name to present to the server. It may only be |
| 6029 | specified (and is mandatory) if the hello command has been |
| 6030 | specified. By default, "localhost" is used. |
| 6031 | |
| 6032 | When "option smtpchk" is set, the health checks will consist in TCP |
| 6033 | connections followed by an SMTP command. By default, this command is |
| 6034 | "HELO localhost". The server's return code is analyzed and only return codes |
| 6035 | starting with a "2" will be considered as valid. All other responses, |
| 6036 | including a lack of response will constitute an error and will indicate a |
| 6037 | dead server. |
| 6038 | |
| 6039 | This test is meant to be used with SMTP servers or relays. Depending on the |
| 6040 | request, it is possible that some servers do not log each connection attempt, |
| 6041 | so you may want to experiment to improve the behaviour. Using telnet on port |
| 6042 | 25 is often easier than adjusting the configuration. |
| 6043 | |
| 6044 | Most often, an incoming SMTP server needs to see the client's IP address for |
| 6045 | various purposes, including spam filtering, anti-spoofing and logging. When |
| 6046 | possible, it is often wise to masquerade the client's IP address when |
| 6047 | connecting to the server using the "usesrc" argument of the "source" keyword, |
Willy Tarreau | 29fbe51 | 2015-08-20 19:35:14 +0200 | [diff] [blame] | 6048 | which requires the transparent proxy feature to be compiled in. |
Willy Tarreau | a453bdd | 2008-01-08 19:50:52 +0100 | [diff] [blame] | 6049 | |
| 6050 | Example : |
| 6051 | option smtpchk HELO mydomain.org |
| 6052 | |
| 6053 | See also : "option httpchk", "source" |
| 6054 | |
Krzysztof Piotr Oledzki | 25b501a | 2008-01-06 16:36:16 +0100 | [diff] [blame] | 6055 | |
Krzysztof Piotr Oledzki | aeebf9b | 2009-10-04 15:43:17 +0200 | [diff] [blame] | 6056 | option socket-stats |
| 6057 | no option socket-stats |
| 6058 | |
| 6059 | Enable or disable collecting & providing separate statistics for each socket. |
| 6060 | May be used in sections : defaults | frontend | listen | backend |
| 6061 | yes | yes | yes | no |
| 6062 | |
| 6063 | Arguments : none |
| 6064 | |
| 6065 | |
Willy Tarreau | ff4f82d | 2009-02-06 11:28:13 +0100 | [diff] [blame] | 6066 | option splice-auto |
| 6067 | no option splice-auto |
| 6068 | Enable or disable automatic kernel acceleration on sockets in both directions |
| 6069 | May be used in sections : defaults | frontend | listen | backend |
| 6070 | yes | yes | yes | yes |
| 6071 | Arguments : none |
| 6072 | |
| 6073 | When this option is enabled either on a frontend or on a backend, haproxy |
| 6074 | will automatically evaluate the opportunity to use kernel tcp splicing to |
| 6075 | forward data between the client and the server, in either direction. Haproxy |
| 6076 | uses heuristics to estimate if kernel splicing might improve performance or |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 6077 | not. Both directions are handled independently. Note that the heuristics used |
Willy Tarreau | ff4f82d | 2009-02-06 11:28:13 +0100 | [diff] [blame] | 6078 | are not much aggressive in order to limit excessive use of splicing. This |
| 6079 | option requires splicing to be enabled at compile time, and may be globally |
| 6080 | disabled with the global option "nosplice". Since splice uses pipes, using it |
| 6081 | requires that there are enough spare pipes. |
| 6082 | |
| 6083 | Important note: kernel-based TCP splicing is a Linux-specific feature which |
| 6084 | first appeared in kernel 2.6.25. It offers kernel-based acceleration to |
| 6085 | transfer data between sockets without copying these data to user-space, thus |
| 6086 | providing noticeable performance gains and CPU cycles savings. Since many |
| 6087 | early implementations are buggy, corrupt data and/or are inefficient, this |
| 6088 | feature is not enabled by default, and it should be used with extreme care. |
| 6089 | While it is not possible to detect the correctness of an implementation, |
| 6090 | 2.6.29 is the first version offering a properly working implementation. In |
| 6091 | case of doubt, splicing may be globally disabled using the global "nosplice" |
| 6092 | keyword. |
| 6093 | |
| 6094 | Example : |
| 6095 | option splice-auto |
| 6096 | |
| 6097 | If this option has been enabled in a "defaults" section, it can be disabled |
| 6098 | in a specific instance by prepending the "no" keyword before it. |
| 6099 | |
| 6100 | See also : "option splice-request", "option splice-response", and global |
| 6101 | options "nosplice" and "maxpipes" |
| 6102 | |
| 6103 | |
| 6104 | option splice-request |
| 6105 | no option splice-request |
| 6106 | Enable or disable automatic kernel acceleration on sockets for requests |
| 6107 | May be used in sections : defaults | frontend | listen | backend |
| 6108 | yes | yes | yes | yes |
| 6109 | Arguments : none |
| 6110 | |
| 6111 | When this option is enabled either on a frontend or on a backend, haproxy |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 6112 | will use kernel tcp splicing whenever possible to forward data going from |
Willy Tarreau | ff4f82d | 2009-02-06 11:28:13 +0100 | [diff] [blame] | 6113 | the client to the server. It might still use the recv/send scheme if there |
| 6114 | are no spare pipes left. This option requires splicing to be enabled at |
| 6115 | compile time, and may be globally disabled with the global option "nosplice". |
| 6116 | Since splice uses pipes, using it requires that there are enough spare pipes. |
| 6117 | |
| 6118 | Important note: see "option splice-auto" for usage limitations. |
| 6119 | |
| 6120 | Example : |
| 6121 | option splice-request |
| 6122 | |
| 6123 | If this option has been enabled in a "defaults" section, it can be disabled |
| 6124 | in a specific instance by prepending the "no" keyword before it. |
| 6125 | |
| 6126 | See also : "option splice-auto", "option splice-response", and global options |
| 6127 | "nosplice" and "maxpipes" |
| 6128 | |
| 6129 | |
| 6130 | option splice-response |
| 6131 | no option splice-response |
| 6132 | Enable or disable automatic kernel acceleration on sockets for responses |
| 6133 | May be used in sections : defaults | frontend | listen | backend |
| 6134 | yes | yes | yes | yes |
| 6135 | Arguments : none |
| 6136 | |
| 6137 | When this option is enabled either on a frontend or on a backend, haproxy |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 6138 | will use kernel tcp splicing whenever possible to forward data going from |
Willy Tarreau | ff4f82d | 2009-02-06 11:28:13 +0100 | [diff] [blame] | 6139 | the server to the client. It might still use the recv/send scheme if there |
| 6140 | are no spare pipes left. This option requires splicing to be enabled at |
| 6141 | compile time, and may be globally disabled with the global option "nosplice". |
| 6142 | Since splice uses pipes, using it requires that there are enough spare pipes. |
| 6143 | |
| 6144 | Important note: see "option splice-auto" for usage limitations. |
| 6145 | |
| 6146 | Example : |
| 6147 | option splice-response |
| 6148 | |
| 6149 | If this option has been enabled in a "defaults" section, it can be disabled |
| 6150 | in a specific instance by prepending the "no" keyword before it. |
| 6151 | |
| 6152 | See also : "option splice-auto", "option splice-request", and global options |
| 6153 | "nosplice" and "maxpipes" |
| 6154 | |
| 6155 | |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 6156 | option srvtcpka |
| 6157 | no option srvtcpka |
| 6158 | Enable or disable the sending of TCP keepalive packets on the server side |
| 6159 | May be used in sections : defaults | frontend | listen | backend |
| 6160 | yes | no | yes | yes |
| 6161 | Arguments : none |
| 6162 | |
| 6163 | When there is a firewall or any session-aware component between a client and |
| 6164 | a server, and when the protocol involves very long sessions with long idle |
| 6165 | periods (eg: remote desktops), there is a risk that one of the intermediate |
| 6166 | components decides to expire a session which has remained idle for too long. |
| 6167 | |
| 6168 | Enabling socket-level TCP keep-alives makes the system regularly send packets |
| 6169 | to the other end of the connection, leaving it active. The delay between |
| 6170 | keep-alive probes is controlled by the system only and depends both on the |
| 6171 | operating system and its tuning parameters. |
| 6172 | |
| 6173 | It is important to understand that keep-alive packets are neither emitted nor |
| 6174 | received at the application level. It is only the network stacks which sees |
| 6175 | them. For this reason, even if one side of the proxy already uses keep-alives |
| 6176 | to maintain its connection alive, those keep-alive packets will not be |
| 6177 | forwarded to the other side of the proxy. |
| 6178 | |
| 6179 | Please note that this has nothing to do with HTTP keep-alive. |
| 6180 | |
| 6181 | Using option "srvtcpka" enables the emission of TCP keep-alive probes on the |
| 6182 | server side of a connection, which should help when session expirations are |
| 6183 | noticed between HAProxy and a server. |
| 6184 | |
| 6185 | If this option has been enabled in a "defaults" section, it can be disabled |
| 6186 | in a specific instance by prepending the "no" keyword before it. |
| 6187 | |
| 6188 | See also : "option clitcpka", "option tcpka" |
| 6189 | |
| 6190 | |
Willy Tarreau | a453bdd | 2008-01-08 19:50:52 +0100 | [diff] [blame] | 6191 | option ssl-hello-chk |
| 6192 | Use SSLv3 client hello health checks for server testing |
| 6193 | May be used in sections : defaults | frontend | listen | backend |
| 6194 | yes | no | yes | yes |
| 6195 | Arguments : none |
| 6196 | |
| 6197 | When some SSL-based protocols are relayed in TCP mode through HAProxy, it is |
| 6198 | possible to test that the server correctly talks SSL instead of just testing |
| 6199 | that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure |
| 6200 | SSLv3 client hello messages are sent once the connection is established to |
| 6201 | the server, and the response is analyzed to find an SSL server hello message. |
| 6202 | The server is considered valid only when the response contains this server |
| 6203 | hello message. |
| 6204 | |
| 6205 | All servers tested till there correctly reply to SSLv3 client hello messages, |
| 6206 | and most servers tested do not even log the requests containing only hello |
| 6207 | messages, which is appreciable. |
| 6208 | |
Willy Tarreau | 763a95b | 2012-10-04 23:15:39 +0200 | [diff] [blame] | 6209 | Note that this check works even when SSL support was not built into haproxy |
| 6210 | because it forges the SSL message. When SSL support is available, it is best |
| 6211 | to use native SSL health checks instead of this one. |
Willy Tarreau | a453bdd | 2008-01-08 19:50:52 +0100 | [diff] [blame] | 6212 | |
Willy Tarreau | 763a95b | 2012-10-04 23:15:39 +0200 | [diff] [blame] | 6213 | See also: "option httpchk", "check-ssl" |
| 6214 | |
Willy Tarreau | a453bdd | 2008-01-08 19:50:52 +0100 | [diff] [blame] | 6215 | |
Willy Tarreau | ed17985 | 2013-12-16 01:07:00 +0100 | [diff] [blame] | 6216 | option tcp-check |
| 6217 | Perform health checks using tcp-check send/expect sequences |
| 6218 | May be used in sections: defaults | frontend | listen | backend |
| 6219 | yes | no | yes | yes |
| 6220 | |
| 6221 | This health check method is intended to be combined with "tcp-check" command |
| 6222 | lists in order to support send/expect types of health check sequences. |
| 6223 | |
| 6224 | TCP checks currently support 4 modes of operations : |
| 6225 | - no "tcp-check" directive : the health check only consists in a connection |
| 6226 | attempt, which remains the default mode. |
| 6227 | |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 6228 | - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is |
Willy Tarreau | ed17985 | 2013-12-16 01:07:00 +0100 | [diff] [blame] | 6229 | used to send a string along with a connection opening. With some |
| 6230 | protocols, it helps sending a "QUIT" message for example that prevents |
| 6231 | the server from logging a connection error for each health check. The |
| 6232 | check result will still be based on the ability to open the connection |
| 6233 | only. |
| 6234 | |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 6235 | - "tcp-check expect" only is mentioned : this is used to test a banner. |
Willy Tarreau | ed17985 | 2013-12-16 01:07:00 +0100 | [diff] [blame] | 6236 | The connection is opened and haproxy waits for the server to present some |
| 6237 | contents which must validate some rules. The check result will be based |
| 6238 | on the matching between the contents and the rules. This is suited for |
| 6239 | POP, IMAP, SMTP, FTP, SSH, TELNET. |
| 6240 | |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 6241 | - both "tcp-check send" and "tcp-check expect" are mentioned : this is |
Willy Tarreau | ed17985 | 2013-12-16 01:07:00 +0100 | [diff] [blame] | 6242 | used to test a hello-type protocol. Haproxy sends a message, the server |
| 6243 | responds and its response is analysed. the check result will be based on |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 6244 | the matching between the response contents and the rules. This is often |
Willy Tarreau | ed17985 | 2013-12-16 01:07:00 +0100 | [diff] [blame] | 6245 | suited for protocols which require a binding or a request/response model. |
| 6246 | LDAP, MySQL, Redis and SSL are example of such protocols, though they |
| 6247 | already all have their dedicated checks with a deeper understanding of |
| 6248 | the respective protocols. |
| 6249 | In this mode, many questions may be sent and many answers may be |
| 6250 | analysed. |
| 6251 | |
Baptiste Assmann | d60a9e5 | 2015-04-25 16:27:23 +0200 | [diff] [blame] | 6252 | A fifth mode can be used to insert comments in different steps of the |
| 6253 | script. |
| 6254 | |
| 6255 | For each tcp-check rule you create, you can add a "comment" directive, |
| 6256 | followed by a string. This string will be reported in the log and stderr |
| 6257 | in debug mode. It is useful to make user-friendly error reporting. |
| 6258 | The "comment" is of course optional. |
| 6259 | |
| 6260 | |
Willy Tarreau | ed17985 | 2013-12-16 01:07:00 +0100 | [diff] [blame] | 6261 | Examples : |
| 6262 | # perform a POP check (analyse only server's banner) |
| 6263 | option tcp-check |
Baptiste Assmann | d60a9e5 | 2015-04-25 16:27:23 +0200 | [diff] [blame] | 6264 | tcp-check expect string +OK\ POP3\ ready comment POP\ protocol |
Willy Tarreau | ed17985 | 2013-12-16 01:07:00 +0100 | [diff] [blame] | 6265 | |
| 6266 | # perform an IMAP check (analyse only server's banner) |
| 6267 | option tcp-check |
Baptiste Assmann | d60a9e5 | 2015-04-25 16:27:23 +0200 | [diff] [blame] | 6268 | tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol |
Willy Tarreau | ed17985 | 2013-12-16 01:07:00 +0100 | [diff] [blame] | 6269 | |
| 6270 | # look for the redis master server after ensuring it speaks well |
| 6271 | # redis protocol, then it exits properly. |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 6272 | # (send a command then analyse the response 3 times) |
Willy Tarreau | ed17985 | 2013-12-16 01:07:00 +0100 | [diff] [blame] | 6273 | option tcp-check |
Baptiste Assmann | d60a9e5 | 2015-04-25 16:27:23 +0200 | [diff] [blame] | 6274 | tcp-check comment PING\ phase |
Willy Tarreau | ed17985 | 2013-12-16 01:07:00 +0100 | [diff] [blame] | 6275 | tcp-check send PING\r\n |
Baptiste Assmann | a332299 | 2015-08-04 10:12:18 +0200 | [diff] [blame] | 6276 | tcp-check expect string +PONG |
Baptiste Assmann | d60a9e5 | 2015-04-25 16:27:23 +0200 | [diff] [blame] | 6277 | tcp-check comment role\ check |
Willy Tarreau | ed17985 | 2013-12-16 01:07:00 +0100 | [diff] [blame] | 6278 | tcp-check send info\ replication\r\n |
| 6279 | tcp-check expect string role:master |
Baptiste Assmann | d60a9e5 | 2015-04-25 16:27:23 +0200 | [diff] [blame] | 6280 | tcp-check comment QUIT\ phase |
Willy Tarreau | ed17985 | 2013-12-16 01:07:00 +0100 | [diff] [blame] | 6281 | tcp-check send QUIT\r\n |
| 6282 | tcp-check expect string +OK |
| 6283 | |
| 6284 | forge a HTTP request, then analyse the response |
| 6285 | (send many headers before analyzing) |
| 6286 | option tcp-check |
Baptiste Assmann | d60a9e5 | 2015-04-25 16:27:23 +0200 | [diff] [blame] | 6287 | tcp-check comment forge\ and\ send\ HTTP\ request |
Willy Tarreau | ed17985 | 2013-12-16 01:07:00 +0100 | [diff] [blame] | 6288 | tcp-check send HEAD\ /\ HTTP/1.1\r\n |
| 6289 | tcp-check send Host:\ www.mydomain.com\r\n |
| 6290 | tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n |
| 6291 | tcp-check send \r\n |
Baptiste Assmann | d60a9e5 | 2015-04-25 16:27:23 +0200 | [diff] [blame] | 6292 | tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response |
Willy Tarreau | ed17985 | 2013-12-16 01:07:00 +0100 | [diff] [blame] | 6293 | |
| 6294 | |
| 6295 | See also : "tcp-check expect", "tcp-check send" |
| 6296 | |
| 6297 | |
Willy Tarreau | 9ea05a7 | 2009-06-14 12:07:01 +0200 | [diff] [blame] | 6298 | option tcp-smart-accept |
| 6299 | no option tcp-smart-accept |
| 6300 | Enable or disable the saving of one ACK packet during the accept sequence |
| 6301 | May be used in sections : defaults | frontend | listen | backend |
| 6302 | yes | yes | yes | no |
| 6303 | Arguments : none |
| 6304 | |
| 6305 | When an HTTP connection request comes in, the system acknowledges it on |
| 6306 | behalf of HAProxy, then the client immediately sends its request, and the |
| 6307 | system acknowledges it too while it is notifying HAProxy about the new |
| 6308 | connection. HAProxy then reads the request and responds. This means that we |
| 6309 | have one TCP ACK sent by the system for nothing, because the request could |
| 6310 | very well be acknowledged by HAProxy when it sends its response. |
| 6311 | |
| 6312 | For this reason, in HTTP mode, HAProxy automatically asks the system to avoid |
| 6313 | sending this useless ACK on platforms which support it (currently at least |
| 6314 | Linux). It must not cause any problem, because the system will send it anyway |
| 6315 | after 40 ms if the response takes more time than expected to come. |
| 6316 | |
| 6317 | During complex network debugging sessions, it may be desirable to disable |
| 6318 | this optimization because delayed ACKs can make troubleshooting more complex |
| 6319 | when trying to identify where packets are delayed. It is then possible to |
| 6320 | fall back to normal behaviour by specifying "no option tcp-smart-accept". |
| 6321 | |
| 6322 | It is also possible to force it for non-HTTP proxies by simply specifying |
| 6323 | "option tcp-smart-accept". For instance, it can make sense with some services |
| 6324 | such as SMTP where the server speaks first. |
| 6325 | |
| 6326 | It is recommended to avoid forcing this option in a defaults section. In case |
| 6327 | of doubt, consider setting it back to automatic values by prepending the |
| 6328 | "default" keyword before it, or disabling it using the "no" keyword. |
| 6329 | |
Willy Tarreau | d88edf2 | 2009-06-14 15:48:17 +0200 | [diff] [blame] | 6330 | See also : "option tcp-smart-connect" |
| 6331 | |
| 6332 | |
| 6333 | option tcp-smart-connect |
| 6334 | no option tcp-smart-connect |
| 6335 | Enable or disable the saving of one ACK packet during the connect sequence |
| 6336 | May be used in sections : defaults | frontend | listen | backend |
| 6337 | yes | no | yes | yes |
| 6338 | Arguments : none |
| 6339 | |
| 6340 | On certain systems (at least Linux), HAProxy can ask the kernel not to |
| 6341 | immediately send an empty ACK upon a connection request, but to directly |
| 6342 | send the buffer request instead. This saves one packet on the network and |
| 6343 | thus boosts performance. It can also be useful for some servers, because they |
| 6344 | immediately get the request along with the incoming connection. |
| 6345 | |
| 6346 | This feature is enabled when "option tcp-smart-connect" is set in a backend. |
| 6347 | It is not enabled by default because it makes network troubleshooting more |
| 6348 | complex. |
| 6349 | |
| 6350 | It only makes sense to enable it with protocols where the client speaks first |
| 6351 | such as HTTP. In other situations, if there is no data to send in place of |
| 6352 | the ACK, a normal ACK is sent. |
| 6353 | |
| 6354 | If this option has been enabled in a "defaults" section, it can be disabled |
| 6355 | in a specific instance by prepending the "no" keyword before it. |
| 6356 | |
| 6357 | See also : "option tcp-smart-accept" |
| 6358 | |
Willy Tarreau | 9ea05a7 | 2009-06-14 12:07:01 +0200 | [diff] [blame] | 6359 | |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 6360 | option tcpka |
| 6361 | Enable or disable the sending of TCP keepalive packets on both sides |
| 6362 | May be used in sections : defaults | frontend | listen | backend |
| 6363 | yes | yes | yes | yes |
| 6364 | Arguments : none |
| 6365 | |
| 6366 | When there is a firewall or any session-aware component between a client and |
| 6367 | a server, and when the protocol involves very long sessions with long idle |
| 6368 | periods (eg: remote desktops), there is a risk that one of the intermediate |
| 6369 | components decides to expire a session which has remained idle for too long. |
| 6370 | |
| 6371 | Enabling socket-level TCP keep-alives makes the system regularly send packets |
| 6372 | to the other end of the connection, leaving it active. The delay between |
| 6373 | keep-alive probes is controlled by the system only and depends both on the |
| 6374 | operating system and its tuning parameters. |
| 6375 | |
| 6376 | It is important to understand that keep-alive packets are neither emitted nor |
| 6377 | received at the application level. It is only the network stacks which sees |
| 6378 | them. For this reason, even if one side of the proxy already uses keep-alives |
| 6379 | to maintain its connection alive, those keep-alive packets will not be |
| 6380 | forwarded to the other side of the proxy. |
| 6381 | |
| 6382 | Please note that this has nothing to do with HTTP keep-alive. |
| 6383 | |
| 6384 | Using option "tcpka" enables the emission of TCP keep-alive probes on both |
| 6385 | the client and server sides of a connection. Note that this is meaningful |
| 6386 | only in "defaults" or "listen" sections. If this option is used in a |
| 6387 | frontend, only the client side will get keep-alives, and if this option is |
| 6388 | used in a backend, only the server side will get keep-alives. For this |
| 6389 | reason, it is strongly recommended to explicitly use "option clitcpka" and |
| 6390 | "option srvtcpka" when the configuration is split between frontends and |
| 6391 | backends. |
| 6392 | |
| 6393 | See also : "option clitcpka", "option srvtcpka" |
| 6394 | |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 6395 | |
| 6396 | option tcplog |
| 6397 | Enable advanced logging of TCP connections with session state and timers |
| 6398 | May be used in sections : defaults | frontend | listen | backend |
| 6399 | yes | yes | yes | yes |
| 6400 | Arguments : none |
| 6401 | |
| 6402 | By default, the log output format is very poor, as it only contains the |
| 6403 | source and destination addresses, and the instance name. By specifying |
| 6404 | "option tcplog", each log line turns into a much richer format including, but |
| 6405 | not limited to, the connection timers, the session status, the connections |
| 6406 | numbers, the frontend, backend and server name, and of course the source |
| 6407 | address and ports. This option is useful for pure TCP proxies in order to |
| 6408 | find which of the client or server disconnects or times out. For normal HTTP |
| 6409 | proxies, it's better to use "option httplog" which is even more complete. |
| 6410 | |
| 6411 | This option may be set either in the frontend or the backend. |
| 6412 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 6413 | See also : "option httplog", and section 8 about logging. |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 6414 | |
| 6415 | |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 6416 | option transparent |
| 6417 | no option transparent |
| 6418 | Enable client-side transparent proxying |
| 6419 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | 4b1f859 | 2008-12-23 23:13:55 +0100 | [diff] [blame] | 6420 | yes | no | yes | yes |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 6421 | Arguments : none |
| 6422 | |
| 6423 | This option was introduced in order to provide layer 7 persistence to layer 3 |
| 6424 | load balancers. The idea is to use the OS's ability to redirect an incoming |
| 6425 | connection for a remote address to a local process (here HAProxy), and let |
| 6426 | this process know what address was initially requested. When this option is |
| 6427 | used, sessions without cookies will be forwarded to the original destination |
| 6428 | IP address of the incoming request (which should match that of another |
| 6429 | equipment), while requests with cookies will still be forwarded to the |
| 6430 | appropriate server. |
| 6431 | |
| 6432 | Note that contrary to a common belief, this option does NOT make HAProxy |
| 6433 | present the client's IP to the server when establishing the connection. |
| 6434 | |
Willy Tarreau | a114605 | 2011-03-01 09:51:54 +0100 | [diff] [blame] | 6435 | See also: the "usesrc" argument of the "source" keyword, and the |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 6436 | "transparent" option of the "bind" keyword. |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 6437 | |
Willy Tarreau | bf1f816 | 2007-12-28 17:42:56 +0100 | [diff] [blame] | 6438 | |
Simon Horman | 98637e5 | 2014-06-20 12:30:16 +0900 | [diff] [blame] | 6439 | external-check command <command> |
| 6440 | Executable to run when performing an external-check |
| 6441 | May be used in sections : defaults | frontend | listen | backend |
| 6442 | yes | no | yes | yes |
| 6443 | |
| 6444 | Arguments : |
| 6445 | <command> is the external command to run |
| 6446 | |
Simon Horman | 98637e5 | 2014-06-20 12:30:16 +0900 | [diff] [blame] | 6447 | The arguments passed to the to the command are: |
| 6448 | |
Cyril Bonté | 777be86 | 2014-12-02 21:21:35 +0100 | [diff] [blame] | 6449 | <proxy_address> <proxy_port> <server_address> <server_port> |
Simon Horman | 98637e5 | 2014-06-20 12:30:16 +0900 | [diff] [blame] | 6450 | |
Cyril Bonté | 777be86 | 2014-12-02 21:21:35 +0100 | [diff] [blame] | 6451 | The <proxy_address> and <proxy_port> are derived from the first listener |
| 6452 | that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket |
| 6453 | listener the proxy_address will be the path of the socket and the |
| 6454 | <proxy_port> will be the string "NOT_USED". In a backend section, it's not |
| 6455 | possible to determine a listener, and both <proxy_address> and <proxy_port> |
| 6456 | will have the string value "NOT_USED". |
Simon Horman | 98637e5 | 2014-06-20 12:30:16 +0900 | [diff] [blame] | 6457 | |
Cyril Bonté | 72cda2a | 2014-12-27 22:28:39 +0100 | [diff] [blame] | 6458 | Some values are also provided through environment variables. |
| 6459 | |
| 6460 | Environment variables : |
| 6461 | HAPROXY_PROXY_ADDR The first bind address if available (or empty if not |
| 6462 | applicable, for example in a "backend" section). |
| 6463 | |
| 6464 | HAPROXY_PROXY_ID The backend id. |
| 6465 | |
| 6466 | HAPROXY_PROXY_NAME The backend name. |
| 6467 | |
| 6468 | HAPROXY_PROXY_PORT The first bind port if available (or empty if not |
| 6469 | applicable, for example in a "backend" section or |
| 6470 | for a UNIX socket). |
| 6471 | |
| 6472 | HAPROXY_SERVER_ADDR The server address. |
| 6473 | |
| 6474 | HAPROXY_SERVER_CURCONN The current number of connections on the server. |
| 6475 | |
| 6476 | HAPROXY_SERVER_ID The server id. |
| 6477 | |
| 6478 | HAPROXY_SERVER_MAXCONN The server max connections. |
| 6479 | |
| 6480 | HAPROXY_SERVER_NAME The server name. |
| 6481 | |
| 6482 | HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX |
| 6483 | socket). |
| 6484 | |
| 6485 | PATH The PATH environment variable used when executing |
| 6486 | the command may be set using "external-check path". |
| 6487 | |
Simon Horman | 98637e5 | 2014-06-20 12:30:16 +0900 | [diff] [blame] | 6488 | If the command executed and exits with a zero status then the check is |
| 6489 | considered to have passed, otherwise the check is considered to have |
| 6490 | failed. |
| 6491 | |
| 6492 | Example : |
| 6493 | external-check command /bin/true |
| 6494 | |
| 6495 | See also : "external-check", "option external-check", "external-check path" |
| 6496 | |
| 6497 | |
| 6498 | external-check path <path> |
| 6499 | The value of the PATH environment variable used when running an external-check |
| 6500 | May be used in sections : defaults | frontend | listen | backend |
| 6501 | yes | no | yes | yes |
| 6502 | |
| 6503 | Arguments : |
| 6504 | <path> is the path used when executing external command to run |
| 6505 | |
| 6506 | The default path is "". |
| 6507 | |
| 6508 | Example : |
| 6509 | external-check path "/usr/bin:/bin" |
| 6510 | |
| 6511 | See also : "external-check", "option external-check", |
| 6512 | "external-check command" |
| 6513 | |
| 6514 | |
Emeric Brun | 647caf1 | 2009-06-30 17:57:00 +0200 | [diff] [blame] | 6515 | persist rdp-cookie |
Hervé COMMOWICK | a3eb39c | 2011-08-05 18:48:51 +0200 | [diff] [blame] | 6516 | persist rdp-cookie(<name>) |
Emeric Brun | 647caf1 | 2009-06-30 17:57:00 +0200 | [diff] [blame] | 6517 | Enable RDP cookie-based persistence |
| 6518 | May be used in sections : defaults | frontend | listen | backend |
| 6519 | yes | no | yes | yes |
| 6520 | Arguments : |
| 6521 | <name> is the optional name of the RDP cookie to check. If omitted, the |
Willy Tarreau | 61e28f2 | 2010-05-16 22:31:05 +0200 | [diff] [blame] | 6522 | default cookie name "msts" will be used. There currently is no |
| 6523 | valid reason to change this name. |
Emeric Brun | 647caf1 | 2009-06-30 17:57:00 +0200 | [diff] [blame] | 6524 | |
| 6525 | This statement enables persistence based on an RDP cookie. The RDP cookie |
| 6526 | contains all information required to find the server in the list of known |
| 6527 | servers. So when this option is set in the backend, the request is analysed |
| 6528 | and if an RDP cookie is found, it is decoded. If it matches a known server |
| 6529 | which is still UP (or if "option persist" is set), then the connection is |
| 6530 | forwarded to this server. |
| 6531 | |
| 6532 | Note that this only makes sense in a TCP backend, but for this to work, the |
| 6533 | frontend must have waited long enough to ensure that an RDP cookie is present |
| 6534 | in the request buffer. This is the same requirement as with the "rdp-cookie" |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 6535 | load-balancing method. Thus it is highly recommended to put all statements in |
Emeric Brun | 647caf1 | 2009-06-30 17:57:00 +0200 | [diff] [blame] | 6536 | a single "listen" section. |
| 6537 | |
Willy Tarreau | 61e28f2 | 2010-05-16 22:31:05 +0200 | [diff] [blame] | 6538 | Also, it is important to understand that the terminal server will emit this |
| 6539 | RDP cookie only if it is configured for "token redirection mode", which means |
| 6540 | that the "IP address redirection" option is disabled. |
| 6541 | |
Emeric Brun | 647caf1 | 2009-06-30 17:57:00 +0200 | [diff] [blame] | 6542 | Example : |
| 6543 | listen tse-farm |
| 6544 | bind :3389 |
| 6545 | # wait up to 5s for an RDP cookie in the request |
| 6546 | tcp-request inspect-delay 5s |
| 6547 | tcp-request content accept if RDP_COOKIE |
| 6548 | # apply RDP cookie persistence |
| 6549 | persist rdp-cookie |
| 6550 | # if server is unknown, let's balance on the same cookie. |
Cyril Bonté | dc4d903 | 2012-04-08 21:57:39 +0200 | [diff] [blame] | 6551 | # alternatively, "balance leastconn" may be useful too. |
Emeric Brun | 647caf1 | 2009-06-30 17:57:00 +0200 | [diff] [blame] | 6552 | balance rdp-cookie |
| 6553 | server srv1 1.1.1.1:3389 |
| 6554 | server srv2 1.1.1.2:3389 |
| 6555 | |
Simon Horman | ab814e0 | 2011-06-24 14:50:20 +0900 | [diff] [blame] | 6556 | See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and |
| 6557 | the rdp_cookie pattern fetch function. |
Emeric Brun | 647caf1 | 2009-06-30 17:57:00 +0200 | [diff] [blame] | 6558 | |
| 6559 | |
Willy Tarreau | 3a7d207 | 2009-03-05 23:48:25 +0100 | [diff] [blame] | 6560 | rate-limit sessions <rate> |
| 6561 | Set a limit on the number of new sessions accepted per second on a frontend |
| 6562 | May be used in sections : defaults | frontend | listen | backend |
| 6563 | yes | yes | yes | no |
| 6564 | Arguments : |
| 6565 | <rate> The <rate> parameter is an integer designating the maximum number |
| 6566 | of new sessions per second to accept on the frontend. |
| 6567 | |
| 6568 | When the frontend reaches the specified number of new sessions per second, it |
| 6569 | stops accepting new connections until the rate drops below the limit again. |
| 6570 | During this time, the pending sessions will be kept in the socket's backlog |
| 6571 | (in system buffers) and haproxy will not even be aware that sessions are |
| 6572 | pending. When applying very low limit on a highly loaded service, it may make |
| 6573 | sense to increase the socket's backlog using the "backlog" keyword. |
| 6574 | |
| 6575 | This feature is particularly efficient at blocking connection-based attacks |
| 6576 | or service abuse on fragile servers. Since the session rate is measured every |
| 6577 | millisecond, it is extremely accurate. Also, the limit applies immediately, |
| 6578 | no delay is needed at all to detect the threshold. |
| 6579 | |
| 6580 | Example : limit the connection rate on SMTP to 10 per second max |
| 6581 | listen smtp |
| 6582 | mode tcp |
| 6583 | bind :25 |
| 6584 | rate-limit sessions 10 |
Panagiotis Panagiotopoulos | 7282d8e | 2016-02-11 16:37:15 +0200 | [diff] [blame] | 6585 | server smtp1 127.0.0.1:1025 |
Willy Tarreau | 3a7d207 | 2009-03-05 23:48:25 +0100 | [diff] [blame] | 6586 | |
Willy Tarreau | a17c2d9 | 2011-07-25 08:16:20 +0200 | [diff] [blame] | 6587 | Note : when the maximum rate is reached, the frontend's status is not changed |
| 6588 | but its sockets appear as "WAITING" in the statistics if the |
| 6589 | "socket-stats" option is enabled. |
Willy Tarreau | 3a7d207 | 2009-03-05 23:48:25 +0100 | [diff] [blame] | 6590 | |
| 6591 | See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion. |
| 6592 | |
| 6593 | |
Willy Tarreau | 2e1dca8 | 2012-09-12 08:43:15 +0200 | [diff] [blame] | 6594 | redirect location <loc> [code <code>] <option> [{if | unless} <condition>] |
| 6595 | redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>] |
| 6596 | redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>] |
Willy Tarreau | b463dfb | 2008-06-07 23:08:56 +0200 | [diff] [blame] | 6597 | Return an HTTP redirection if/unless a condition is matched |
| 6598 | May be used in sections : defaults | frontend | listen | backend |
| 6599 | no | yes | yes | yes |
| 6600 | |
| 6601 | If/unless the condition is matched, the HTTP request will lead to a redirect |
Willy Tarreau | f285f54 | 2010-01-03 20:03:03 +0100 | [diff] [blame] | 6602 | response. If no condition is specified, the redirect applies unconditionally. |
Willy Tarreau | b463dfb | 2008-06-07 23:08:56 +0200 | [diff] [blame] | 6603 | |
Willy Tarreau | 0140f25 | 2008-11-19 21:07:09 +0100 | [diff] [blame] | 6604 | Arguments : |
Willy Tarreau | 2e1dca8 | 2012-09-12 08:43:15 +0200 | [diff] [blame] | 6605 | <loc> With "redirect location", the exact value in <loc> is placed into |
Thierry FOURNIER | d18cd0f | 2013-11-29 12:15:45 +0100 | [diff] [blame] | 6606 | the HTTP "Location" header. When used in an "http-request" rule, |
| 6607 | <loc> value follows the log-format rules and can include some |
| 6608 | dynamic values (see Custom Log Format in section 8.2.4). |
Willy Tarreau | 2e1dca8 | 2012-09-12 08:43:15 +0200 | [diff] [blame] | 6609 | |
| 6610 | <pfx> With "redirect prefix", the "Location" header is built from the |
| 6611 | concatenation of <pfx> and the complete URI path, including the |
| 6612 | query string, unless the "drop-query" option is specified (see |
| 6613 | below). As a special case, if <pfx> equals exactly "/", then |
| 6614 | nothing is inserted before the original URI. It allows one to |
Thierry FOURNIER | d18cd0f | 2013-11-29 12:15:45 +0100 | [diff] [blame] | 6615 | redirect to the same URL (for instance, to insert a cookie). When |
| 6616 | used in an "http-request" rule, <pfx> value follows the log-format |
| 6617 | rules and can include some dynamic values (see Custom Log Format |
| 6618 | in section 8.2.4). |
Willy Tarreau | 2e1dca8 | 2012-09-12 08:43:15 +0200 | [diff] [blame] | 6619 | |
| 6620 | <sch> With "redirect scheme", then the "Location" header is built by |
| 6621 | concatenating <sch> with "://" then the first occurrence of the |
| 6622 | "Host" header, and then the URI path, including the query string |
| 6623 | unless the "drop-query" option is specified (see below). If no |
| 6624 | path is found or if the path is "*", then "/" is used instead. If |
| 6625 | no "Host" header is found, then an empty host component will be |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 6626 | returned, which most recent browsers interpret as redirecting to |
Willy Tarreau | 2e1dca8 | 2012-09-12 08:43:15 +0200 | [diff] [blame] | 6627 | the same host. This directive is mostly used to redirect HTTP to |
Thierry FOURNIER | d18cd0f | 2013-11-29 12:15:45 +0100 | [diff] [blame] | 6628 | HTTPS. When used in an "http-request" rule, <sch> value follows |
| 6629 | the log-format rules and can include some dynamic values (see |
| 6630 | Custom Log Format in section 8.2.4). |
Willy Tarreau | 0140f25 | 2008-11-19 21:07:09 +0100 | [diff] [blame] | 6631 | |
| 6632 | <code> The code is optional. It indicates which type of HTTP redirection |
Willy Tarreau | b67fdc4 | 2013-03-29 19:28:11 +0100 | [diff] [blame] | 6633 | is desired. Only codes 301, 302, 303, 307 and 308 are supported, |
| 6634 | with 302 used by default if no code is specified. 301 means |
| 6635 | "Moved permanently", and a browser may cache the Location. 302 |
Baptiste Assmann | ea849c0 | 2015-08-03 11:42:50 +0200 | [diff] [blame] | 6636 | means "Moved temporarily" and means that the browser should not |
Willy Tarreau | b67fdc4 | 2013-03-29 19:28:11 +0100 | [diff] [blame] | 6637 | cache the redirection. 303 is equivalent to 302 except that the |
| 6638 | browser will fetch the location with a GET method. 307 is just |
| 6639 | like 302 but makes it clear that the same method must be reused. |
| 6640 | Likewise, 308 replaces 301 if the same method must be used. |
Willy Tarreau | 0140f25 | 2008-11-19 21:07:09 +0100 | [diff] [blame] | 6641 | |
| 6642 | <option> There are several options which can be specified to adjust the |
| 6643 | expected behaviour of a redirection : |
| 6644 | |
| 6645 | - "drop-query" |
| 6646 | When this keyword is used in a prefix-based redirection, then the |
| 6647 | location will be set without any possible query-string, which is useful |
| 6648 | for directing users to a non-secure page for instance. It has no effect |
| 6649 | with a location-type redirect. |
| 6650 | |
Willy Tarreau | 81e3b4f | 2010-01-10 00:42:19 +0100 | [diff] [blame] | 6651 | - "append-slash" |
| 6652 | This keyword may be used in conjunction with "drop-query" to redirect |
| 6653 | users who use a URL not ending with a '/' to the same one with the '/'. |
| 6654 | It can be useful to ensure that search engines will only see one URL. |
| 6655 | For this, a return code 301 is preferred. |
| 6656 | |
Willy Tarreau | 0140f25 | 2008-11-19 21:07:09 +0100 | [diff] [blame] | 6657 | - "set-cookie NAME[=value]" |
| 6658 | A "Set-Cookie" header will be added with NAME (and optionally "=value") |
| 6659 | to the response. This is sometimes used to indicate that a user has |
| 6660 | been seen, for instance to protect against some types of DoS. No other |
| 6661 | cookie option is added, so the cookie will be a session cookie. Note |
| 6662 | that for a browser, a sole cookie name without an equal sign is |
| 6663 | different from a cookie with an equal sign. |
| 6664 | |
| 6665 | - "clear-cookie NAME[=]" |
| 6666 | A "Set-Cookie" header will be added with NAME (and optionally "="), but |
| 6667 | with the "Max-Age" attribute set to zero. This will tell the browser to |
| 6668 | delete this cookie. It is useful for instance on logout pages. It is |
| 6669 | important to note that clearing the cookie "NAME" will not remove a |
| 6670 | cookie set with "NAME=value". You have to clear the cookie "NAME=" for |
| 6671 | that, because the browser makes the difference. |
Willy Tarreau | b463dfb | 2008-06-07 23:08:56 +0200 | [diff] [blame] | 6672 | |
| 6673 | Example: move the login URL only to HTTPS. |
| 6674 | acl clear dst_port 80 |
| 6675 | acl secure dst_port 8080 |
| 6676 | acl login_page url_beg /login |
Willy Tarreau | 0140f25 | 2008-11-19 21:07:09 +0100 | [diff] [blame] | 6677 | acl logout url_beg /logout |
Willy Tarreau | 79da469 | 2008-11-19 20:03:04 +0100 | [diff] [blame] | 6678 | acl uid_given url_reg /login?userid=[^&]+ |
Willy Tarreau | 0140f25 | 2008-11-19 21:07:09 +0100 | [diff] [blame] | 6679 | acl cookie_set hdr_sub(cookie) SEEN=1 |
| 6680 | |
| 6681 | redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set |
Willy Tarreau | 79da469 | 2008-11-19 20:03:04 +0100 | [diff] [blame] | 6682 | redirect prefix https://mysite.com if login_page !secure |
| 6683 | redirect prefix http://mysite.com drop-query if login_page !uid_given |
| 6684 | redirect location http://mysite.com/ if !login_page secure |
Willy Tarreau | 0140f25 | 2008-11-19 21:07:09 +0100 | [diff] [blame] | 6685 | redirect location / clear-cookie USERID= if logout |
Willy Tarreau | b463dfb | 2008-06-07 23:08:56 +0200 | [diff] [blame] | 6686 | |
Willy Tarreau | 81e3b4f | 2010-01-10 00:42:19 +0100 | [diff] [blame] | 6687 | Example: send redirects for request for articles without a '/'. |
| 6688 | acl missing_slash path_reg ^/article/[^/]*$ |
| 6689 | redirect code 301 prefix / drop-query append-slash if missing_slash |
| 6690 | |
Willy Tarreau | 2e1dca8 | 2012-09-12 08:43:15 +0200 | [diff] [blame] | 6691 | Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy. |
David BERARD | e715304 | 2012-11-03 00:11:31 +0100 | [diff] [blame] | 6692 | redirect scheme https if !{ ssl_fc } |
Willy Tarreau | 2e1dca8 | 2012-09-12 08:43:15 +0200 | [diff] [blame] | 6693 | |
Thierry FOURNIER | d18cd0f | 2013-11-29 12:15:45 +0100 | [diff] [blame] | 6694 | Example: append 'www.' prefix in front of all hosts not having it |
Coen Rosdorff | 596659b | 2016-04-11 11:33:49 +0200 | [diff] [blame] | 6695 | http-request redirect code 301 location \ |
| 6696 | http://www.%[hdr(host)]%[capture.req.uri] \ |
| 6697 | unless { hdr_beg(host) -i www } |
Thierry FOURNIER | d18cd0f | 2013-11-29 12:15:45 +0100 | [diff] [blame] | 6698 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 6699 | See section 7 about ACL usage. |
Willy Tarreau | b463dfb | 2008-06-07 23:08:56 +0200 | [diff] [blame] | 6700 | |
| 6701 | |
Krzysztof Piotr Oledzki | 25b501a | 2008-01-06 16:36:16 +0100 | [diff] [blame] | 6702 | redisp (deprecated) |
| 6703 | redispatch (deprecated) |
| 6704 | Enable or disable session redistribution in case of connection failure |
| 6705 | May be used in sections: defaults | frontend | listen | backend |
| 6706 | yes | no | yes | yes |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 6707 | Arguments : none |
Krzysztof Piotr Oledzki | 25b501a | 2008-01-06 16:36:16 +0100 | [diff] [blame] | 6708 | |
| 6709 | In HTTP mode, if a server designated by a cookie is down, clients may |
| 6710 | definitely stick to it because they cannot flush the cookie, so they will not |
| 6711 | be able to access the service anymore. |
| 6712 | |
| 6713 | Specifying "redispatch" will allow the proxy to break their persistence and |
| 6714 | redistribute them to a working server. |
| 6715 | |
| 6716 | It also allows to retry last connection to another server in case of multiple |
| 6717 | connection failures. Of course, it requires having "retries" set to a nonzero |
| 6718 | value. |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 6719 | |
Krzysztof Piotr Oledzki | 25b501a | 2008-01-06 16:36:16 +0100 | [diff] [blame] | 6720 | This form is deprecated, do not use it in any new configuration, use the new |
| 6721 | "option redispatch" instead. |
| 6722 | |
| 6723 | See also : "option redispatch" |
| 6724 | |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 6725 | |
Willy Tarreau | 8abd4cd | 2010-01-31 14:30:44 +0100 | [diff] [blame] | 6726 | reqadd <string> [{if | unless} <cond>] |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6727 | Add a header at the end of the HTTP request |
| 6728 | May be used in sections : defaults | frontend | listen | backend |
| 6729 | no | yes | yes | yes |
| 6730 | Arguments : |
| 6731 | <string> is the complete line to be added. Any space or known delimiter |
| 6732 | must be escaped using a backslash ('\'). Please refer to section |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 6733 | 6 about HTTP header manipulation for more information. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6734 | |
Willy Tarreau | 8abd4cd | 2010-01-31 14:30:44 +0100 | [diff] [blame] | 6735 | <cond> is an optional matching condition built from ACLs. It makes it |
| 6736 | possible to ignore this rule when other conditions are not met. |
| 6737 | |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6738 | A new line consisting in <string> followed by a line feed will be added after |
| 6739 | the last header of an HTTP request. |
| 6740 | |
| 6741 | Header transformations only apply to traffic which passes through HAProxy, |
| 6742 | and not to traffic generated by HAProxy, such as health-checks or error |
| 6743 | responses. |
| 6744 | |
Willy Tarreau | 8abd4cd | 2010-01-31 14:30:44 +0100 | [diff] [blame] | 6745 | Example : add "X-Proto: SSL" to requests coming via port 81 |
| 6746 | acl is-ssl dst_port 81 |
| 6747 | reqadd X-Proto:\ SSL if is-ssl |
| 6748 | |
Ruoshan Huang | eb5a363 | 2015-12-08 21:00:23 +0800 | [diff] [blame] | 6749 | See also: "rspadd", "http-request", section 6 about HTTP header manipulation, |
| 6750 | and section 7 about ACLs. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6751 | |
| 6752 | |
Willy Tarreau | 5321c42 | 2010-01-28 20:35:13 +0100 | [diff] [blame] | 6753 | reqallow <search> [{if | unless} <cond>] |
| 6754 | reqiallow <search> [{if | unless} <cond>] (ignore case) |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6755 | Definitely allow an HTTP request if a line matches a regular expression |
| 6756 | May be used in sections : defaults | frontend | listen | backend |
| 6757 | no | yes | yes | yes |
| 6758 | Arguments : |
| 6759 | <search> is the regular expression applied to HTTP headers and to the |
| 6760 | request line. This is an extended regular expression. Parenthesis |
| 6761 | grouping is supported and no preliminary backslash is required. |
| 6762 | Any space or known delimiter must be escaped using a backslash |
| 6763 | ('\'). The pattern applies to a full line at a time. The |
| 6764 | "reqallow" keyword strictly matches case while "reqiallow" |
| 6765 | ignores case. |
| 6766 | |
Willy Tarreau | 5321c42 | 2010-01-28 20:35:13 +0100 | [diff] [blame] | 6767 | <cond> is an optional matching condition built from ACLs. It makes it |
| 6768 | possible to ignore this rule when other conditions are not met. |
| 6769 | |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6770 | A request containing any line which matches extended regular expression |
| 6771 | <search> will mark the request as allowed, even if any later test would |
| 6772 | result in a deny. The test applies both to the request line and to request |
| 6773 | headers. Keep in mind that URLs in request line are case-sensitive while |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 6774 | header names are not. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6775 | |
| 6776 | It is easier, faster and more powerful to use ACLs to write access policies. |
| 6777 | Reqdeny, reqallow and reqpass should be avoided in new designs. |
| 6778 | |
| 6779 | Example : |
| 6780 | # allow www.* but refuse *.local |
| 6781 | reqiallow ^Host:\ www\. |
| 6782 | reqideny ^Host:\ .*\.local |
| 6783 | |
Ruoshan Huang | eb5a363 | 2015-12-08 21:00:23 +0800 | [diff] [blame] | 6784 | See also: "reqdeny", "block", "http-request", section 6 about HTTP header |
| 6785 | manipulation, and section 7 about ACLs. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6786 | |
| 6787 | |
Willy Tarreau | 5321c42 | 2010-01-28 20:35:13 +0100 | [diff] [blame] | 6788 | reqdel <search> [{if | unless} <cond>] |
| 6789 | reqidel <search> [{if | unless} <cond>] (ignore case) |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6790 | Delete all headers matching a regular expression in an HTTP request |
| 6791 | May be used in sections : defaults | frontend | listen | backend |
| 6792 | no | yes | yes | yes |
| 6793 | Arguments : |
| 6794 | <search> is the regular expression applied to HTTP headers and to the |
| 6795 | request line. This is an extended regular expression. Parenthesis |
| 6796 | grouping is supported and no preliminary backslash is required. |
| 6797 | Any space or known delimiter must be escaped using a backslash |
| 6798 | ('\'). The pattern applies to a full line at a time. The "reqdel" |
| 6799 | keyword strictly matches case while "reqidel" ignores case. |
| 6800 | |
Willy Tarreau | 5321c42 | 2010-01-28 20:35:13 +0100 | [diff] [blame] | 6801 | <cond> is an optional matching condition built from ACLs. It makes it |
| 6802 | possible to ignore this rule when other conditions are not met. |
| 6803 | |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6804 | Any header line matching extended regular expression <search> in the request |
| 6805 | will be completely deleted. Most common use of this is to remove unwanted |
| 6806 | and/or dangerous headers or cookies from a request before passing it to the |
| 6807 | next servers. |
| 6808 | |
| 6809 | Header transformations only apply to traffic which passes through HAProxy, |
| 6810 | and not to traffic generated by HAProxy, such as health-checks or error |
| 6811 | responses. Keep in mind that header names are not case-sensitive. |
| 6812 | |
| 6813 | Example : |
| 6814 | # remove X-Forwarded-For header and SERVER cookie |
| 6815 | reqidel ^X-Forwarded-For:.* |
| 6816 | reqidel ^Cookie:.*SERVER= |
| 6817 | |
Ruoshan Huang | eb5a363 | 2015-12-08 21:00:23 +0800 | [diff] [blame] | 6818 | See also: "reqadd", "reqrep", "rspdel", "http-request", section 6 about |
| 6819 | HTTP header manipulation, and section 7 about ACLs. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6820 | |
| 6821 | |
Willy Tarreau | 5321c42 | 2010-01-28 20:35:13 +0100 | [diff] [blame] | 6822 | reqdeny <search> [{if | unless} <cond>] |
| 6823 | reqideny <search> [{if | unless} <cond>] (ignore case) |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6824 | Deny an HTTP request if a line matches a regular expression |
| 6825 | May be used in sections : defaults | frontend | listen | backend |
| 6826 | no | yes | yes | yes |
| 6827 | Arguments : |
| 6828 | <search> is the regular expression applied to HTTP headers and to the |
| 6829 | request line. This is an extended regular expression. Parenthesis |
| 6830 | grouping is supported and no preliminary backslash is required. |
| 6831 | Any space or known delimiter must be escaped using a backslash |
| 6832 | ('\'). The pattern applies to a full line at a time. The |
| 6833 | "reqdeny" keyword strictly matches case while "reqideny" ignores |
| 6834 | case. |
| 6835 | |
Willy Tarreau | 5321c42 | 2010-01-28 20:35:13 +0100 | [diff] [blame] | 6836 | <cond> is an optional matching condition built from ACLs. It makes it |
| 6837 | possible to ignore this rule when other conditions are not met. |
| 6838 | |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6839 | A request containing any line which matches extended regular expression |
| 6840 | <search> will mark the request as denied, even if any later test would |
| 6841 | result in an allow. The test applies both to the request line and to request |
| 6842 | headers. Keep in mind that URLs in request line are case-sensitive while |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 6843 | header names are not. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6844 | |
Willy Tarreau | ced2701 | 2008-01-17 20:35:34 +0100 | [diff] [blame] | 6845 | A denied request will generate an "HTTP 403 forbidden" response once the |
Willy Tarreau | d2a4aa2 | 2008-01-31 15:28:22 +0100 | [diff] [blame] | 6846 | complete request has been parsed. This is consistent with what is practiced |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 6847 | using ACLs. |
Willy Tarreau | ced2701 | 2008-01-17 20:35:34 +0100 | [diff] [blame] | 6848 | |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6849 | It is easier, faster and more powerful to use ACLs to write access policies. |
| 6850 | Reqdeny, reqallow and reqpass should be avoided in new designs. |
| 6851 | |
| 6852 | Example : |
| 6853 | # refuse *.local, then allow www.* |
| 6854 | reqideny ^Host:\ .*\.local |
| 6855 | reqiallow ^Host:\ www\. |
| 6856 | |
Ruoshan Huang | eb5a363 | 2015-12-08 21:00:23 +0800 | [diff] [blame] | 6857 | See also: "reqallow", "rspdeny", "block", "http-request", section 6 about |
| 6858 | HTTP header manipulation, and section 7 about ACLs. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6859 | |
| 6860 | |
Willy Tarreau | 5321c42 | 2010-01-28 20:35:13 +0100 | [diff] [blame] | 6861 | reqpass <search> [{if | unless} <cond>] |
| 6862 | reqipass <search> [{if | unless} <cond>] (ignore case) |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6863 | Ignore any HTTP request line matching a regular expression in next rules |
| 6864 | May be used in sections : defaults | frontend | listen | backend |
| 6865 | no | yes | yes | yes |
| 6866 | Arguments : |
| 6867 | <search> is the regular expression applied to HTTP headers and to the |
| 6868 | request line. This is an extended regular expression. Parenthesis |
| 6869 | grouping is supported and no preliminary backslash is required. |
| 6870 | Any space or known delimiter must be escaped using a backslash |
| 6871 | ('\'). The pattern applies to a full line at a time. The |
| 6872 | "reqpass" keyword strictly matches case while "reqipass" ignores |
| 6873 | case. |
| 6874 | |
Willy Tarreau | 5321c42 | 2010-01-28 20:35:13 +0100 | [diff] [blame] | 6875 | <cond> is an optional matching condition built from ACLs. It makes it |
| 6876 | possible to ignore this rule when other conditions are not met. |
| 6877 | |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6878 | A request containing any line which matches extended regular expression |
| 6879 | <search> will skip next rules, without assigning any deny or allow verdict. |
| 6880 | The test applies both to the request line and to request headers. Keep in |
| 6881 | mind that URLs in request line are case-sensitive while header names are not. |
| 6882 | |
| 6883 | It is easier, faster and more powerful to use ACLs to write access policies. |
| 6884 | Reqdeny, reqallow and reqpass should be avoided in new designs. |
| 6885 | |
| 6886 | Example : |
| 6887 | # refuse *.local, then allow www.*, but ignore "www.private.local" |
| 6888 | reqipass ^Host:\ www.private\.local |
| 6889 | reqideny ^Host:\ .*\.local |
| 6890 | reqiallow ^Host:\ www\. |
| 6891 | |
Ruoshan Huang | eb5a363 | 2015-12-08 21:00:23 +0800 | [diff] [blame] | 6892 | See also: "reqallow", "reqdeny", "block", "http-request", section 6 about |
| 6893 | HTTP header manipulation, and section 7 about ACLs. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6894 | |
| 6895 | |
Willy Tarreau | 5321c42 | 2010-01-28 20:35:13 +0100 | [diff] [blame] | 6896 | reqrep <search> <string> [{if | unless} <cond>] |
| 6897 | reqirep <search> <string> [{if | unless} <cond>] (ignore case) |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6898 | Replace a regular expression with a string in an HTTP request line |
| 6899 | May be used in sections : defaults | frontend | listen | backend |
| 6900 | no | yes | yes | yes |
| 6901 | Arguments : |
| 6902 | <search> is the regular expression applied to HTTP headers and to the |
| 6903 | request line. This is an extended regular expression. Parenthesis |
| 6904 | grouping is supported and no preliminary backslash is required. |
| 6905 | Any space or known delimiter must be escaped using a backslash |
| 6906 | ('\'). The pattern applies to a full line at a time. The "reqrep" |
| 6907 | keyword strictly matches case while "reqirep" ignores case. |
| 6908 | |
| 6909 | <string> is the complete line to be added. Any space or known delimiter |
| 6910 | must be escaped using a backslash ('\'). References to matched |
| 6911 | pattern groups are possible using the common \N form, with N |
| 6912 | being a single digit between 0 and 9. Please refer to section |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 6913 | 6 about HTTP header manipulation for more information. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6914 | |
Willy Tarreau | 5321c42 | 2010-01-28 20:35:13 +0100 | [diff] [blame] | 6915 | <cond> is an optional matching condition built from ACLs. It makes it |
| 6916 | possible to ignore this rule when other conditions are not met. |
| 6917 | |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6918 | Any line matching extended regular expression <search> in the request (both |
| 6919 | the request line and header lines) will be completely replaced with <string>. |
| 6920 | Most common use of this is to rewrite URLs or domain names in "Host" headers. |
| 6921 | |
| 6922 | Header transformations only apply to traffic which passes through HAProxy, |
| 6923 | and not to traffic generated by HAProxy, such as health-checks or error |
| 6924 | responses. Note that for increased readability, it is suggested to add enough |
| 6925 | spaces between the request and the response. Keep in mind that URLs in |
| 6926 | request line are case-sensitive while header names are not. |
| 6927 | |
| 6928 | Example : |
| 6929 | # replace "/static/" with "/" at the beginning of any request path. |
Dmitry Sivachenko | 7823de3 | 2012-05-16 14:00:26 +0400 | [diff] [blame] | 6930 | reqrep ^([^\ :]*)\ /static/(.*) \1\ /\2 |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6931 | # replace "www.mydomain.com" with "www" in the host name. |
| 6932 | reqirep ^Host:\ www.mydomain.com Host:\ www |
| 6933 | |
Ruoshan Huang | eb5a363 | 2015-12-08 21:00:23 +0800 | [diff] [blame] | 6934 | See also: "reqadd", "reqdel", "rsprep", "tune.bufsize", "http-request", |
| 6935 | section 6 about HTTP header manipulation, and section 7 about ACLs. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6936 | |
| 6937 | |
Willy Tarreau | 5321c42 | 2010-01-28 20:35:13 +0100 | [diff] [blame] | 6938 | reqtarpit <search> [{if | unless} <cond>] |
| 6939 | reqitarpit <search> [{if | unless} <cond>] (ignore case) |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6940 | Tarpit an HTTP request containing a line matching a regular expression |
| 6941 | May be used in sections : defaults | frontend | listen | backend |
| 6942 | no | yes | yes | yes |
| 6943 | Arguments : |
| 6944 | <search> is the regular expression applied to HTTP headers and to the |
| 6945 | request line. This is an extended regular expression. Parenthesis |
| 6946 | grouping is supported and no preliminary backslash is required. |
| 6947 | Any space or known delimiter must be escaped using a backslash |
| 6948 | ('\'). The pattern applies to a full line at a time. The |
| 6949 | "reqtarpit" keyword strictly matches case while "reqitarpit" |
| 6950 | ignores case. |
| 6951 | |
Willy Tarreau | 5321c42 | 2010-01-28 20:35:13 +0100 | [diff] [blame] | 6952 | <cond> is an optional matching condition built from ACLs. It makes it |
| 6953 | possible to ignore this rule when other conditions are not met. |
| 6954 | |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6955 | A request containing any line which matches extended regular expression |
| 6956 | <search> will be tarpitted, which means that it will connect to nowhere, will |
Willy Tarreau | ced2701 | 2008-01-17 20:35:34 +0100 | [diff] [blame] | 6957 | be kept open for a pre-defined time, then will return an HTTP error 500 so |
| 6958 | that the attacker does not suspect it has been tarpitted. The status 500 will |
| 6959 | be reported in the logs, but the completion flags will indicate "PT". The |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6960 | delay is defined by "timeout tarpit", or "timeout connect" if the former is |
| 6961 | not set. |
| 6962 | |
| 6963 | The goal of the tarpit is to slow down robots attacking servers with |
| 6964 | identifiable requests. Many robots limit their outgoing number of connections |
| 6965 | and stay connected waiting for a reply which can take several minutes to |
| 6966 | come. Depending on the environment and attack, it may be particularly |
| 6967 | efficient at reducing the load on the network and firewalls. |
| 6968 | |
Willy Tarreau | 5321c42 | 2010-01-28 20:35:13 +0100 | [diff] [blame] | 6969 | Examples : |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6970 | # ignore user-agents reporting any flavour of "Mozilla" or "MSIE", but |
| 6971 | # block all others. |
| 6972 | reqipass ^User-Agent:\.*(Mozilla|MSIE) |
| 6973 | reqitarpit ^User-Agent: |
| 6974 | |
Willy Tarreau | 5321c42 | 2010-01-28 20:35:13 +0100 | [diff] [blame] | 6975 | # block bad guys |
| 6976 | acl badguys src 10.1.0.3 172.16.13.20/28 |
| 6977 | reqitarpit . if badguys |
| 6978 | |
Ruoshan Huang | eb5a363 | 2015-12-08 21:00:23 +0800 | [diff] [blame] | 6979 | See also: "reqallow", "reqdeny", "reqpass", "http-request", section 6 |
| 6980 | about HTTP header manipulation, and section 7 about ACLs. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 6981 | |
| 6982 | |
Willy Tarreau | e5c5ce9 | 2008-06-20 17:27:19 +0200 | [diff] [blame] | 6983 | retries <value> |
| 6984 | Set the number of retries to perform on a server after a connection failure |
| 6985 | May be used in sections: defaults | frontend | listen | backend |
| 6986 | yes | no | yes | yes |
| 6987 | Arguments : |
| 6988 | <value> is the number of times a connection attempt should be retried on |
| 6989 | a server when a connection either is refused or times out. The |
| 6990 | default value is 3. |
| 6991 | |
| 6992 | It is important to understand that this value applies to the number of |
| 6993 | connection attempts, not full requests. When a connection has effectively |
| 6994 | been established to a server, there will be no more retry. |
| 6995 | |
| 6996 | In order to avoid immediate reconnections to a server which is restarting, |
Joseph Lynch | 726ab71 | 2015-05-11 23:25:34 -0700 | [diff] [blame] | 6997 | a turn-around timer of min("timeout connect", one second) is applied before |
| 6998 | a retry occurs. |
Willy Tarreau | e5c5ce9 | 2008-06-20 17:27:19 +0200 | [diff] [blame] | 6999 | |
| 7000 | When "option redispatch" is set, the last retry may be performed on another |
| 7001 | server even if a cookie references a different server. |
| 7002 | |
| 7003 | See also : "option redispatch" |
| 7004 | |
| 7005 | |
Willy Tarreau | fdb563c | 2010-01-31 15:43:27 +0100 | [diff] [blame] | 7006 | rspadd <string> [{if | unless} <cond>] |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7007 | Add a header at the end of the HTTP response |
| 7008 | May be used in sections : defaults | frontend | listen | backend |
| 7009 | no | yes | yes | yes |
| 7010 | Arguments : |
| 7011 | <string> is the complete line to be added. Any space or known delimiter |
| 7012 | must be escaped using a backslash ('\'). Please refer to section |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 7013 | 6 about HTTP header manipulation for more information. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7014 | |
Willy Tarreau | fdb563c | 2010-01-31 15:43:27 +0100 | [diff] [blame] | 7015 | <cond> is an optional matching condition built from ACLs. It makes it |
| 7016 | possible to ignore this rule when other conditions are not met. |
| 7017 | |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7018 | A new line consisting in <string> followed by a line feed will be added after |
| 7019 | the last header of an HTTP response. |
| 7020 | |
| 7021 | Header transformations only apply to traffic which passes through HAProxy, |
| 7022 | and not to traffic generated by HAProxy, such as health-checks or error |
| 7023 | responses. |
| 7024 | |
Ruoshan Huang | eb5a363 | 2015-12-08 21:00:23 +0800 | [diff] [blame] | 7025 | See also: "rspdel" "reqadd", "http-response", section 6 about HTTP header |
| 7026 | manipulation, and section 7 about ACLs. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7027 | |
| 7028 | |
Willy Tarreau | fdb563c | 2010-01-31 15:43:27 +0100 | [diff] [blame] | 7029 | rspdel <search> [{if | unless} <cond>] |
| 7030 | rspidel <search> [{if | unless} <cond>] (ignore case) |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7031 | Delete all headers matching a regular expression in an HTTP response |
| 7032 | May be used in sections : defaults | frontend | listen | backend |
| 7033 | no | yes | yes | yes |
| 7034 | Arguments : |
| 7035 | <search> is the regular expression applied to HTTP headers and to the |
| 7036 | response line. This is an extended regular expression, so |
| 7037 | parenthesis grouping is supported and no preliminary backslash |
| 7038 | is required. Any space or known delimiter must be escaped using |
| 7039 | a backslash ('\'). The pattern applies to a full line at a time. |
| 7040 | The "rspdel" keyword strictly matches case while "rspidel" |
| 7041 | ignores case. |
| 7042 | |
Willy Tarreau | fdb563c | 2010-01-31 15:43:27 +0100 | [diff] [blame] | 7043 | <cond> is an optional matching condition built from ACLs. It makes it |
| 7044 | possible to ignore this rule when other conditions are not met. |
| 7045 | |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7046 | Any header line matching extended regular expression <search> in the response |
| 7047 | will be completely deleted. Most common use of this is to remove unwanted |
Willy Tarreau | 3c92c5f | 2011-08-28 09:45:47 +0200 | [diff] [blame] | 7048 | and/or sensitive headers or cookies from a response before passing it to the |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7049 | client. |
| 7050 | |
| 7051 | Header transformations only apply to traffic which passes through HAProxy, |
| 7052 | and not to traffic generated by HAProxy, such as health-checks or error |
| 7053 | responses. Keep in mind that header names are not case-sensitive. |
| 7054 | |
| 7055 | Example : |
| 7056 | # remove the Server header from responses |
Willy Tarreau | 5e80e02 | 2013-05-25 08:31:25 +0200 | [diff] [blame] | 7057 | rspidel ^Server:.* |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7058 | |
Ruoshan Huang | eb5a363 | 2015-12-08 21:00:23 +0800 | [diff] [blame] | 7059 | See also: "rspadd", "rsprep", "reqdel", "http-response", section 6 about |
| 7060 | HTTP header manipulation, and section 7 about ACLs. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7061 | |
| 7062 | |
Willy Tarreau | fdb563c | 2010-01-31 15:43:27 +0100 | [diff] [blame] | 7063 | rspdeny <search> [{if | unless} <cond>] |
| 7064 | rspideny <search> [{if | unless} <cond>] (ignore case) |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7065 | Block an HTTP response if a line matches a regular expression |
| 7066 | May be used in sections : defaults | frontend | listen | backend |
| 7067 | no | yes | yes | yes |
| 7068 | Arguments : |
| 7069 | <search> is the regular expression applied to HTTP headers and to the |
| 7070 | response line. This is an extended regular expression, so |
| 7071 | parenthesis grouping is supported and no preliminary backslash |
| 7072 | is required. Any space or known delimiter must be escaped using |
| 7073 | a backslash ('\'). The pattern applies to a full line at a time. |
| 7074 | The "rspdeny" keyword strictly matches case while "rspideny" |
| 7075 | ignores case. |
| 7076 | |
Willy Tarreau | fdb563c | 2010-01-31 15:43:27 +0100 | [diff] [blame] | 7077 | <cond> is an optional matching condition built from ACLs. It makes it |
| 7078 | possible to ignore this rule when other conditions are not met. |
| 7079 | |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7080 | A response containing any line which matches extended regular expression |
| 7081 | <search> will mark the request as denied. The test applies both to the |
| 7082 | response line and to response headers. Keep in mind that header names are not |
| 7083 | case-sensitive. |
| 7084 | |
| 7085 | Main use of this keyword is to prevent sensitive information leak and to |
Willy Tarreau | ced2701 | 2008-01-17 20:35:34 +0100 | [diff] [blame] | 7086 | block the response before it reaches the client. If a response is denied, it |
| 7087 | will be replaced with an HTTP 502 error so that the client never retrieves |
| 7088 | any sensitive data. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7089 | |
| 7090 | It is easier, faster and more powerful to use ACLs to write access policies. |
| 7091 | Rspdeny should be avoided in new designs. |
| 7092 | |
| 7093 | Example : |
| 7094 | # Ensure that no content type matching ms-word will leak |
| 7095 | rspideny ^Content-type:\.*/ms-word |
| 7096 | |
Ruoshan Huang | eb5a363 | 2015-12-08 21:00:23 +0800 | [diff] [blame] | 7097 | See also: "reqdeny", "acl", "block", "http-response", section 6 about |
| 7098 | HTTP header manipulation and section 7 about ACLs. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7099 | |
| 7100 | |
Willy Tarreau | fdb563c | 2010-01-31 15:43:27 +0100 | [diff] [blame] | 7101 | rsprep <search> <string> [{if | unless} <cond>] |
| 7102 | rspirep <search> <string> [{if | unless} <cond>] (ignore case) |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7103 | Replace a regular expression with a string in an HTTP response line |
| 7104 | May be used in sections : defaults | frontend | listen | backend |
| 7105 | no | yes | yes | yes |
| 7106 | Arguments : |
| 7107 | <search> is the regular expression applied to HTTP headers and to the |
| 7108 | response line. This is an extended regular expression, so |
| 7109 | parenthesis grouping is supported and no preliminary backslash |
| 7110 | is required. Any space or known delimiter must be escaped using |
| 7111 | a backslash ('\'). The pattern applies to a full line at a time. |
| 7112 | The "rsprep" keyword strictly matches case while "rspirep" |
| 7113 | ignores case. |
| 7114 | |
| 7115 | <string> is the complete line to be added. Any space or known delimiter |
| 7116 | must be escaped using a backslash ('\'). References to matched |
| 7117 | pattern groups are possible using the common \N form, with N |
| 7118 | being a single digit between 0 and 9. Please refer to section |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 7119 | 6 about HTTP header manipulation for more information. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7120 | |
Willy Tarreau | fdb563c | 2010-01-31 15:43:27 +0100 | [diff] [blame] | 7121 | <cond> is an optional matching condition built from ACLs. It makes it |
| 7122 | possible to ignore this rule when other conditions are not met. |
| 7123 | |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7124 | Any line matching extended regular expression <search> in the response (both |
| 7125 | the response line and header lines) will be completely replaced with |
| 7126 | <string>. Most common use of this is to rewrite Location headers. |
| 7127 | |
| 7128 | Header transformations only apply to traffic which passes through HAProxy, |
| 7129 | and not to traffic generated by HAProxy, such as health-checks or error |
| 7130 | responses. Note that for increased readability, it is suggested to add enough |
| 7131 | spaces between the request and the response. Keep in mind that header names |
| 7132 | are not case-sensitive. |
| 7133 | |
| 7134 | Example : |
| 7135 | # replace "Location: 127.0.0.1:8080" with "Location: www.mydomain.com" |
| 7136 | rspirep ^Location:\ 127.0.0.1:8080 Location:\ www.mydomain.com |
| 7137 | |
Ruoshan Huang | eb5a363 | 2015-12-08 21:00:23 +0800 | [diff] [blame] | 7138 | See also: "rspadd", "rspdel", "reqrep", "http-response", section 6 about |
| 7139 | HTTP header manipulation, and section 7 about ACLs. |
Willy Tarreau | 303c035 | 2008-01-17 19:01:39 +0100 | [diff] [blame] | 7140 | |
| 7141 | |
David du Colombier | 486df47 | 2011-03-17 10:40:26 +0100 | [diff] [blame] | 7142 | server <name> <address>[:[port]] [param*] |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7143 | Declare a server in a backend |
| 7144 | May be used in sections : defaults | frontend | listen | backend |
| 7145 | no | no | yes | yes |
| 7146 | Arguments : |
| 7147 | <name> is the internal name assigned to this server. This name will |
Cyril Bonté | 941a0c6 | 2012-10-15 19:44:24 +0200 | [diff] [blame] | 7148 | appear in logs and alerts. If "http-send-name-header" is |
Mark Lamourine | c2247f0 | 2012-01-04 13:02:01 -0500 | [diff] [blame] | 7149 | set, it will be added to the request header sent to the server. |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7150 | |
David du Colombier | 486df47 | 2011-03-17 10:40:26 +0100 | [diff] [blame] | 7151 | <address> is the IPv4 or IPv6 address of the server. Alternatively, a |
| 7152 | resolvable hostname is supported, but this name will be resolved |
| 7153 | during start-up. Address "0.0.0.0" or "*" has a special meaning. |
| 7154 | It indicates that the connection will be forwarded to the same IP |
Willy Tarreau | d669a4f | 2010-07-13 14:49:50 +0200 | [diff] [blame] | 7155 | address as the one from the client connection. This is useful in |
| 7156 | transparent proxy architectures where the client's connection is |
| 7157 | intercepted and haproxy must forward to the original destination |
| 7158 | address. This is more or less what the "transparent" keyword does |
| 7159 | except that with a server it's possible to limit concurrency and |
Willy Tarreau | 2470928 | 2013-03-10 21:32:12 +0100 | [diff] [blame] | 7160 | to report statistics. Optionally, an address family prefix may be |
| 7161 | used before the address to force the family regardless of the |
| 7162 | address format, which can be useful to specify a path to a unix |
| 7163 | socket with no slash ('/'). Currently supported prefixes are : |
| 7164 | - 'ipv4@' -> address is always IPv4 |
| 7165 | - 'ipv6@' -> address is always IPv6 |
| 7166 | - 'unix@' -> address is a path to a local unix socket |
Willy Tarreau | ccfccef | 2014-05-10 01:49:15 +0200 | [diff] [blame] | 7167 | - 'abns@' -> address is in abstract namespace (Linux only) |
William Lallemand | b2f0745 | 2015-05-12 14:27:13 +0200 | [diff] [blame] | 7168 | You may want to reference some environment variables in the |
| 7169 | address parameter, see section 2.3 about environment |
| 7170 | variables. |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7171 | |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 7172 | <port> is an optional port specification. If set, all connections will |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7173 | be sent to this port. If unset, the same port the client |
| 7174 | connected to will be used. The port may also be prefixed by a "+" |
| 7175 | or a "-". In this case, the server's port will be determined by |
| 7176 | adding this value to the client's port. |
| 7177 | |
| 7178 | <param*> is a list of parameters for this server. The "server" keywords |
| 7179 | accepts an important number of options and has a complete section |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 7180 | dedicated to it. Please refer to section 5 for more details. |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7181 | |
| 7182 | Examples : |
| 7183 | server first 10.1.1.1:1080 cookie first check inter 1000 |
| 7184 | server second 10.1.1.2:1080 cookie second check inter 1000 |
Willy Tarreau | 2470928 | 2013-03-10 21:32:12 +0100 | [diff] [blame] | 7185 | server transp ipv4@ |
William Lallemand | b2f0745 | 2015-05-12 14:27:13 +0200 | [diff] [blame] | 7186 | server backup "${SRV_BACKUP}:1080" backup |
| 7187 | server www1_dc1 "${LAN_DC1}.101:80" |
| 7188 | server www1_dc2 "${LAN_DC2}.101:80" |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7189 | |
Willy Tarreau | 55dcaf6 | 2015-09-27 15:03:15 +0200 | [diff] [blame] | 7190 | Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole |
| 7191 | sun_path length is used for the address length. Some other programs |
| 7192 | such as socat use the string length only by default. Pass the option |
| 7193 | ",unix-tightsocklen=0" to any abstract socket definition in socat to |
| 7194 | make it compatible with HAProxy's. |
| 7195 | |
Mark Lamourine | c2247f0 | 2012-01-04 13:02:01 -0500 | [diff] [blame] | 7196 | See also: "default-server", "http-send-name-header" and section 5 about |
| 7197 | server options |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7198 | |
Baptiste Assmann | 01c6cc3 | 2015-08-23 11:45:29 +0200 | [diff] [blame] | 7199 | server-state-file-name [<file>] |
| 7200 | Set the server state file to read, load and apply to servers available in |
| 7201 | this backend. It only applies when the directive "load-server-state-from-file" |
| 7202 | is set to "local". When <file> is not provided or if this directive is not |
| 7203 | set, then backend name is used. If <file> starts with a slash '/', then it is |
| 7204 | considered as an absolute path. Otherwise, <file> is concatenated to the |
| 7205 | global directive "server-state-file-base". |
| 7206 | |
| 7207 | Example: the minimal configuration below would make HAProxy look for the |
| 7208 | state server file '/etc/haproxy/states/bk': |
| 7209 | |
| 7210 | global |
| 7211 | server-state-file-base /etc/haproxy/states |
| 7212 | |
| 7213 | backend bk |
| 7214 | load-server-state-from-file |
| 7215 | |
| 7216 | See also: "server-state-file-base", "load-server-state-from-file", and |
| 7217 | "show servers state" |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7218 | |
| 7219 | source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ] |
Willy Tarreau | bce7088 | 2009-09-07 11:51:47 +0200 | [diff] [blame] | 7220 | source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ] |
Willy Tarreau | d53f96b | 2009-02-04 18:46:54 +0100 | [diff] [blame] | 7221 | source <addr>[:<port>] [interface <name>] |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7222 | Set the source address for outgoing connections |
| 7223 | May be used in sections : defaults | frontend | listen | backend |
| 7224 | yes | no | yes | yes |
| 7225 | Arguments : |
| 7226 | <addr> is the IPv4 address HAProxy will bind to before connecting to a |
| 7227 | server. This address is also used as a source for health checks. |
Willy Tarreau | 2470928 | 2013-03-10 21:32:12 +0100 | [diff] [blame] | 7228 | |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7229 | The default value of 0.0.0.0 means that the system will select |
Willy Tarreau | 2470928 | 2013-03-10 21:32:12 +0100 | [diff] [blame] | 7230 | the most appropriate address to reach its destination. Optionally |
| 7231 | an address family prefix may be used before the address to force |
| 7232 | the family regardless of the address format, which can be useful |
| 7233 | to specify a path to a unix socket with no slash ('/'). Currently |
| 7234 | supported prefixes are : |
| 7235 | - 'ipv4@' -> address is always IPv4 |
| 7236 | - 'ipv6@' -> address is always IPv6 |
| 7237 | - 'unix@' -> address is a path to a local unix socket |
Willy Tarreau | ccfccef | 2014-05-10 01:49:15 +0200 | [diff] [blame] | 7238 | - 'abns@' -> address is in abstract namespace (Linux only) |
Cyril Bonté | 307ee1e | 2015-09-28 23:16:06 +0200 | [diff] [blame] | 7239 | You may want to reference some environment variables in the |
| 7240 | address parameter, see section 2.3 about environment variables. |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7241 | |
| 7242 | <port> is an optional port. It is normally not needed but may be useful |
| 7243 | in some very specific contexts. The default value of zero means |
Willy Tarreau | c6f4ce8 | 2009-06-10 11:09:37 +0200 | [diff] [blame] | 7244 | the system will select a free port. Note that port ranges are not |
| 7245 | supported in the backend. If you want to force port ranges, you |
| 7246 | have to specify them on each "server" line. |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7247 | |
| 7248 | <addr2> is the IP address to present to the server when connections are |
| 7249 | forwarded in full transparent proxy mode. This is currently only |
| 7250 | supported on some patched Linux kernels. When this address is |
| 7251 | specified, clients connecting to the server will be presented |
| 7252 | with this address, while health checks will still use the address |
| 7253 | <addr>. |
| 7254 | |
| 7255 | <port2> is the optional port to present to the server when connections |
| 7256 | are forwarded in full transparent proxy mode (see <addr2> above). |
| 7257 | The default value of zero means the system will select a free |
| 7258 | port. |
| 7259 | |
Willy Tarreau | bce7088 | 2009-09-07 11:51:47 +0200 | [diff] [blame] | 7260 | <hdr> is the name of a HTTP header in which to fetch the IP to bind to. |
| 7261 | This is the name of a comma-separated header list which can |
| 7262 | contain multiple IP addresses. By default, the last occurrence is |
| 7263 | used. This is designed to work with the X-Forwarded-For header |
Baptiste Assmann | ea3e73b | 2013-02-02 23:47:49 +0100 | [diff] [blame] | 7264 | and to automatically bind to the client's IP address as seen |
Willy Tarreau | bce7088 | 2009-09-07 11:51:47 +0200 | [diff] [blame] | 7265 | by previous proxy, typically Stunnel. In order to use another |
| 7266 | occurrence from the last one, please see the <occ> parameter |
| 7267 | below. When the header (or occurrence) is not found, no binding |
| 7268 | is performed so that the proxy's default IP address is used. Also |
| 7269 | keep in mind that the header name is case insensitive, as for any |
| 7270 | HTTP header. |
| 7271 | |
| 7272 | <occ> is the occurrence number of a value to be used in a multi-value |
| 7273 | header. This is to be used in conjunction with "hdr_ip(<hdr>)", |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 7274 | in order to specify which occurrence to use for the source IP |
Willy Tarreau | bce7088 | 2009-09-07 11:51:47 +0200 | [diff] [blame] | 7275 | address. Positive values indicate a position from the first |
| 7276 | occurrence, 1 being the first one. Negative values indicate |
| 7277 | positions relative to the last one, -1 being the last one. This |
| 7278 | is helpful for situations where an X-Forwarded-For header is set |
| 7279 | at the entry point of an infrastructure and must be used several |
| 7280 | proxy layers away. When this value is not specified, -1 is |
| 7281 | assumed. Passing a zero here disables the feature. |
| 7282 | |
Willy Tarreau | d53f96b | 2009-02-04 18:46:54 +0100 | [diff] [blame] | 7283 | <name> is an optional interface name to which to bind to for outgoing |
| 7284 | traffic. On systems supporting this features (currently, only |
| 7285 | Linux), this allows one to bind all traffic to the server to |
| 7286 | this interface even if it is not the one the system would select |
| 7287 | based on routing tables. This should be used with extreme care. |
| 7288 | Note that using this option requires root privileges. |
| 7289 | |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7290 | The "source" keyword is useful in complex environments where a specific |
| 7291 | address only is allowed to connect to the servers. It may be needed when a |
| 7292 | private address must be used through a public gateway for instance, and it is |
| 7293 | known that the system cannot determine the adequate source address by itself. |
| 7294 | |
| 7295 | An extension which is available on certain patched Linux kernels may be used |
| 7296 | through the "usesrc" optional keyword. It makes it possible to connect to the |
| 7297 | servers with an IP address which does not belong to the system itself. This |
| 7298 | is called "full transparent proxy mode". For this to work, the destination |
| 7299 | servers have to route their traffic back to this address through the machine |
| 7300 | running HAProxy, and IP forwarding must generally be enabled on this machine. |
| 7301 | |
| 7302 | In this "full transparent proxy" mode, it is possible to force a specific IP |
| 7303 | address to be presented to the servers. This is not much used in fact. A more |
| 7304 | common use is to tell HAProxy to present the client's IP address. For this, |
| 7305 | there are two methods : |
| 7306 | |
| 7307 | - present the client's IP and port addresses. This is the most transparent |
| 7308 | mode, but it can cause problems when IP connection tracking is enabled on |
| 7309 | the machine, because a same connection may be seen twice with different |
| 7310 | states. However, this solution presents the huge advantage of not |
| 7311 | limiting the system to the 64k outgoing address+port couples, because all |
| 7312 | of the client ranges may be used. |
| 7313 | |
| 7314 | - present only the client's IP address and select a spare port. This |
| 7315 | solution is still quite elegant but slightly less transparent (downstream |
| 7316 | firewalls logs will not match upstream's). It also presents the downside |
| 7317 | of limiting the number of concurrent connections to the usual 64k ports. |
| 7318 | However, since the upstream and downstream ports are different, local IP |
| 7319 | connection tracking on the machine will not be upset by the reuse of the |
| 7320 | same session. |
| 7321 | |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7322 | This option sets the default source for all servers in the backend. It may |
| 7323 | also be specified in a "defaults" section. Finer source address specification |
| 7324 | is possible at the server level using the "source" server option. Refer to |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 7325 | section 5 for more information. |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7326 | |
Baptiste Assmann | 91bd337 | 2015-07-17 21:59:42 +0200 | [diff] [blame] | 7327 | In order to work, "usesrc" requires root privileges. |
| 7328 | |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7329 | Examples : |
| 7330 | backend private |
| 7331 | # Connect to the servers using our 192.168.1.200 source address |
| 7332 | source 192.168.1.200 |
| 7333 | |
| 7334 | backend transparent_ssl1 |
| 7335 | # Connect to the SSL farm from the client's source address |
| 7336 | source 192.168.1.200 usesrc clientip |
| 7337 | |
| 7338 | backend transparent_ssl2 |
| 7339 | # Connect to the SSL farm from the client's source address and port |
| 7340 | # not recommended if IP conntrack is present on the local machine. |
| 7341 | source 192.168.1.200 usesrc client |
| 7342 | |
| 7343 | backend transparent_ssl3 |
| 7344 | # Connect to the SSL farm from the client's source address. It |
| 7345 | # is more conntrack-friendly. |
| 7346 | source 192.168.1.200 usesrc clientip |
| 7347 | |
| 7348 | backend transparent_smtp |
| 7349 | # Connect to the SMTP farm from the client's source address/port |
| 7350 | # with Tproxy version 4. |
| 7351 | source 0.0.0.0 usesrc clientip |
| 7352 | |
Willy Tarreau | bce7088 | 2009-09-07 11:51:47 +0200 | [diff] [blame] | 7353 | backend transparent_http |
| 7354 | # Connect to the servers using the client's IP as seen by previous |
| 7355 | # proxy. |
| 7356 | source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1) |
| 7357 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 7358 | See also : the "source" server option in section 5, the Tproxy patches for |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7359 | the Linux kernel on www.balabit.com, the "bind" keyword. |
| 7360 | |
Krzysztof Piotr Oledzki | 25b501a | 2008-01-06 16:36:16 +0100 | [diff] [blame] | 7361 | |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 7362 | srvtimeout <timeout> (deprecated) |
| 7363 | Set the maximum inactivity time on the server side. |
| 7364 | May be used in sections : defaults | frontend | listen | backend |
| 7365 | yes | no | yes | yes |
| 7366 | Arguments : |
| 7367 | <timeout> is the timeout value specified in milliseconds by default, but |
| 7368 | can be in any other unit if the number is suffixed by the unit, |
| 7369 | as explained at the top of this document. |
| 7370 | |
| 7371 | The inactivity timeout applies when the server is expected to acknowledge or |
| 7372 | send data. In HTTP mode, this timeout is particularly important to consider |
| 7373 | during the first phase of the server's response, when it has to send the |
| 7374 | headers, as it directly represents the server's processing time for the |
| 7375 | request. To find out what value to put there, it's often good to start with |
| 7376 | what would be considered as unacceptable response times, then check the logs |
| 7377 | to observe the response time distribution, and adjust the value accordingly. |
| 7378 | |
| 7379 | The value is specified in milliseconds by default, but can be in any other |
| 7380 | unit if the number is suffixed by the unit, as specified at the top of this |
| 7381 | document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly |
| 7382 | recommended that the client timeout remains equal to the server timeout in |
| 7383 | order to avoid complex situations to debug. Whatever the expected server |
Willy Tarreau | d2a4aa2 | 2008-01-31 15:28:22 +0100 | [diff] [blame] | 7384 | response times, it is a good practice to cover at least one or several TCP |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 7385 | packet losses by specifying timeouts that are slightly above multiples of 3 |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 7386 | seconds (eg: 4 or 5 seconds minimum). |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 7387 | |
| 7388 | This parameter is specific to backends, but can be specified once for all in |
| 7389 | "defaults" sections. This is in fact one of the easiest solutions not to |
| 7390 | forget about it. An unspecified timeout results in an infinite timeout, which |
| 7391 | is not recommended. Such a usage is accepted and works but reports a warning |
| 7392 | during startup because it may results in accumulation of expired sessions in |
| 7393 | the system if the system's timeouts are not configured either. |
| 7394 | |
| 7395 | This parameter is provided for compatibility but is currently deprecated. |
| 7396 | Please use "timeout server" instead. |
| 7397 | |
Willy Tarreau | ce887fd | 2012-05-12 12:50:00 +0200 | [diff] [blame] | 7398 | See also : "timeout server", "timeout tunnel", "timeout client" and |
| 7399 | "clitimeout". |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 7400 | |
| 7401 | |
Cyril Bonté | 66c327d | 2010-10-12 00:14:37 +0200 | [diff] [blame] | 7402 | stats admin { if | unless } <cond> |
| 7403 | Enable statistics admin level if/unless a condition is matched |
| 7404 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | ed2119c | 2014-04-24 22:10:39 +0200 | [diff] [blame] | 7405 | no | yes | yes | yes |
Cyril Bonté | 66c327d | 2010-10-12 00:14:37 +0200 | [diff] [blame] | 7406 | |
| 7407 | This statement enables the statistics admin level if/unless a condition is |
| 7408 | matched. |
| 7409 | |
| 7410 | The admin level allows to enable/disable servers from the web interface. By |
| 7411 | default, statistics page is read-only for security reasons. |
| 7412 | |
Cyril Bonté | 02ff8ef | 2010-12-14 22:48:49 +0100 | [diff] [blame] | 7413 | Note : Consider not using this feature in multi-process mode (nbproc > 1) |
| 7414 | unless you know what you do : memory is not shared between the |
| 7415 | processes, which can result in random behaviours. |
| 7416 | |
Cyril Bonté | 23b39d9 | 2011-02-10 22:54:44 +0100 | [diff] [blame] | 7417 | Currently, the POST request is limited to the buffer size minus the reserved |
| 7418 | buffer space, which means that if the list of servers is too long, the |
| 7419 | request won't be processed. It is recommended to alter few servers at a |
| 7420 | time. |
Cyril Bonté | 66c327d | 2010-10-12 00:14:37 +0200 | [diff] [blame] | 7421 | |
| 7422 | Example : |
| 7423 | # statistics admin level only for localhost |
| 7424 | backend stats_localhost |
| 7425 | stats enable |
| 7426 | stats admin if LOCALHOST |
| 7427 | |
| 7428 | Example : |
| 7429 | # statistics admin level always enabled because of the authentication |
| 7430 | backend stats_auth |
| 7431 | stats enable |
| 7432 | stats auth admin:AdMiN123 |
| 7433 | stats admin if TRUE |
| 7434 | |
| 7435 | Example : |
| 7436 | # statistics admin level depends on the authenticated user |
| 7437 | userlist stats-auth |
| 7438 | group admin users admin |
| 7439 | user admin insecure-password AdMiN123 |
| 7440 | group readonly users haproxy |
| 7441 | user haproxy insecure-password haproxy |
| 7442 | |
| 7443 | backend stats_auth |
| 7444 | stats enable |
| 7445 | acl AUTH http_auth(stats-auth) |
| 7446 | acl AUTH_ADMIN http_auth_group(stats-auth) admin |
| 7447 | stats http-request auth unless AUTH |
| 7448 | stats admin if AUTH_ADMIN |
| 7449 | |
Cyril Bonté | 02ff8ef | 2010-12-14 22:48:49 +0100 | [diff] [blame] | 7450 | See also : "stats enable", "stats auth", "stats http-request", "nbproc", |
| 7451 | "bind-process", section 3.4 about userlists and section 7 about |
| 7452 | ACL usage. |
Cyril Bonté | 66c327d | 2010-10-12 00:14:37 +0200 | [diff] [blame] | 7453 | |
| 7454 | |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7455 | stats auth <user>:<passwd> |
| 7456 | Enable statistics with authentication and grant access to an account |
| 7457 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | ed2119c | 2014-04-24 22:10:39 +0200 | [diff] [blame] | 7458 | yes | yes | yes | yes |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7459 | Arguments : |
| 7460 | <user> is a user name to grant access to |
| 7461 | |
| 7462 | <passwd> is the cleartext password associated to this user |
| 7463 | |
| 7464 | This statement enables statistics with default settings, and restricts access |
| 7465 | to declared users only. It may be repeated as many times as necessary to |
| 7466 | allow as many users as desired. When a user tries to access the statistics |
| 7467 | without a valid account, a "401 Forbidden" response will be returned so that |
| 7468 | the browser asks the user to provide a valid user and password. The real |
| 7469 | which will be returned to the browser is configurable using "stats realm". |
| 7470 | |
| 7471 | Since the authentication method is HTTP Basic Authentication, the passwords |
| 7472 | circulate in cleartext on the network. Thus, it was decided that the |
| 7473 | configuration file would also use cleartext passwords to remind the users |
Willy Tarreau | 3c92c5f | 2011-08-28 09:45:47 +0200 | [diff] [blame] | 7474 | that those ones should not be sensitive and not shared with any other account. |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7475 | |
| 7476 | It is also possible to reduce the scope of the proxies which appear in the |
| 7477 | report using "stats scope". |
| 7478 | |
| 7479 | Though this statement alone is enough to enable statistics reporting, it is |
| 7480 | recommended to set all other settings in order to avoid relying on default |
| 7481 | unobvious parameters. |
| 7482 | |
| 7483 | Example : |
| 7484 | # public access (limited to this backend only) |
| 7485 | backend public_www |
| 7486 | server srv1 192.168.0.1:80 |
| 7487 | stats enable |
| 7488 | stats hide-version |
| 7489 | stats scope . |
| 7490 | stats uri /admin?stats |
| 7491 | stats realm Haproxy\ Statistics |
| 7492 | stats auth admin1:AdMiN123 |
| 7493 | stats auth admin2:AdMiN321 |
| 7494 | |
| 7495 | # internal monitoring access (unlimited) |
| 7496 | backend private_monitoring |
| 7497 | stats enable |
| 7498 | stats uri /admin?stats |
| 7499 | stats refresh 5s |
| 7500 | |
| 7501 | See also : "stats enable", "stats realm", "stats scope", "stats uri" |
| 7502 | |
| 7503 | |
| 7504 | stats enable |
| 7505 | Enable statistics reporting with default settings |
| 7506 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | ed2119c | 2014-04-24 22:10:39 +0200 | [diff] [blame] | 7507 | yes | yes | yes | yes |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7508 | Arguments : none |
| 7509 | |
| 7510 | This statement enables statistics reporting with default settings defined |
| 7511 | at build time. Unless stated otherwise, these settings are used : |
| 7512 | - stats uri : /haproxy?stats |
| 7513 | - stats realm : "HAProxy Statistics" |
| 7514 | - stats auth : no authentication |
| 7515 | - stats scope : no restriction |
| 7516 | |
| 7517 | Though this statement alone is enough to enable statistics reporting, it is |
| 7518 | recommended to set all other settings in order to avoid relying on default |
| 7519 | unobvious parameters. |
| 7520 | |
| 7521 | Example : |
| 7522 | # public access (limited to this backend only) |
| 7523 | backend public_www |
| 7524 | server srv1 192.168.0.1:80 |
| 7525 | stats enable |
| 7526 | stats hide-version |
| 7527 | stats scope . |
| 7528 | stats uri /admin?stats |
| 7529 | stats realm Haproxy\ Statistics |
| 7530 | stats auth admin1:AdMiN123 |
| 7531 | stats auth admin2:AdMiN321 |
| 7532 | |
| 7533 | # internal monitoring access (unlimited) |
| 7534 | backend private_monitoring |
| 7535 | stats enable |
| 7536 | stats uri /admin?stats |
| 7537 | stats refresh 5s |
| 7538 | |
| 7539 | See also : "stats auth", "stats realm", "stats uri" |
| 7540 | |
| 7541 | |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7542 | stats hide-version |
| 7543 | Enable statistics and hide HAProxy version reporting |
Willy Tarreau | 1d45b7c | 2009-08-16 10:29:18 +0200 | [diff] [blame] | 7544 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | ed2119c | 2014-04-24 22:10:39 +0200 | [diff] [blame] | 7545 | yes | yes | yes | yes |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7546 | Arguments : none |
Willy Tarreau | 1d45b7c | 2009-08-16 10:29:18 +0200 | [diff] [blame] | 7547 | |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7548 | By default, the stats page reports some useful status information along with |
| 7549 | the statistics. Among them is HAProxy's version. However, it is generally |
| 7550 | considered dangerous to report precise version to anyone, as it can help them |
| 7551 | target known weaknesses with specific attacks. The "stats hide-version" |
| 7552 | statement removes the version from the statistics report. This is recommended |
| 7553 | for public sites or any site with a weak login/password. |
Willy Tarreau | 1d45b7c | 2009-08-16 10:29:18 +0200 | [diff] [blame] | 7554 | |
Krzysztof Piotr Oledzki | 48cb2ae | 2009-10-02 22:51:14 +0200 | [diff] [blame] | 7555 | Though this statement alone is enough to enable statistics reporting, it is |
| 7556 | recommended to set all other settings in order to avoid relying on default |
| 7557 | unobvious parameters. |
| 7558 | |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7559 | Example : |
| 7560 | # public access (limited to this backend only) |
| 7561 | backend public_www |
| 7562 | server srv1 192.168.0.1:80 |
Krzysztof Piotr Oledzki | 48cb2ae | 2009-10-02 22:51:14 +0200 | [diff] [blame] | 7563 | stats enable |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7564 | stats hide-version |
| 7565 | stats scope . |
| 7566 | stats uri /admin?stats |
| 7567 | stats realm Haproxy\ Statistics |
| 7568 | stats auth admin1:AdMiN123 |
| 7569 | stats auth admin2:AdMiN321 |
Willy Tarreau | 1d45b7c | 2009-08-16 10:29:18 +0200 | [diff] [blame] | 7570 | |
Willy Tarreau | 1d45b7c | 2009-08-16 10:29:18 +0200 | [diff] [blame] | 7571 | # internal monitoring access (unlimited) |
| 7572 | backend private_monitoring |
| 7573 | stats enable |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7574 | stats uri /admin?stats |
| 7575 | stats refresh 5s |
Krzysztof Piotr Oledzki | 15514c2 | 2010-01-04 16:03:09 +0100 | [diff] [blame] | 7576 | |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7577 | See also : "stats auth", "stats enable", "stats realm", "stats uri" |
Willy Tarreau | 1d45b7c | 2009-08-16 10:29:18 +0200 | [diff] [blame] | 7578 | |
Willy Tarreau | 983e01e | 2010-01-11 18:42:06 +0100 | [diff] [blame] | 7579 | |
Cyril Bonté | 2be1b3f | 2010-09-30 23:46:30 +0200 | [diff] [blame] | 7580 | stats http-request { allow | deny | auth [realm <realm>] } |
| 7581 | [ { if | unless } <condition> ] |
| 7582 | Access control for statistics |
| 7583 | |
| 7584 | May be used in sections: defaults | frontend | listen | backend |
| 7585 | no | no | yes | yes |
| 7586 | |
| 7587 | As "http-request", these set of options allow to fine control access to |
| 7588 | statistics. Each option may be followed by if/unless and acl. |
| 7589 | First option with matched condition (or option without condition) is final. |
| 7590 | For "deny" a 403 error will be returned, for "allow" normal processing is |
| 7591 | performed, for "auth" a 401/407 error code is returned so the client |
| 7592 | should be asked to enter a username and password. |
| 7593 | |
| 7594 | There is no fixed limit to the number of http-request statements per |
| 7595 | instance. |
| 7596 | |
| 7597 | See also : "http-request", section 3.4 about userlists and section 7 |
| 7598 | about ACL usage. |
| 7599 | |
| 7600 | |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7601 | stats realm <realm> |
| 7602 | Enable statistics and set authentication realm |
| 7603 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | ed2119c | 2014-04-24 22:10:39 +0200 | [diff] [blame] | 7604 | yes | yes | yes | yes |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7605 | Arguments : |
| 7606 | <realm> is the name of the HTTP Basic Authentication realm reported to |
| 7607 | the browser. The browser uses it to display it in the pop-up |
| 7608 | inviting the user to enter a valid username and password. |
| 7609 | |
| 7610 | The realm is read as a single word, so any spaces in it should be escaped |
| 7611 | using a backslash ('\'). |
| 7612 | |
| 7613 | This statement is useful only in conjunction with "stats auth" since it is |
| 7614 | only related to authentication. |
| 7615 | |
| 7616 | Though this statement alone is enough to enable statistics reporting, it is |
| 7617 | recommended to set all other settings in order to avoid relying on default |
| 7618 | unobvious parameters. |
| 7619 | |
| 7620 | Example : |
| 7621 | # public access (limited to this backend only) |
| 7622 | backend public_www |
| 7623 | server srv1 192.168.0.1:80 |
| 7624 | stats enable |
| 7625 | stats hide-version |
| 7626 | stats scope . |
| 7627 | stats uri /admin?stats |
| 7628 | stats realm Haproxy\ Statistics |
| 7629 | stats auth admin1:AdMiN123 |
| 7630 | stats auth admin2:AdMiN321 |
| 7631 | |
| 7632 | # internal monitoring access (unlimited) |
| 7633 | backend private_monitoring |
| 7634 | stats enable |
| 7635 | stats uri /admin?stats |
| 7636 | stats refresh 5s |
| 7637 | |
| 7638 | See also : "stats auth", "stats enable", "stats uri" |
| 7639 | |
| 7640 | |
| 7641 | stats refresh <delay> |
| 7642 | Enable statistics with automatic refresh |
| 7643 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | ed2119c | 2014-04-24 22:10:39 +0200 | [diff] [blame] | 7644 | yes | yes | yes | yes |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7645 | Arguments : |
| 7646 | <delay> is the suggested refresh delay, specified in seconds, which will |
| 7647 | be returned to the browser consulting the report page. While the |
| 7648 | browser is free to apply any delay, it will generally respect it |
| 7649 | and refresh the page this every seconds. The refresh interval may |
| 7650 | be specified in any other non-default time unit, by suffixing the |
| 7651 | unit after the value, as explained at the top of this document. |
| 7652 | |
| 7653 | This statement is useful on monitoring displays with a permanent page |
| 7654 | reporting the load balancer's activity. When set, the HTML report page will |
| 7655 | include a link "refresh"/"stop refresh" so that the user can select whether |
| 7656 | he wants automatic refresh of the page or not. |
| 7657 | |
| 7658 | Though this statement alone is enough to enable statistics reporting, it is |
| 7659 | recommended to set all other settings in order to avoid relying on default |
| 7660 | unobvious parameters. |
| 7661 | |
| 7662 | Example : |
| 7663 | # public access (limited to this backend only) |
| 7664 | backend public_www |
| 7665 | server srv1 192.168.0.1:80 |
| 7666 | stats enable |
| 7667 | stats hide-version |
| 7668 | stats scope . |
| 7669 | stats uri /admin?stats |
| 7670 | stats realm Haproxy\ Statistics |
| 7671 | stats auth admin1:AdMiN123 |
| 7672 | stats auth admin2:AdMiN321 |
| 7673 | |
| 7674 | # internal monitoring access (unlimited) |
| 7675 | backend private_monitoring |
| 7676 | stats enable |
| 7677 | stats uri /admin?stats |
| 7678 | stats refresh 5s |
| 7679 | |
| 7680 | See also : "stats auth", "stats enable", "stats realm", "stats uri" |
| 7681 | |
| 7682 | |
| 7683 | stats scope { <name> | "." } |
| 7684 | Enable statistics and limit access scope |
| 7685 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | ed2119c | 2014-04-24 22:10:39 +0200 | [diff] [blame] | 7686 | yes | yes | yes | yes |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7687 | Arguments : |
| 7688 | <name> is the name of a listen, frontend or backend section to be |
| 7689 | reported. The special name "." (a single dot) designates the |
| 7690 | section in which the statement appears. |
| 7691 | |
| 7692 | When this statement is specified, only the sections enumerated with this |
| 7693 | statement will appear in the report. All other ones will be hidden. This |
| 7694 | statement may appear as many times as needed if multiple sections need to be |
| 7695 | reported. Please note that the name checking is performed as simple string |
| 7696 | comparisons, and that it is never checked that a give section name really |
| 7697 | exists. |
| 7698 | |
| 7699 | Though this statement alone is enough to enable statistics reporting, it is |
| 7700 | recommended to set all other settings in order to avoid relying on default |
| 7701 | unobvious parameters. |
| 7702 | |
| 7703 | Example : |
| 7704 | # public access (limited to this backend only) |
| 7705 | backend public_www |
| 7706 | server srv1 192.168.0.1:80 |
| 7707 | stats enable |
| 7708 | stats hide-version |
| 7709 | stats scope . |
| 7710 | stats uri /admin?stats |
| 7711 | stats realm Haproxy\ Statistics |
| 7712 | stats auth admin1:AdMiN123 |
| 7713 | stats auth admin2:AdMiN321 |
| 7714 | |
| 7715 | # internal monitoring access (unlimited) |
| 7716 | backend private_monitoring |
| 7717 | stats enable |
| 7718 | stats uri /admin?stats |
| 7719 | stats refresh 5s |
| 7720 | |
| 7721 | See also : "stats auth", "stats enable", "stats realm", "stats uri" |
| 7722 | |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7723 | |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 7724 | stats show-desc [ <desc> ] |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7725 | Enable reporting of a description on the statistics page. |
| 7726 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | ed2119c | 2014-04-24 22:10:39 +0200 | [diff] [blame] | 7727 | yes | yes | yes | yes |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7728 | |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 7729 | <desc> is an optional description to be reported. If unspecified, the |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7730 | description from global section is automatically used instead. |
| 7731 | |
| 7732 | This statement is useful for users that offer shared services to their |
| 7733 | customers, where node or description should be different for each customer. |
| 7734 | |
| 7735 | Though this statement alone is enough to enable statistics reporting, it is |
| 7736 | recommended to set all other settings in order to avoid relying on default |
Dmitry Sivachenko | 7823de3 | 2012-05-16 14:00:26 +0400 | [diff] [blame] | 7737 | unobvious parameters. By default description is not shown. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7738 | |
| 7739 | Example : |
| 7740 | # internal monitoring access (unlimited) |
| 7741 | backend private_monitoring |
| 7742 | stats enable |
| 7743 | stats show-desc Master node for Europe, Asia, Africa |
| 7744 | stats uri /admin?stats |
| 7745 | stats refresh 5s |
| 7746 | |
| 7747 | See also: "show-node", "stats enable", "stats uri" and "description" in |
| 7748 | global section. |
| 7749 | |
| 7750 | |
| 7751 | stats show-legends |
Willy Tarreau | ed2119c | 2014-04-24 22:10:39 +0200 | [diff] [blame] | 7752 | Enable reporting additional information on the statistics page |
| 7753 | May be used in sections : defaults | frontend | listen | backend |
| 7754 | yes | yes | yes | yes |
| 7755 | Arguments : none |
| 7756 | |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 7757 | Enable reporting additional information on the statistics page : |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7758 | - cap: capabilities (proxy) |
| 7759 | - mode: one of tcp, http or health (proxy) |
| 7760 | - id: SNMP ID (proxy, socket, server) |
| 7761 | - IP (socket, server) |
| 7762 | - cookie (backend, server) |
| 7763 | |
| 7764 | Though this statement alone is enough to enable statistics reporting, it is |
| 7765 | recommended to set all other settings in order to avoid relying on default |
Dmitry Sivachenko | 7823de3 | 2012-05-16 14:00:26 +0400 | [diff] [blame] | 7766 | unobvious parameters. Default behaviour is not to show this information. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7767 | |
| 7768 | See also: "stats enable", "stats uri". |
| 7769 | |
| 7770 | |
| 7771 | stats show-node [ <name> ] |
| 7772 | Enable reporting of a host name on the statistics page. |
| 7773 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | ed2119c | 2014-04-24 22:10:39 +0200 | [diff] [blame] | 7774 | yes | yes | yes | yes |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7775 | Arguments: |
| 7776 | <name> is an optional name to be reported. If unspecified, the |
| 7777 | node name from global section is automatically used instead. |
| 7778 | |
| 7779 | This statement is useful for users that offer shared services to their |
| 7780 | customers, where node or description might be different on a stats page |
Dmitry Sivachenko | 7823de3 | 2012-05-16 14:00:26 +0400 | [diff] [blame] | 7781 | provided for each customer. Default behaviour is not to show host name. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7782 | |
| 7783 | Though this statement alone is enough to enable statistics reporting, it is |
| 7784 | recommended to set all other settings in order to avoid relying on default |
| 7785 | unobvious parameters. |
| 7786 | |
| 7787 | Example: |
| 7788 | # internal monitoring access (unlimited) |
| 7789 | backend private_monitoring |
| 7790 | stats enable |
| 7791 | stats show-node Europe-1 |
| 7792 | stats uri /admin?stats |
| 7793 | stats refresh 5s |
| 7794 | |
| 7795 | See also: "show-desc", "stats enable", "stats uri", and "node" in global |
| 7796 | section. |
| 7797 | |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7798 | |
| 7799 | stats uri <prefix> |
| 7800 | Enable statistics and define the URI prefix to access them |
| 7801 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | ed2119c | 2014-04-24 22:10:39 +0200 | [diff] [blame] | 7802 | yes | yes | yes | yes |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7803 | Arguments : |
| 7804 | <prefix> is the prefix of any URI which will be redirected to stats. This |
| 7805 | prefix may contain a question mark ('?') to indicate part of a |
| 7806 | query string. |
| 7807 | |
| 7808 | The statistics URI is intercepted on the relayed traffic, so it appears as a |
| 7809 | page within the normal application. It is strongly advised to ensure that the |
| 7810 | selected URI will never appear in the application, otherwise it will never be |
| 7811 | possible to reach it in the application. |
| 7812 | |
| 7813 | The default URI compiled in haproxy is "/haproxy?stats", but this may be |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 7814 | changed at build time, so it's better to always explicitly specify it here. |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7815 | It is generally a good idea to include a question mark in the URI so that |
| 7816 | intermediate proxies refrain from caching the results. Also, since any string |
| 7817 | beginning with the prefix will be accepted as a stats request, the question |
| 7818 | mark helps ensuring that no valid URI will begin with the same words. |
| 7819 | |
| 7820 | It is sometimes very convenient to use "/" as the URI prefix, and put that |
| 7821 | statement in a "listen" instance of its own. That makes it easy to dedicate |
| 7822 | an address or a port to statistics only. |
| 7823 | |
| 7824 | Though this statement alone is enough to enable statistics reporting, it is |
| 7825 | recommended to set all other settings in order to avoid relying on default |
| 7826 | unobvious parameters. |
| 7827 | |
| 7828 | Example : |
| 7829 | # public access (limited to this backend only) |
| 7830 | backend public_www |
| 7831 | server srv1 192.168.0.1:80 |
| 7832 | stats enable |
| 7833 | stats hide-version |
| 7834 | stats scope . |
| 7835 | stats uri /admin?stats |
| 7836 | stats realm Haproxy\ Statistics |
| 7837 | stats auth admin1:AdMiN123 |
| 7838 | stats auth admin2:AdMiN321 |
| 7839 | |
| 7840 | # internal monitoring access (unlimited) |
| 7841 | backend private_monitoring |
| 7842 | stats enable |
| 7843 | stats uri /admin?stats |
| 7844 | stats refresh 5s |
| 7845 | |
| 7846 | See also : "stats auth", "stats enable", "stats realm" |
| 7847 | |
| 7848 | |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7849 | stick match <pattern> [table <table>] [{if | unless} <cond>] |
| 7850 | Define a request pattern matching condition to stick a user to a server |
Willy Tarreau | eabeafa | 2008-01-16 16:17:06 +0100 | [diff] [blame] | 7851 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 7852 | no | no | yes | yes |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 7853 | |
| 7854 | Arguments : |
Willy Tarreau | be722a2 | 2014-06-13 16:31:59 +0200 | [diff] [blame] | 7855 | <pattern> is a sample expression rule as described in section 7.3. It |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 7856 | describes what elements of the incoming request or connection |
| 7857 | will be analysed in the hope to find a matching entry in a |
| 7858 | stickiness table. This rule is mandatory. |
| 7859 | |
| 7860 | <table> is an optional stickiness table name. If unspecified, the same |
| 7861 | backend's table is used. A stickiness table is declared using |
| 7862 | the "stick-table" statement. |
| 7863 | |
| 7864 | <cond> is an optional matching condition. It makes it possible to match |
| 7865 | on a certain criterion only when other conditions are met (or |
| 7866 | not met). For instance, it could be used to match on a source IP |
| 7867 | address except when a request passes through a known proxy, in |
| 7868 | which case we'd match on a header containing that IP address. |
| 7869 | |
| 7870 | Some protocols or applications require complex stickiness rules and cannot |
| 7871 | always simply rely on cookies nor hashing. The "stick match" statement |
| 7872 | describes a rule to extract the stickiness criterion from an incoming request |
| 7873 | or connection. See section 7 for a complete list of possible patterns and |
| 7874 | transformation rules. |
| 7875 | |
| 7876 | The table has to be declared using the "stick-table" statement. It must be of |
| 7877 | a type compatible with the pattern. By default it is the one which is present |
| 7878 | in the same backend. It is possible to share a table with other backends by |
| 7879 | referencing it using the "table" keyword. If another table is referenced, |
| 7880 | the server's ID inside the backends are used. By default, all server IDs |
| 7881 | start at 1 in each backend, so the server ordering is enough. But in case of |
| 7882 | doubt, it is highly recommended to force server IDs using their "id" setting. |
| 7883 | |
| 7884 | It is possible to restrict the conditions where a "stick match" statement |
| 7885 | will apply, using "if" or "unless" followed by a condition. See section 7 for |
| 7886 | ACL based conditions. |
| 7887 | |
| 7888 | There is no limit on the number of "stick match" statements. The first that |
| 7889 | applies and matches will cause the request to be directed to the same server |
| 7890 | as was used for the request which created the entry. That way, multiple |
| 7891 | matches can be used as fallbacks. |
| 7892 | |
| 7893 | The stick rules are checked after the persistence cookies, so they will not |
| 7894 | affect stickiness if a cookie has already been used to select a server. That |
| 7895 | way, it becomes very easy to insert cookies and match on IP addresses in |
| 7896 | order to maintain stickiness between HTTP and HTTPS. |
| 7897 | |
Cyril Bonté | 02ff8ef | 2010-12-14 22:48:49 +0100 | [diff] [blame] | 7898 | Note : Consider not using this feature in multi-process mode (nbproc > 1) |
| 7899 | unless you know what you do : memory is not shared between the |
| 7900 | processes, which can result in random behaviours. |
| 7901 | |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 7902 | Example : |
| 7903 | # forward SMTP users to the same server they just used for POP in the |
| 7904 | # last 30 minutes |
| 7905 | backend pop |
| 7906 | mode tcp |
| 7907 | balance roundrobin |
| 7908 | stick store-request src |
| 7909 | stick-table type ip size 200k expire 30m |
| 7910 | server s1 192.168.1.1:110 |
| 7911 | server s2 192.168.1.1:110 |
| 7912 | |
| 7913 | backend smtp |
| 7914 | mode tcp |
| 7915 | balance roundrobin |
| 7916 | stick match src table pop |
| 7917 | server s1 192.168.1.1:25 |
| 7918 | server s2 192.168.1.1:25 |
| 7919 | |
Cyril Bonté | 02ff8ef | 2010-12-14 22:48:49 +0100 | [diff] [blame] | 7920 | See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7 |
Willy Tarreau | be722a2 | 2014-06-13 16:31:59 +0200 | [diff] [blame] | 7921 | about ACLs and samples fetching. |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 7922 | |
| 7923 | |
| 7924 | stick on <pattern> [table <table>] [{if | unless} <condition>] |
| 7925 | Define a request pattern to associate a user to a server |
| 7926 | May be used in sections : defaults | frontend | listen | backend |
| 7927 | no | no | yes | yes |
| 7928 | |
| 7929 | Note : This form is exactly equivalent to "stick match" followed by |
| 7930 | "stick store-request", all with the same arguments. Please refer |
| 7931 | to both keywords for details. It is only provided as a convenience |
| 7932 | for writing more maintainable configurations. |
| 7933 | |
Cyril Bonté | 02ff8ef | 2010-12-14 22:48:49 +0100 | [diff] [blame] | 7934 | Note : Consider not using this feature in multi-process mode (nbproc > 1) |
| 7935 | unless you know what you do : memory is not shared between the |
| 7936 | processes, which can result in random behaviours. |
| 7937 | |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 7938 | Examples : |
| 7939 | # The following form ... |
Willy Tarreau | ec579d8 | 2010-02-26 19:15:04 +0100 | [diff] [blame] | 7940 | stick on src table pop if !localhost |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 7941 | |
| 7942 | # ...is strictly equivalent to this one : |
| 7943 | stick match src table pop if !localhost |
| 7944 | stick store-request src table pop if !localhost |
| 7945 | |
| 7946 | |
| 7947 | # Use cookie persistence for HTTP, and stick on source address for HTTPS as |
| 7948 | # well as HTTP without cookie. Share the same table between both accesses. |
| 7949 | backend http |
| 7950 | mode http |
| 7951 | balance roundrobin |
| 7952 | stick on src table https |
| 7953 | cookie SRV insert indirect nocache |
| 7954 | server s1 192.168.1.1:80 cookie s1 |
| 7955 | server s2 192.168.1.1:80 cookie s2 |
| 7956 | |
| 7957 | backend https |
| 7958 | mode tcp |
| 7959 | balance roundrobin |
| 7960 | stick-table type ip size 200k expire 30m |
| 7961 | stick on src |
| 7962 | server s1 192.168.1.1:443 |
| 7963 | server s2 192.168.1.1:443 |
| 7964 | |
Cyril Bonté | 02ff8ef | 2010-12-14 22:48:49 +0100 | [diff] [blame] | 7965 | See also : "stick match", "stick store-request", "nbproc" and "bind-process". |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 7966 | |
| 7967 | |
| 7968 | stick store-request <pattern> [table <table>] [{if | unless} <condition>] |
| 7969 | Define a request pattern used to create an entry in a stickiness table |
| 7970 | May be used in sections : defaults | frontend | listen | backend |
| 7971 | no | no | yes | yes |
| 7972 | |
| 7973 | Arguments : |
Willy Tarreau | be722a2 | 2014-06-13 16:31:59 +0200 | [diff] [blame] | 7974 | <pattern> is a sample expression rule as described in section 7.3. It |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 7975 | describes what elements of the incoming request or connection |
| 7976 | will be analysed, extracted and stored in the table once a |
| 7977 | server is selected. |
| 7978 | |
| 7979 | <table> is an optional stickiness table name. If unspecified, the same |
| 7980 | backend's table is used. A stickiness table is declared using |
| 7981 | the "stick-table" statement. |
| 7982 | |
| 7983 | <cond> is an optional storage condition. It makes it possible to store |
| 7984 | certain criteria only when some conditions are met (or not met). |
| 7985 | For instance, it could be used to store the source IP address |
| 7986 | except when the request passes through a known proxy, in which |
| 7987 | case we'd store a converted form of a header containing that IP |
| 7988 | address. |
| 7989 | |
| 7990 | Some protocols or applications require complex stickiness rules and cannot |
| 7991 | always simply rely on cookies nor hashing. The "stick store-request" statement |
| 7992 | describes a rule to decide what to extract from the request and when to do |
| 7993 | it, in order to store it into a stickiness table for further requests to |
| 7994 | match it using the "stick match" statement. Obviously the extracted part must |
| 7995 | make sense and have a chance to be matched in a further request. Storing a |
| 7996 | client's IP address for instance often makes sense. Storing an ID found in a |
| 7997 | URL parameter also makes sense. Storing a source port will almost never make |
| 7998 | any sense because it will be randomly matched. See section 7 for a complete |
| 7999 | list of possible patterns and transformation rules. |
| 8000 | |
| 8001 | The table has to be declared using the "stick-table" statement. It must be of |
| 8002 | a type compatible with the pattern. By default it is the one which is present |
| 8003 | in the same backend. It is possible to share a table with other backends by |
| 8004 | referencing it using the "table" keyword. If another table is referenced, |
| 8005 | the server's ID inside the backends are used. By default, all server IDs |
| 8006 | start at 1 in each backend, so the server ordering is enough. But in case of |
| 8007 | doubt, it is highly recommended to force server IDs using their "id" setting. |
| 8008 | |
| 8009 | It is possible to restrict the conditions where a "stick store-request" |
| 8010 | statement will apply, using "if" or "unless" followed by a condition. This |
| 8011 | condition will be evaluated while parsing the request, so any criteria can be |
| 8012 | used. See section 7 for ACL based conditions. |
| 8013 | |
| 8014 | There is no limit on the number of "stick store-request" statements, but |
| 8015 | there is a limit of 8 simultaneous stores per request or response. This |
| 8016 | makes it possible to store up to 8 criteria, all extracted from either the |
| 8017 | request or the response, regardless of the number of rules. Only the 8 first |
| 8018 | ones which match will be kept. Using this, it is possible to feed multiple |
| 8019 | tables at once in the hope to increase the chance to recognize a user on |
Willy Tarreau | 9667a80 | 2013-12-09 12:52:13 +0100 | [diff] [blame] | 8020 | another protocol or access method. Using multiple store-request rules with |
| 8021 | the same table is possible and may be used to find the best criterion to rely |
| 8022 | on, by arranging the rules by decreasing preference order. Only the first |
| 8023 | extracted criterion for a given table will be stored. All subsequent store- |
| 8024 | request rules referencing the same table will be skipped and their ACLs will |
| 8025 | not be evaluated. |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 8026 | |
| 8027 | The "store-request" rules are evaluated once the server connection has been |
| 8028 | established, so that the table will contain the real server that processed |
| 8029 | the request. |
| 8030 | |
Cyril Bonté | 02ff8ef | 2010-12-14 22:48:49 +0100 | [diff] [blame] | 8031 | Note : Consider not using this feature in multi-process mode (nbproc > 1) |
| 8032 | unless you know what you do : memory is not shared between the |
| 8033 | processes, which can result in random behaviours. |
| 8034 | |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 8035 | Example : |
| 8036 | # forward SMTP users to the same server they just used for POP in the |
| 8037 | # last 30 minutes |
| 8038 | backend pop |
| 8039 | mode tcp |
| 8040 | balance roundrobin |
| 8041 | stick store-request src |
| 8042 | stick-table type ip size 200k expire 30m |
| 8043 | server s1 192.168.1.1:110 |
| 8044 | server s2 192.168.1.1:110 |
| 8045 | |
| 8046 | backend smtp |
| 8047 | mode tcp |
| 8048 | balance roundrobin |
| 8049 | stick match src table pop |
| 8050 | server s1 192.168.1.1:25 |
| 8051 | server s2 192.168.1.1:25 |
| 8052 | |
Cyril Bonté | 02ff8ef | 2010-12-14 22:48:49 +0100 | [diff] [blame] | 8053 | See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7 |
Willy Tarreau | be722a2 | 2014-06-13 16:31:59 +0200 | [diff] [blame] | 8054 | about ACLs and sample fetching. |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 8055 | |
| 8056 | |
Emeric Brun | 7c6b82e | 2010-09-24 16:34:28 +0200 | [diff] [blame] | 8057 | stick-table type {ip | integer | string [len <length>] | binary [len <length>]} |
Emeric Brun | f099e79 | 2010-09-27 12:05:28 +0200 | [diff] [blame] | 8058 | size <size> [expire <expire>] [nopurge] [peers <peersect>] |
| 8059 | [store <data_type>]* |
Godbach | 64cef79 | 2013-12-04 16:08:22 +0800 | [diff] [blame] | 8060 | Configure the stickiness table for the current section |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 8061 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | c00cdc2 | 2010-06-06 16:48:26 +0200 | [diff] [blame] | 8062 | no | yes | yes | yes |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 8063 | |
| 8064 | Arguments : |
| 8065 | ip a table declared with "type ip" will only store IPv4 addresses. |
| 8066 | This form is very compact (about 50 bytes per entry) and allows |
| 8067 | very fast entry lookup and stores with almost no overhead. This |
| 8068 | is mainly used to store client source IP addresses. |
| 8069 | |
David du Colombier | 9a6d3c9 | 2011-03-17 10:40:24 +0100 | [diff] [blame] | 8070 | ipv6 a table declared with "type ipv6" will only store IPv6 addresses. |
| 8071 | This form is very compact (about 60 bytes per entry) and allows |
| 8072 | very fast entry lookup and stores with almost no overhead. This |
| 8073 | is mainly used to store client source IP addresses. |
| 8074 | |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 8075 | integer a table declared with "type integer" will store 32bit integers |
| 8076 | which can represent a client identifier found in a request for |
| 8077 | instance. |
| 8078 | |
| 8079 | string a table declared with "type string" will store substrings of up |
| 8080 | to <len> characters. If the string provided by the pattern |
| 8081 | extractor is larger than <len>, it will be truncated before |
| 8082 | being stored. During matching, at most <len> characters will be |
| 8083 | compared between the string in the table and the extracted |
| 8084 | pattern. When not specified, the string is automatically limited |
Emeric Brun | 7c6b82e | 2010-09-24 16:34:28 +0200 | [diff] [blame] | 8085 | to 32 characters. |
| 8086 | |
| 8087 | binary a table declared with "type binary" will store binary blocks |
| 8088 | of <len> bytes. If the block provided by the pattern |
| 8089 | extractor is larger than <len>, it will be truncated before |
Willy Tarreau | be722a2 | 2014-06-13 16:31:59 +0200 | [diff] [blame] | 8090 | being stored. If the block provided by the sample expression |
Emeric Brun | 7c6b82e | 2010-09-24 16:34:28 +0200 | [diff] [blame] | 8091 | is shorter than <len>, it will be padded by 0. When not |
| 8092 | specified, the block is automatically limited to 32 bytes. |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 8093 | |
| 8094 | <length> is the maximum number of characters that will be stored in a |
Emeric Brun | 7c6b82e | 2010-09-24 16:34:28 +0200 | [diff] [blame] | 8095 | "string" type table (See type "string" above). Or the number |
| 8096 | of bytes of the block in "binary" type table. Be careful when |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 8097 | changing this parameter as memory usage will proportionally |
| 8098 | increase. |
| 8099 | |
| 8100 | <size> is the maximum number of entries that can fit in the table. This |
Cyril Bonté | 78caf84 | 2010-03-10 22:41:43 +0100 | [diff] [blame] | 8101 | value directly impacts memory usage. Count approximately |
| 8102 | 50 bytes per entry, plus the size of a string if any. The size |
| 8103 | supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors. |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 8104 | |
| 8105 | [nopurge] indicates that we refuse to purge older entries when the table |
| 8106 | is full. When not specified and the table is full when haproxy |
| 8107 | wants to store an entry in it, it will flush a few of the oldest |
| 8108 | entries in order to release some space for the new ones. This is |
| 8109 | most often the desired behaviour. In some specific cases, it |
| 8110 | be desirable to refuse new entries instead of purging the older |
| 8111 | ones. That may be the case when the amount of data to store is |
| 8112 | far above the hardware limits and we prefer not to offer access |
| 8113 | to new clients than to reject the ones already connected. When |
| 8114 | using this parameter, be sure to properly set the "expire" |
| 8115 | parameter (see below). |
| 8116 | |
Emeric Brun | f099e79 | 2010-09-27 12:05:28 +0200 | [diff] [blame] | 8117 | <peersect> is the name of the peers section to use for replication. Entries |
| 8118 | which associate keys to server IDs are kept synchronized with |
| 8119 | the remote peers declared in this section. All entries are also |
| 8120 | automatically learned from the local peer (old process) during a |
| 8121 | soft restart. |
| 8122 | |
Willy Tarreau | 1abc673 | 2015-05-01 19:21:02 +0200 | [diff] [blame] | 8123 | NOTE : each peers section may be referenced only by tables |
| 8124 | belonging to the same unique process. |
Cyril Bonté | 02ff8ef | 2010-12-14 22:48:49 +0100 | [diff] [blame] | 8125 | |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 8126 | <expire> defines the maximum duration of an entry in the table since it |
| 8127 | was last created, refreshed or matched. The expiration delay is |
| 8128 | defined using the standard time format, similarly as the various |
| 8129 | timeouts. The maximum duration is slightly above 24 days. See |
| 8130 | section 2.2 for more information. If this delay is not specified, |
Cyril Bonté | dc4d903 | 2012-04-08 21:57:39 +0200 | [diff] [blame] | 8131 | the session won't automatically expire, but older entries will |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 8132 | be removed once full. Be sure not to use the "nopurge" parameter |
| 8133 | if not expiration delay is specified. |
| 8134 | |
Willy Tarreau | 08d5f98 | 2010-06-06 13:34:54 +0200 | [diff] [blame] | 8135 | <data_type> is used to store additional information in the stick-table. This |
| 8136 | may be used by ACLs in order to control various criteria related |
| 8137 | to the activity of the client matching the stick-table. For each |
| 8138 | item specified here, the size of each entry will be inflated so |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 8139 | that the additional data can fit. Several data types may be |
| 8140 | stored with an entry. Multiple data types may be specified after |
| 8141 | the "store" keyword, as a comma-separated list. Alternatively, |
| 8142 | it is possible to repeat the "store" keyword followed by one or |
| 8143 | several data types. Except for the "server_id" type which is |
| 8144 | automatically detected and enabled, all data types must be |
| 8145 | explicitly declared to be stored. If an ACL references a data |
| 8146 | type which is not stored, the ACL will simply not match. Some |
| 8147 | data types require an argument which must be passed just after |
| 8148 | the type between parenthesis. See below for the supported data |
| 8149 | types and their arguments. |
| 8150 | |
| 8151 | The data types that can be stored with an entry are the following : |
| 8152 | - server_id : this is an integer which holds the numeric ID of the server a |
| 8153 | request was assigned to. It is used by the "stick match", "stick store", |
| 8154 | and "stick on" rules. It is automatically enabled when referenced. |
| 8155 | |
| 8156 | - gpc0 : first General Purpose Counter. It is a positive 32-bit integer |
| 8157 | integer which may be used for anything. Most of the time it will be used |
| 8158 | to put a special tag on some entries, for instance to note that a |
| 8159 | specific behaviour was detected and must be known for future matches. |
| 8160 | |
Willy Tarreau | ba2ffd1 | 2013-05-29 15:54:14 +0200 | [diff] [blame] | 8161 | - gpc0_rate(<period>) : increment rate of the first General Purpose Counter |
| 8162 | over a period. It is a positive 32-bit integer integer which may be used |
| 8163 | for anything. Just like <gpc0>, it counts events, but instead of keeping |
| 8164 | a cumulative count, it maintains the rate at which the counter is |
| 8165 | incremented. Most of the time it will be used to measure the frequency of |
| 8166 | occurrence of certain events (eg: requests to a specific URL). |
| 8167 | |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 8168 | - conn_cnt : Connection Count. It is a positive 32-bit integer which counts |
| 8169 | the absolute number of connections received from clients which matched |
| 8170 | this entry. It does not mean the connections were accepted, just that |
| 8171 | they were received. |
| 8172 | |
| 8173 | - conn_cur : Current Connections. It is a positive 32-bit integer which |
| 8174 | stores the concurrent connection counts for the entry. It is incremented |
| 8175 | once an incoming connection matches the entry, and decremented once the |
| 8176 | connection leaves. That way it is possible to know at any time the exact |
| 8177 | number of concurrent connections for an entry. |
| 8178 | |
| 8179 | - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an |
| 8180 | integer parameter <period> which indicates in milliseconds the length |
| 8181 | of the period over which the average is measured. It reports the average |
| 8182 | incoming connection rate over that period, in connections per period. The |
| 8183 | result is an integer which can be matched using ACLs. |
| 8184 | |
| 8185 | - sess_cnt : Session Count. It is a positive 32-bit integer which counts |
| 8186 | the absolute number of sessions received from clients which matched this |
| 8187 | entry. A session is a connection that was accepted by the layer 4 rules. |
| 8188 | |
| 8189 | - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an |
| 8190 | integer parameter <period> which indicates in milliseconds the length |
| 8191 | of the period over which the average is measured. It reports the average |
| 8192 | incoming session rate over that period, in sessions per period. The |
| 8193 | result is an integer which can be matched using ACLs. |
| 8194 | |
| 8195 | - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which |
| 8196 | counts the absolute number of HTTP requests received from clients which |
| 8197 | matched this entry. It does not matter whether they are valid requests or |
| 8198 | not. Note that this is different from sessions when keep-alive is used on |
| 8199 | the client side. |
| 8200 | |
| 8201 | - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an |
| 8202 | integer parameter <period> which indicates in milliseconds the length |
| 8203 | of the period over which the average is measured. It reports the average |
| 8204 | HTTP request rate over that period, in requests per period. The result is |
| 8205 | an integer which can be matched using ACLs. It does not matter whether |
| 8206 | they are valid requests or not. Note that this is different from sessions |
| 8207 | when keep-alive is used on the client side. |
| 8208 | |
| 8209 | - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which |
| 8210 | counts the absolute number of HTTP requests errors induced by clients |
| 8211 | which matched this entry. Errors are counted on invalid and truncated |
| 8212 | requests, as well as on denied or tarpitted requests, and on failed |
| 8213 | authentications. If the server responds with 4xx, then the request is |
| 8214 | also counted as an error since it's an error triggered by the client |
| 8215 | (eg: vulnerability scan). |
| 8216 | |
| 8217 | - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an |
| 8218 | integer parameter <period> which indicates in milliseconds the length |
| 8219 | of the period over which the average is measured. It reports the average |
| 8220 | HTTP request error rate over that period, in requests per period (see |
| 8221 | http_err_cnt above for what is accounted as an error). The result is an |
| 8222 | integer which can be matched using ACLs. |
| 8223 | |
| 8224 | - bytes_in_cnt : client to server byte count. It is a positive 64-bit |
| 8225 | integer which counts the cumulated amount of bytes received from clients |
| 8226 | which matched this entry. Headers are included in the count. This may be |
| 8227 | used to limit abuse of upload features on photo or video servers. |
| 8228 | |
| 8229 | - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an |
| 8230 | integer parameter <period> which indicates in milliseconds the length |
| 8231 | of the period over which the average is measured. It reports the average |
| 8232 | incoming bytes rate over that period, in bytes per period. It may be used |
| 8233 | to detect users which upload too much and too fast. Warning: with large |
| 8234 | uploads, it is possible that the amount of uploaded data will be counted |
| 8235 | once upon termination, thus causing spikes in the average transfer speed |
| 8236 | instead of having a smooth one. This may partially be smoothed with |
| 8237 | "option contstats" though this is not perfect yet. Use of byte_in_cnt is |
| 8238 | recommended for better fairness. |
| 8239 | |
| 8240 | - bytes_out_cnt : server to client byte count. It is a positive 64-bit |
| 8241 | integer which counts the cumulated amount of bytes sent to clients which |
| 8242 | matched this entry. Headers are included in the count. This may be used |
| 8243 | to limit abuse of bots sucking the whole site. |
| 8244 | |
| 8245 | - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes |
| 8246 | an integer parameter <period> which indicates in milliseconds the length |
| 8247 | of the period over which the average is measured. It reports the average |
| 8248 | outgoing bytes rate over that period, in bytes per period. It may be used |
| 8249 | to detect users which download too much and too fast. Warning: with large |
| 8250 | transfers, it is possible that the amount of transferred data will be |
| 8251 | counted once upon termination, thus causing spikes in the average |
| 8252 | transfer speed instead of having a smooth one. This may partially be |
| 8253 | smoothed with "option contstats" though this is not perfect yet. Use of |
| 8254 | byte_out_cnt is recommended for better fairness. |
Willy Tarreau | 08d5f98 | 2010-06-06 13:34:54 +0200 | [diff] [blame] | 8255 | |
Willy Tarreau | c00cdc2 | 2010-06-06 16:48:26 +0200 | [diff] [blame] | 8256 | There is only one stick-table per proxy. At the moment of writing this doc, |
| 8257 | it does not seem useful to have multiple tables per proxy. If this happens |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 8258 | to be required, simply create a dummy backend with a stick-table in it and |
| 8259 | reference it. |
| 8260 | |
| 8261 | It is important to understand that stickiness based on learning information |
| 8262 | has some limitations, including the fact that all learned associations are |
Baptiste Assmann | 123ff04 | 2016-03-06 23:29:28 +0100 | [diff] [blame] | 8263 | lost upon restart unless peers are properly configured to transfer such |
| 8264 | information upon restart (recommended). In general it can be good as a |
| 8265 | complement but not always as an exclusive stickiness. |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 8266 | |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 8267 | Last, memory requirements may be important when storing many data types. |
| 8268 | Indeed, storing all indicators above at once in each entry requires 116 bytes |
| 8269 | per entry, or 116 MB for a 1-million entries table. This is definitely not |
| 8270 | something that can be ignored. |
| 8271 | |
| 8272 | Example: |
| 8273 | # Keep track of counters of up to 1 million IP addresses over 5 minutes |
| 8274 | # and store a general purpose counter and the average connection rate |
| 8275 | # computed over a sliding window of 30 seconds. |
| 8276 | stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s) |
| 8277 | |
| 8278 | See also : "stick match", "stick on", "stick store-request", section 2.2 |
David du Colombier | a13d1b9 | 2011-03-17 10:40:22 +0100 | [diff] [blame] | 8279 | about time format and section 7 about ACLs. |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 8280 | |
| 8281 | |
Emeric Brun | 6a1cefa | 2010-09-24 18:15:17 +0200 | [diff] [blame] | 8282 | stick store-response <pattern> [table <table>] [{if | unless} <condition>] |
Baptiste Assmann | 2f2d2ec | 2016-03-06 23:27:24 +0100 | [diff] [blame] | 8283 | Define a response pattern used to create an entry in a stickiness table |
Emeric Brun | 6a1cefa | 2010-09-24 18:15:17 +0200 | [diff] [blame] | 8284 | May be used in sections : defaults | frontend | listen | backend |
| 8285 | no | no | yes | yes |
| 8286 | |
| 8287 | Arguments : |
Willy Tarreau | be722a2 | 2014-06-13 16:31:59 +0200 | [diff] [blame] | 8288 | <pattern> is a sample expression rule as described in section 7.3. It |
Emeric Brun | 6a1cefa | 2010-09-24 18:15:17 +0200 | [diff] [blame] | 8289 | describes what elements of the response or connection will |
| 8290 | be analysed, extracted and stored in the table once a |
| 8291 | server is selected. |
| 8292 | |
| 8293 | <table> is an optional stickiness table name. If unspecified, the same |
| 8294 | backend's table is used. A stickiness table is declared using |
| 8295 | the "stick-table" statement. |
| 8296 | |
| 8297 | <cond> is an optional storage condition. It makes it possible to store |
| 8298 | certain criteria only when some conditions are met (or not met). |
| 8299 | For instance, it could be used to store the SSL session ID only |
| 8300 | when the response is a SSL server hello. |
| 8301 | |
| 8302 | Some protocols or applications require complex stickiness rules and cannot |
| 8303 | always simply rely on cookies nor hashing. The "stick store-response" |
| 8304 | statement describes a rule to decide what to extract from the response and |
| 8305 | when to do it, in order to store it into a stickiness table for further |
| 8306 | requests to match it using the "stick match" statement. Obviously the |
| 8307 | extracted part must make sense and have a chance to be matched in a further |
Cyril Bonté | 108cf6e | 2012-04-21 23:30:29 +0200 | [diff] [blame] | 8308 | request. Storing an ID found in a header of a response makes sense. |
Emeric Brun | 6a1cefa | 2010-09-24 18:15:17 +0200 | [diff] [blame] | 8309 | See section 7 for a complete list of possible patterns and transformation |
| 8310 | rules. |
| 8311 | |
| 8312 | The table has to be declared using the "stick-table" statement. It must be of |
| 8313 | a type compatible with the pattern. By default it is the one which is present |
| 8314 | in the same backend. It is possible to share a table with other backends by |
| 8315 | referencing it using the "table" keyword. If another table is referenced, |
| 8316 | the server's ID inside the backends are used. By default, all server IDs |
| 8317 | start at 1 in each backend, so the server ordering is enough. But in case of |
| 8318 | doubt, it is highly recommended to force server IDs using their "id" setting. |
| 8319 | |
| 8320 | It is possible to restrict the conditions where a "stick store-response" |
| 8321 | statement will apply, using "if" or "unless" followed by a condition. This |
| 8322 | condition will be evaluated while parsing the response, so any criteria can |
| 8323 | be used. See section 7 for ACL based conditions. |
| 8324 | |
| 8325 | There is no limit on the number of "stick store-response" statements, but |
| 8326 | there is a limit of 8 simultaneous stores per request or response. This |
| 8327 | makes it possible to store up to 8 criteria, all extracted from either the |
| 8328 | request or the response, regardless of the number of rules. Only the 8 first |
| 8329 | ones which match will be kept. Using this, it is possible to feed multiple |
| 8330 | tables at once in the hope to increase the chance to recognize a user on |
Willy Tarreau | 9667a80 | 2013-12-09 12:52:13 +0100 | [diff] [blame] | 8331 | another protocol or access method. Using multiple store-response rules with |
| 8332 | the same table is possible and may be used to find the best criterion to rely |
| 8333 | on, by arranging the rules by decreasing preference order. Only the first |
| 8334 | extracted criterion for a given table will be stored. All subsequent store- |
| 8335 | response rules referencing the same table will be skipped and their ACLs will |
| 8336 | not be evaluated. However, even if a store-request rule references a table, a |
| 8337 | store-response rule may also use the same table. This means that each table |
| 8338 | may learn exactly one element from the request and one element from the |
| 8339 | response at once. |
Emeric Brun | 6a1cefa | 2010-09-24 18:15:17 +0200 | [diff] [blame] | 8340 | |
| 8341 | The table will contain the real server that processed the request. |
| 8342 | |
| 8343 | Example : |
| 8344 | # Learn SSL session ID from both request and response and create affinity. |
| 8345 | backend https |
| 8346 | mode tcp |
| 8347 | balance roundrobin |
Cyril Bonté | dc4d903 | 2012-04-08 21:57:39 +0200 | [diff] [blame] | 8348 | # maximum SSL session ID length is 32 bytes. |
Emeric Brun | 6a1cefa | 2010-09-24 18:15:17 +0200 | [diff] [blame] | 8349 | stick-table type binary len 32 size 30k expire 30m |
Cyril Bonté | 108cf6e | 2012-04-21 23:30:29 +0200 | [diff] [blame] | 8350 | |
Emeric Brun | 6a1cefa | 2010-09-24 18:15:17 +0200 | [diff] [blame] | 8351 | acl clienthello req_ssl_hello_type 1 |
| 8352 | acl serverhello rep_ssl_hello_type 2 |
| 8353 | |
| 8354 | # use tcp content accepts to detects ssl client and server hello. |
| 8355 | tcp-request inspect-delay 5s |
| 8356 | tcp-request content accept if clienthello |
| 8357 | |
| 8358 | # no timeout on response inspect delay by default. |
| 8359 | tcp-response content accept if serverhello |
Cyril Bonté | 108cf6e | 2012-04-21 23:30:29 +0200 | [diff] [blame] | 8360 | |
Emeric Brun | 6a1cefa | 2010-09-24 18:15:17 +0200 | [diff] [blame] | 8361 | # SSL session ID (SSLID) may be present on a client or server hello. |
| 8362 | # Its length is coded on 1 byte at offset 43 and its value starts |
| 8363 | # at offset 44. |
| 8364 | |
| 8365 | # Match and learn on request if client hello. |
| 8366 | stick on payload_lv(43,1) if clienthello |
| 8367 | |
| 8368 | # Learn on response if server hello. |
| 8369 | stick store-response payload_lv(43,1) if serverhello |
Cyril Bonté | dc4d903 | 2012-04-08 21:57:39 +0200 | [diff] [blame] | 8370 | |
Emeric Brun | 6a1cefa | 2010-09-24 18:15:17 +0200 | [diff] [blame] | 8371 | server s1 192.168.1.1:443 |
| 8372 | server s2 192.168.1.1:443 |
| 8373 | |
| 8374 | See also : "stick-table", "stick on", and section 7 about ACLs and pattern |
| 8375 | extraction. |
| 8376 | |
| 8377 | |
Willy Tarreau | 938c7fe | 2014-04-25 14:21:39 +0200 | [diff] [blame] | 8378 | tcp-check connect [params*] |
| 8379 | Opens a new connection |
| 8380 | May be used in sections: defaults | frontend | listen | backend |
| 8381 | no | no | yes | yes |
| 8382 | |
| 8383 | When an application lies on more than a single TCP port or when HAProxy |
| 8384 | load-balance many services in a single backend, it makes sense to probe all |
| 8385 | the services individually before considering a server as operational. |
| 8386 | |
| 8387 | When there are no TCP port configured on the server line neither server port |
| 8388 | directive, then the 'tcp-check connect port <port>' must be the first step |
| 8389 | of the sequence. |
| 8390 | |
| 8391 | In a tcp-check ruleset a 'connect' is required, it is also mandatory to start |
| 8392 | the ruleset with a 'connect' rule. Purpose is to ensure admin know what they |
| 8393 | do. |
| 8394 | |
| 8395 | Parameters : |
| 8396 | They are optional and can be used to describe how HAProxy should open and |
| 8397 | use the TCP connection. |
| 8398 | |
| 8399 | port if not set, check port or server port is used. |
| 8400 | It tells HAProxy where to open the connection to. |
| 8401 | <port> must be a valid TCP port source integer, from 1 to 65535. |
| 8402 | |
| 8403 | send-proxy send a PROXY protocol string |
| 8404 | |
| 8405 | ssl opens a ciphered connection |
| 8406 | |
| 8407 | Examples: |
| 8408 | # check HTTP and HTTPs services on a server. |
| 8409 | # first open port 80 thanks to server line port directive, then |
| 8410 | # tcp-check opens port 443, ciphered and run a request on it: |
| 8411 | option tcp-check |
| 8412 | tcp-check connect |
| 8413 | tcp-check send GET\ /\ HTTP/1.0\r\n |
| 8414 | tcp-check send Host:\ haproxy.1wt.eu\r\n |
| 8415 | tcp-check send \r\n |
| 8416 | tcp-check expect rstring (2..|3..) |
| 8417 | tcp-check connect port 443 ssl |
| 8418 | tcp-check send GET\ /\ HTTP/1.0\r\n |
| 8419 | tcp-check send Host:\ haproxy.1wt.eu\r\n |
| 8420 | tcp-check send \r\n |
| 8421 | tcp-check expect rstring (2..|3..) |
| 8422 | server www 10.0.0.1 check port 80 |
| 8423 | |
| 8424 | # check both POP and IMAP from a single server: |
| 8425 | option tcp-check |
| 8426 | tcp-check connect port 110 |
| 8427 | tcp-check expect string +OK\ POP3\ ready |
| 8428 | tcp-check connect port 143 |
| 8429 | tcp-check expect string *\ OK\ IMAP4\ ready |
| 8430 | server mail 10.0.0.1 check |
| 8431 | |
| 8432 | See also : "option tcp-check", "tcp-check send", "tcp-check expect" |
| 8433 | |
| 8434 | |
| 8435 | tcp-check expect [!] <match> <pattern> |
| 8436 | Specify data to be collected and analysed during a generic health check |
| 8437 | May be used in sections: defaults | frontend | listen | backend |
| 8438 | no | no | yes | yes |
| 8439 | |
| 8440 | Arguments : |
| 8441 | <match> is a keyword indicating how to look for a specific pattern in the |
| 8442 | response. The keyword may be one of "string", "rstring" or |
| 8443 | binary. |
| 8444 | The keyword may be preceded by an exclamation mark ("!") to negate |
| 8445 | the match. Spaces are allowed between the exclamation mark and the |
| 8446 | keyword. See below for more details on the supported keywords. |
| 8447 | |
| 8448 | <pattern> is the pattern to look for. It may be a string or a regular |
| 8449 | expression. If the pattern contains spaces, they must be escaped |
| 8450 | with the usual backslash ('\'). |
| 8451 | If the match is set to binary, then the pattern must be passed as |
| 8452 | a serie of hexadecimal digits in an even number. Each sequence of |
| 8453 | two digits will represent a byte. The hexadecimal digits may be |
| 8454 | used upper or lower case. |
| 8455 | |
| 8456 | |
| 8457 | The available matches are intentionally similar to their http-check cousins : |
| 8458 | |
| 8459 | string <string> : test the exact string matches in the response buffer. |
| 8460 | A health check response will be considered valid if the |
| 8461 | response's buffer contains this exact string. If the |
| 8462 | "string" keyword is prefixed with "!", then the response |
| 8463 | will be considered invalid if the body contains this |
| 8464 | string. This can be used to look for a mandatory pattern |
| 8465 | in a protocol response, or to detect a failure when a |
| 8466 | specific error appears in a protocol banner. |
| 8467 | |
| 8468 | rstring <regex> : test a regular expression on the response buffer. |
| 8469 | A health check response will be considered valid if the |
| 8470 | response's buffer matches this expression. If the |
| 8471 | "rstring" keyword is prefixed with "!", then the response |
| 8472 | will be considered invalid if the body matches the |
| 8473 | expression. |
| 8474 | |
| 8475 | binary <hexstring> : test the exact string in its hexadecimal form matches |
| 8476 | in the response buffer. A health check response will |
| 8477 | be considered valid if the response's buffer contains |
| 8478 | this exact hexadecimal string. |
| 8479 | Purpose is to match data on binary protocols. |
| 8480 | |
| 8481 | It is important to note that the responses will be limited to a certain size |
| 8482 | defined by the global "tune.chksize" option, which defaults to 16384 bytes. |
| 8483 | Thus, too large responses may not contain the mandatory pattern when using |
| 8484 | "string", "rstring" or binary. If a large response is absolutely required, it |
| 8485 | is possible to change the default max size by setting the global variable. |
| 8486 | However, it is worth keeping in mind that parsing very large responses can |
| 8487 | waste some CPU cycles, especially when regular expressions are used, and that |
| 8488 | it is always better to focus the checks on smaller resources. Also, in its |
| 8489 | current state, the check will not find any string nor regex past a null |
| 8490 | character in the response. Similarly it is not possible to request matching |
| 8491 | the null character. |
| 8492 | |
| 8493 | Examples : |
| 8494 | # perform a POP check |
| 8495 | option tcp-check |
| 8496 | tcp-check expect string +OK\ POP3\ ready |
| 8497 | |
| 8498 | # perform an IMAP check |
| 8499 | option tcp-check |
| 8500 | tcp-check expect string *\ OK\ IMAP4\ ready |
| 8501 | |
| 8502 | # look for the redis master server |
| 8503 | option tcp-check |
| 8504 | tcp-check send PING\r\n |
Baptiste Assmann | a332299 | 2015-08-04 10:12:18 +0200 | [diff] [blame] | 8505 | tcp-check expect string +PONG |
Willy Tarreau | 938c7fe | 2014-04-25 14:21:39 +0200 | [diff] [blame] | 8506 | tcp-check send info\ replication\r\n |
| 8507 | tcp-check expect string role:master |
| 8508 | tcp-check send QUIT\r\n |
| 8509 | tcp-check expect string +OK |
| 8510 | |
| 8511 | |
| 8512 | See also : "option tcp-check", "tcp-check connect", "tcp-check send", |
| 8513 | "tcp-check send-binary", "http-check expect", tune.chksize |
| 8514 | |
| 8515 | |
| 8516 | tcp-check send <data> |
| 8517 | Specify a string to be sent as a question during a generic health check |
| 8518 | May be used in sections: defaults | frontend | listen | backend |
| 8519 | no | no | yes | yes |
| 8520 | |
| 8521 | <data> : the data to be sent as a question during a generic health check |
| 8522 | session. For now, <data> must be a string. |
| 8523 | |
| 8524 | Examples : |
| 8525 | # look for the redis master server |
| 8526 | option tcp-check |
| 8527 | tcp-check send info\ replication\r\n |
| 8528 | tcp-check expect string role:master |
| 8529 | |
| 8530 | See also : "option tcp-check", "tcp-check connect", "tcp-check expect", |
| 8531 | "tcp-check send-binary", tune.chksize |
| 8532 | |
| 8533 | |
| 8534 | tcp-check send-binary <hexastring> |
| 8535 | Specify an hexa digits string to be sent as a binary question during a raw |
| 8536 | tcp health check |
| 8537 | May be used in sections: defaults | frontend | listen | backend |
| 8538 | no | no | yes | yes |
| 8539 | |
| 8540 | <data> : the data to be sent as a question during a generic health check |
| 8541 | session. For now, <data> must be a string. |
| 8542 | <hexastring> : test the exact string in its hexadecimal form matches in the |
| 8543 | response buffer. A health check response will be considered |
| 8544 | valid if the response's buffer contains this exact |
| 8545 | hexadecimal string. |
| 8546 | Purpose is to send binary data to ask on binary protocols. |
| 8547 | |
| 8548 | Examples : |
| 8549 | # redis check in binary |
| 8550 | option tcp-check |
| 8551 | tcp-check send-binary 50494e470d0a # PING\r\n |
| 8552 | tcp-check expect binary 2b504F4e47 # +PONG |
| 8553 | |
| 8554 | |
| 8555 | See also : "option tcp-check", "tcp-check connect", "tcp-check expect", |
| 8556 | "tcp-check send", tune.chksize |
| 8557 | |
| 8558 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8559 | tcp-request connection <action> [{if | unless} <condition>] |
| 8560 | Perform an action on an incoming connection depending on a layer 4 condition |
Willy Tarreau | 1a68794 | 2010-05-23 22:40:30 +0200 | [diff] [blame] | 8561 | May be used in sections : defaults | frontend | listen | backend |
| 8562 | no | yes | yes | no |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8563 | Arguments : |
Willy Tarreau | c870bfd | 2015-09-28 18:47:38 +0200 | [diff] [blame] | 8564 | <action> defines the action to perform if the condition applies. See |
| 8565 | below. |
Willy Tarreau | 1a68794 | 2010-05-23 22:40:30 +0200 | [diff] [blame] | 8566 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8567 | <condition> is a standard layer4-only ACL-based condition (see section 7). |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8568 | |
| 8569 | Immediately after acceptance of a new incoming connection, it is possible to |
| 8570 | evaluate some conditions to decide whether this connection must be accepted |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8571 | or dropped or have its counters tracked. Those conditions cannot make use of |
| 8572 | any data contents because the connection has not been read from yet, and the |
| 8573 | buffers are not yet allocated. This is used to selectively and very quickly |
| 8574 | accept or drop connections from various sources with a very low overhead. If |
| 8575 | some contents need to be inspected in order to take the decision, the |
| 8576 | "tcp-request content" statements must be used instead. |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8577 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8578 | The "tcp-request connection" rules are evaluated in their exact declaration |
| 8579 | order. If no rule matches or if there is no rule, the default action is to |
| 8580 | accept the incoming connection. There is no specific limit to the number of |
| 8581 | rules which may be inserted. |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8582 | |
Willy Tarreau | a9083d0 | 2015-05-08 15:27:59 +0200 | [diff] [blame] | 8583 | Four types of actions are supported : |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8584 | - accept : |
| 8585 | accepts the connection if the condition is true (when used with "if") |
| 8586 | or false (when used with "unless"). The first such rule executed ends |
| 8587 | the rules evaluation. |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8588 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8589 | - reject : |
| 8590 | rejects the connection if the condition is true (when used with "if") |
| 8591 | or false (when used with "unless"). The first such rule executed ends |
| 8592 | the rules evaluation. Rejected connections do not even become a |
| 8593 | session, which is why they are accounted separately for in the stats, |
| 8594 | as "denied connections". They are not considered for the session |
| 8595 | rate-limit and are not logged either. The reason is that these rules |
| 8596 | should only be used to filter extremely high connection rates such as |
| 8597 | the ones encountered during a massive DDoS attack. Under these extreme |
| 8598 | conditions, the simple action of logging each event would make the |
| 8599 | system collapse and would considerably lower the filtering capacity. If |
| 8600 | logging is absolutely desired, then "tcp-request content" rules should |
| 8601 | be used instead. |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8602 | |
Willy Tarreau | 4f0d919 | 2013-06-11 20:40:55 +0200 | [diff] [blame] | 8603 | - expect-proxy layer4 : |
| 8604 | configures the client-facing connection to receive a PROXY protocol |
| 8605 | header before any byte is read from the socket. This is equivalent to |
| 8606 | having the "accept-proxy" keyword on the "bind" line, except that using |
| 8607 | the TCP rule allows the PROXY protocol to be accepted only for certain |
| 8608 | IP address ranges using an ACL. This is convenient when multiple layers |
| 8609 | of load balancers are passed through by traffic coming from public |
| 8610 | hosts. |
| 8611 | |
Willy Tarreau | 18bf01e | 2014-06-13 16:18:52 +0200 | [diff] [blame] | 8612 | - capture <sample> len <length> : |
| 8613 | This only applies to "tcp-request content" rules. It captures sample |
| 8614 | expression <sample> from the request buffer, and converts it to a |
| 8615 | string of at most <len> characters. The resulting string is stored into |
| 8616 | the next request "capture" slot, so it will possibly appear next to |
| 8617 | some captured HTTP headers. It will then automatically appear in the |
| 8618 | logs, and it will be possible to extract it using sample fetch rules to |
| 8619 | feed it into headers or anything. The length should be limited given |
| 8620 | that this size will be allocated for each capture during the whole |
Willy Tarreau | a9083d0 | 2015-05-08 15:27:59 +0200 | [diff] [blame] | 8621 | session life. Please check section 7.3 (Fetching samples) and "capture |
| 8622 | request header" for more information. |
Willy Tarreau | 18bf01e | 2014-06-13 16:18:52 +0200 | [diff] [blame] | 8623 | |
Willy Tarreau | be4a3ef | 2013-06-17 15:04:07 +0200 | [diff] [blame] | 8624 | - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] : |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8625 | enables tracking of sticky counters from current connection. These |
Willy Tarreau | 09448f7 | 2014-06-25 18:12:15 +0200 | [diff] [blame] | 8626 | rules do not stop evaluation and do not change default action. 3 sets |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8627 | of counters may be simultaneously tracked by the same connection. The |
Willy Tarreau | be4a3ef | 2013-06-17 15:04:07 +0200 | [diff] [blame] | 8628 | first "track-sc0" rule executed enables tracking of the counters of the |
| 8629 | specified table as the first set. The first "track-sc1" rule executed |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8630 | enables tracking of the counters of the specified table as the second |
Willy Tarreau | be4a3ef | 2013-06-17 15:04:07 +0200 | [diff] [blame] | 8631 | set. The first "track-sc2" rule executed enables tracking of the |
| 8632 | counters of the specified table as the third set. It is a recommended |
| 8633 | practice to use the first set of counters for the per-frontend counters |
| 8634 | and the second set for the per-backend ones. But this is just a |
| 8635 | guideline, all may be used everywhere. |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8636 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8637 | These actions take one or two arguments : |
Willy Tarreau | be722a2 | 2014-06-13 16:31:59 +0200 | [diff] [blame] | 8638 | <key> is mandatory, and is a sample expression rule as described |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 8639 | in section 7.3. It describes what elements of the incoming |
Willy Tarreau | 5d5b5d8 | 2012-12-09 12:00:04 +0100 | [diff] [blame] | 8640 | request or connection will be analysed, extracted, combined, |
| 8641 | and used to select which table entry to update the counters. |
| 8642 | Note that "tcp-request connection" cannot use content-based |
| 8643 | fetches. |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8644 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8645 | <table> is an optional table to be used instead of the default one, |
| 8646 | which is the stick-table declared in the current proxy. All |
| 8647 | the counters for the matches and updates for the key will |
| 8648 | then be performed in that table until the session ends. |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8649 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8650 | Once a "track-sc*" rule is executed, the key is looked up in the table |
| 8651 | and if it is not found, an entry is allocated for it. Then a pointer to |
| 8652 | that entry is kept during all the session's life, and this entry's |
| 8653 | counters are updated as often as possible, every time the session's |
| 8654 | counters are updated, and also systematically when the session ends. |
Willy Tarreau | 5d5b5d8 | 2012-12-09 12:00:04 +0100 | [diff] [blame] | 8655 | Counters are only updated for events that happen after the tracking has |
| 8656 | been started. For example, connection counters will not be updated when |
| 8657 | tracking layer 7 information, since the connection event happens before |
| 8658 | layer7 information is extracted. |
| 8659 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8660 | If the entry tracks concurrent connection counters, one connection is |
| 8661 | counted for as long as the entry is tracked, and the entry will not |
| 8662 | expire during that time. Tracking counters also provides a performance |
| 8663 | advantage over just checking the keys, because only one table lookup is |
| 8664 | performed for all ACL checks that make use of it. |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8665 | |
Thierry FOURNIER | e0627bd | 2015-08-04 08:20:33 +0200 | [diff] [blame] | 8666 | - sc-inc-gpc0(<sc-id>): |
| 8667 | The "sc-inc-gpc0" increments the GPC0 counter according to the sticky |
| 8668 | counter designated by <sc-id>. If an error occurs, this action silently |
| 8669 | fails and the actions evaluation continues. |
| 8670 | |
Thierry FOURNIER | 236657b | 2015-08-19 08:25:14 +0200 | [diff] [blame] | 8671 | - sc-set-gpt0(<sc-id>) <int>: |
| 8672 | This action sets the GPT0 tag according to the sticky counter designated |
| 8673 | by <sc-id> and the value of <int>. The expected result is a boolean. If |
| 8674 | an error occurs, this action silently fails and the actions evaluation |
| 8675 | continues. |
| 8676 | |
Willy Tarreau | 2d392c2 | 2015-08-24 01:43:45 +0200 | [diff] [blame] | 8677 | - "silent-drop" : |
| 8678 | This stops the evaluation of the rules and makes the client-facing |
| 8679 | connection suddenly disappear using a system-dependant way that tries |
| 8680 | to prevent the client from being notified. The effect it then that the |
| 8681 | client still sees an established connection while there's none on |
| 8682 | HAProxy. The purpose is to achieve a comparable effect to "tarpit" |
| 8683 | except that it doesn't use any local resource at all on the machine |
| 8684 | running HAProxy. It can resist much higher loads than "tarpit", and |
| 8685 | slow down stronger attackers. It is important to undestand the impact |
| 8686 | of using this mechanism. All stateful equipments placed between the |
| 8687 | client and HAProxy (firewalls, proxies, load balancers) will also keep |
| 8688 | the established connection for a long time and may suffer from this |
| 8689 | action. On modern Linux systems running with enough privileges, the |
| 8690 | TCP_REPAIR socket option is used to block the emission of a TCP |
| 8691 | reset. On other systems, the socket's TTL is reduced to 1 so that the |
| 8692 | TCP reset doesn't pass the first router, though it's still delivered to |
| 8693 | local networks. Do not use it unless you fully understand how it works. |
| 8694 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8695 | Note that the "if/unless" condition is optional. If no condition is set on |
| 8696 | the action, it is simply performed unconditionally. That can be useful for |
| 8697 | "track-sc*" actions as well as for changing the default action to a reject. |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8698 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8699 | Example: accept all connections from white-listed hosts, reject too fast |
| 8700 | connection without counting them, and track accepted connections. |
| 8701 | This results in connection rate being capped from abusive sources. |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8702 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8703 | tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst } |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8704 | tcp-request connection reject if { src_conn_rate gt 10 } |
Willy Tarreau | be4a3ef | 2013-06-17 15:04:07 +0200 | [diff] [blame] | 8705 | tcp-request connection track-sc0 src |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8706 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8707 | Example: accept all connections from white-listed hosts, count all other |
| 8708 | connections and reject too fast ones. This results in abusive ones |
| 8709 | being blocked as long as they don't slow down. |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8710 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8711 | tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst } |
Willy Tarreau | be4a3ef | 2013-06-17 15:04:07 +0200 | [diff] [blame] | 8712 | tcp-request connection track-sc0 src |
| 8713 | tcp-request connection reject if { sc0_conn_rate gt 10 } |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8714 | |
Willy Tarreau | 4f0d919 | 2013-06-11 20:40:55 +0200 | [diff] [blame] | 8715 | Example: enable the PROXY protocol for traffic coming from all known proxies. |
| 8716 | |
| 8717 | tcp-request connection expect-proxy layer4 if { src -f proxies.lst } |
| 8718 | |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8719 | See section 7 about ACL usage. |
| 8720 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8721 | See also : "tcp-request content", "stick-table" |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8722 | |
| 8723 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8724 | tcp-request content <action> [{if | unless} <condition>] |
| 8725 | Perform an action on a new session depending on a layer 4-7 condition |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8726 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | fb35620 | 2010-08-03 14:02:05 +0200 | [diff] [blame] | 8727 | no | yes | yes | yes |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8728 | Arguments : |
Willy Tarreau | c870bfd | 2015-09-28 18:47:38 +0200 | [diff] [blame] | 8729 | <action> defines the action to perform if the condition applies. See |
| 8730 | below. |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8731 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8732 | <condition> is a standard layer 4-7 ACL-based condition (see section 7). |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8733 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8734 | A request's contents can be analysed at an early stage of request processing |
| 8735 | called "TCP content inspection". During this stage, ACL-based rules are |
| 8736 | evaluated every time the request contents are updated, until either an |
| 8737 | "accept" or a "reject" rule matches, or the TCP request inspection delay |
| 8738 | expires with no matching rule. |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8739 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8740 | The first difference between these rules and "tcp-request connection" rules |
| 8741 | is that "tcp-request content" rules can make use of contents to take a |
| 8742 | decision. Most often, these decisions will consider a protocol recognition or |
| 8743 | validity. The second difference is that content-based rules can be used in |
Willy Tarreau | f333834 | 2014-01-28 21:40:28 +0100 | [diff] [blame] | 8744 | both frontends and backends. In case of HTTP keep-alive with the client, all |
| 8745 | tcp-request content rules are evaluated again, so haproxy keeps a record of |
| 8746 | what sticky counters were assigned by a "tcp-request connection" versus a |
| 8747 | "tcp-request content" rule, and flushes all the content-related ones after |
| 8748 | processing an HTTP request, so that they may be evaluated again by the rules |
| 8749 | being evaluated again for the next request. This is of particular importance |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 8750 | when the rule tracks some L7 information or when it is conditioned by an |
Willy Tarreau | f333834 | 2014-01-28 21:40:28 +0100 | [diff] [blame] | 8751 | L7-based ACL, since tracking may change between requests. |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8752 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8753 | Content-based rules are evaluated in their exact declaration order. If no |
| 8754 | rule matches or if there is no rule, the default action is to accept the |
| 8755 | contents. There is no specific limit to the number of rules which may be |
| 8756 | inserted. |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8757 | |
Thierry FOURNIER | 236657b | 2015-08-19 08:25:14 +0200 | [diff] [blame] | 8758 | Several types of actions are supported : |
Willy Tarreau | 18bf01e | 2014-06-13 16:18:52 +0200 | [diff] [blame] | 8759 | - accept : the request is accepted |
| 8760 | - reject : the request is rejected and the connection is closed |
| 8761 | - capture : the specified sample expression is captured |
Willy Tarreau | be4a3ef | 2013-06-17 15:04:07 +0200 | [diff] [blame] | 8762 | - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] |
Thierry FOURNIER | e0627bd | 2015-08-04 08:20:33 +0200 | [diff] [blame] | 8763 | - sc-inc-gpc0(<sc-id>) |
Thierry Fournier | b912567 | 2016-03-29 19:34:37 +0200 | [diff] [blame] | 8764 | - sc-set-gpt0(<sc-id>) <int> |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 8765 | - set-var(<var-name>) <expr> |
Willy Tarreau | 2d392c2 | 2015-08-24 01:43:45 +0200 | [diff] [blame] | 8766 | - silent-drop |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8767 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8768 | They have the same meaning as their counter-parts in "tcp-request connection" |
| 8769 | so please refer to that section for a complete description. |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8770 | |
Willy Tarreau | f333834 | 2014-01-28 21:40:28 +0100 | [diff] [blame] | 8771 | While there is nothing mandatory about it, it is recommended to use the |
| 8772 | track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request |
| 8773 | content" rules in the frontend, and track-sc2 for "tcp-request content" |
| 8774 | rules in the backend, because that makes the configuration more readable |
| 8775 | and easier to troubleshoot, but this is just a guideline and all counters |
| 8776 | may be used everywhere. |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8777 | |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 8778 | Note that the "if/unless" condition is optional. If no condition is set on |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8779 | the action, it is simply performed unconditionally. That can be useful for |
| 8780 | "track-sc*" actions as well as for changing the default action to a reject. |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8781 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8782 | It is perfectly possible to match layer 7 contents with "tcp-request content" |
Willy Tarreau | c0239e0 | 2012-04-16 14:42:55 +0200 | [diff] [blame] | 8783 | rules, since HTTP-specific ACL matches are able to preliminarily parse the |
| 8784 | contents of a buffer before extracting the required data. If the buffered |
| 8785 | contents do not parse as a valid HTTP message, then the ACL does not match. |
| 8786 | The parser which is involved there is exactly the same as for all other HTTP |
Willy Tarreau | f333834 | 2014-01-28 21:40:28 +0100 | [diff] [blame] | 8787 | processing, so there is no risk of parsing something differently. In an HTTP |
| 8788 | backend connected to from an HTTP frontend, it is guaranteed that HTTP |
| 8789 | contents will always be immediately present when the rule is evaluated first. |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8790 | |
Willy Tarreau | 5d5b5d8 | 2012-12-09 12:00:04 +0100 | [diff] [blame] | 8791 | Tracking layer7 information is also possible provided that the information |
Willy Tarreau | 4d54c7c | 2014-09-16 15:48:15 +0200 | [diff] [blame] | 8792 | are present when the rule is processed. The rule processing engine is able to |
| 8793 | wait until the inspect delay expires when the data to be tracked is not yet |
| 8794 | available. |
Willy Tarreau | 5d5b5d8 | 2012-12-09 12:00:04 +0100 | [diff] [blame] | 8795 | |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 8796 | The "set-var" is used to set the content of a variable. The variable is |
| 8797 | declared inline. |
| 8798 | |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 8799 | <var-name> The name of the variable starts with an indication about |
| 8800 | its scope. The scopes allowed are: |
| 8801 | "sess" : the variable is shared with the whole session |
| 8802 | "txn" : the variable is shared with the transaction |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 8803 | (request and response) |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 8804 | "req" : the variable is shared only during request |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 8805 | processing |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 8806 | "res" : the variable is shared only during response |
| 8807 | processing |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 8808 | This prefix is followed by a name. The separator is a '.'. |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 8809 | The name may only contain characters 'a-z', 'A-Z', '0-9' |
| 8810 | and '_'. |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 8811 | |
| 8812 | <expr> Is a standard HAProxy expression formed by a sample-fetch |
| 8813 | followed by some converters. |
| 8814 | |
| 8815 | Example: |
| 8816 | |
| 8817 | tcp-request content set-var(sess.my_var) src |
| 8818 | |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8819 | Example: |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8820 | # Accept HTTP requests containing a Host header saying "example.com" |
| 8821 | # and reject everything else. |
| 8822 | acl is_host_com hdr(Host) -i example.com |
| 8823 | tcp-request inspect-delay 30s |
Willy Tarreau | c0239e0 | 2012-04-16 14:42:55 +0200 | [diff] [blame] | 8824 | tcp-request content accept if is_host_com |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8825 | tcp-request content reject |
| 8826 | |
| 8827 | Example: |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8828 | # reject SMTP connection if client speaks first |
| 8829 | tcp-request inspect-delay 30s |
| 8830 | acl content_present req_len gt 0 |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8831 | tcp-request content reject if content_present |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8832 | |
| 8833 | # Forward HTTPS connection only if client speaks |
| 8834 | tcp-request inspect-delay 30s |
| 8835 | acl content_present req_len gt 0 |
Willy Tarreau | 68c03ab | 2010-08-06 15:08:45 +0200 | [diff] [blame] | 8836 | tcp-request content accept if content_present |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8837 | tcp-request content reject |
| 8838 | |
Willy Tarreau | 5d5b5d8 | 2012-12-09 12:00:04 +0100 | [diff] [blame] | 8839 | Example: |
| 8840 | # Track the last IP from X-Forwarded-For |
| 8841 | tcp-request inspect-delay 10s |
Willy Tarreau | 4d54c7c | 2014-09-16 15:48:15 +0200 | [diff] [blame] | 8842 | tcp-request content track-sc0 hdr(x-forwarded-for,-1) |
Willy Tarreau | 5d5b5d8 | 2012-12-09 12:00:04 +0100 | [diff] [blame] | 8843 | |
| 8844 | Example: |
| 8845 | # track request counts per "base" (concatenation of Host+URL) |
| 8846 | tcp-request inspect-delay 10s |
Willy Tarreau | 4d54c7c | 2014-09-16 15:48:15 +0200 | [diff] [blame] | 8847 | tcp-request content track-sc0 base table req-rate |
Willy Tarreau | 5d5b5d8 | 2012-12-09 12:00:04 +0100 | [diff] [blame] | 8848 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8849 | Example: track per-frontend and per-backend counters, block abusers at the |
| 8850 | frontend when the backend detects abuse. |
| 8851 | |
| 8852 | frontend http |
Willy Tarreau | be4a3ef | 2013-06-17 15:04:07 +0200 | [diff] [blame] | 8853 | # Use General Purpose Couter 0 in SC0 as a global abuse counter |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8854 | # protecting all our sites |
| 8855 | stick-table type ip size 1m expire 5m store gpc0 |
Willy Tarreau | be4a3ef | 2013-06-17 15:04:07 +0200 | [diff] [blame] | 8856 | tcp-request connection track-sc0 src |
| 8857 | tcp-request connection reject if { sc0_get_gpc0 gt 0 } |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8858 | ... |
| 8859 | use_backend http_dynamic if { path_end .php } |
| 8860 | |
| 8861 | backend http_dynamic |
| 8862 | # if a source makes too fast requests to this dynamic site (tracked |
Willy Tarreau | be4a3ef | 2013-06-17 15:04:07 +0200 | [diff] [blame] | 8863 | # by SC1), block it globally in the frontend. |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8864 | stick-table type ip size 1m expire 5m store http_req_rate(10s) |
Willy Tarreau | be4a3ef | 2013-06-17 15:04:07 +0200 | [diff] [blame] | 8865 | acl click_too_fast sc1_http_req_rate gt 10 |
| 8866 | acl mark_as_abuser sc0_inc_gpc0 gt 0 |
| 8867 | tcp-request content track-sc1 src |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8868 | tcp-request content reject if click_too_fast mark_as_abuser |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8869 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 8870 | See section 7 about ACL usage. |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8871 | |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 8872 | See also : "tcp-request connection", "tcp-request inspect-delay" |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8873 | |
| 8874 | |
| 8875 | tcp-request inspect-delay <timeout> |
| 8876 | Set the maximum allowed time to wait for data during content inspection |
| 8877 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | fb35620 | 2010-08-03 14:02:05 +0200 | [diff] [blame] | 8878 | no | yes | yes | yes |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8879 | Arguments : |
| 8880 | <timeout> is the timeout value specified in milliseconds by default, but |
| 8881 | can be in any other unit if the number is suffixed by the unit, |
| 8882 | as explained at the top of this document. |
| 8883 | |
| 8884 | People using haproxy primarily as a TCP relay are often worried about the |
| 8885 | risk of passing any type of protocol to a server without any analysis. In |
| 8886 | order to be able to analyze the request contents, we must first withhold |
| 8887 | the data then analyze them. This statement simply enables withholding of |
| 8888 | data for at most the specified amount of time. |
| 8889 | |
Willy Tarreau | fb35620 | 2010-08-03 14:02:05 +0200 | [diff] [blame] | 8890 | TCP content inspection applies very early when a connection reaches a |
| 8891 | frontend, then very early when the connection is forwarded to a backend. This |
| 8892 | means that a connection may experience a first delay in the frontend and a |
| 8893 | second delay in the backend if both have tcp-request rules. |
| 8894 | |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8895 | Note that when performing content inspection, haproxy will evaluate the whole |
| 8896 | rules for every new chunk which gets in, taking into account the fact that |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 8897 | those data are partial. If no rule matches before the aforementioned delay, |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8898 | a last check is performed upon expiration, this time considering that the |
Willy Tarreau | d869b24 | 2009-03-15 14:43:58 +0100 | [diff] [blame] | 8899 | contents are definitive. If no delay is set, haproxy will not wait at all |
| 8900 | and will immediately apply a verdict based on the available information. |
| 8901 | Obviously this is unlikely to be very useful and might even be racy, so such |
| 8902 | setups are not recommended. |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8903 | |
| 8904 | As soon as a rule matches, the request is released and continues as usual. If |
| 8905 | the timeout is reached and no rule matches, the default policy will be to let |
| 8906 | it pass through unaffected. |
| 8907 | |
| 8908 | For most protocols, it is enough to set it to a few seconds, as most clients |
| 8909 | send the full request immediately upon connection. Add 3 or more seconds to |
| 8910 | cover TCP retransmits but that's all. For some protocols, it may make sense |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 8911 | to use large values, for instance to ensure that the client never talks |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8912 | before the server (eg: SMTP), or to wait for a client to talk before passing |
| 8913 | data to the server (eg: SSL). Note that the client timeout must cover at |
Willy Tarreau | b824b00 | 2010-09-29 16:36:16 +0200 | [diff] [blame] | 8914 | least the inspection delay, otherwise it will expire first. If the client |
| 8915 | closes the connection or if the buffer is full, the delay immediately expires |
| 8916 | since the contents will not be able to change anymore. |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8917 | |
Willy Tarreau | 55165fe | 2009-05-10 12:02:55 +0200 | [diff] [blame] | 8918 | See also : "tcp-request content accept", "tcp-request content reject", |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 8919 | "timeout client". |
| 8920 | |
| 8921 | |
Emeric Brun | 0a3b67f | 2010-09-24 15:34:53 +0200 | [diff] [blame] | 8922 | tcp-response content <action> [{if | unless} <condition>] |
| 8923 | Perform an action on a session response depending on a layer 4-7 condition |
| 8924 | May be used in sections : defaults | frontend | listen | backend |
| 8925 | no | no | yes | yes |
| 8926 | Arguments : |
Willy Tarreau | c870bfd | 2015-09-28 18:47:38 +0200 | [diff] [blame] | 8927 | <action> defines the action to perform if the condition applies. See |
| 8928 | below. |
Emeric Brun | 0a3b67f | 2010-09-24 15:34:53 +0200 | [diff] [blame] | 8929 | |
| 8930 | <condition> is a standard layer 4-7 ACL-based condition (see section 7). |
| 8931 | |
| 8932 | Response contents can be analysed at an early stage of response processing |
| 8933 | called "TCP content inspection". During this stage, ACL-based rules are |
| 8934 | evaluated every time the response contents are updated, until either an |
Willy Tarreau | cc1e04b | 2013-09-11 23:20:29 +0200 | [diff] [blame] | 8935 | "accept", "close" or a "reject" rule matches, or a TCP response inspection |
| 8936 | delay is set and expires with no matching rule. |
Emeric Brun | 0a3b67f | 2010-09-24 15:34:53 +0200 | [diff] [blame] | 8937 | |
| 8938 | Most often, these decisions will consider a protocol recognition or validity. |
| 8939 | |
| 8940 | Content-based rules are evaluated in their exact declaration order. If no |
| 8941 | rule matches or if there is no rule, the default action is to accept the |
| 8942 | contents. There is no specific limit to the number of rules which may be |
| 8943 | inserted. |
| 8944 | |
Thierry FOURNIER | 236657b | 2015-08-19 08:25:14 +0200 | [diff] [blame] | 8945 | Several types of actions are supported : |
Emeric Brun | 0a3b67f | 2010-09-24 15:34:53 +0200 | [diff] [blame] | 8946 | - accept : |
| 8947 | accepts the response if the condition is true (when used with "if") |
| 8948 | or false (when used with "unless"). The first such rule executed ends |
| 8949 | the rules evaluation. |
| 8950 | |
Willy Tarreau | cc1e04b | 2013-09-11 23:20:29 +0200 | [diff] [blame] | 8951 | - close : |
| 8952 | immediately closes the connection with the server if the condition is |
| 8953 | true (when used with "if"), or false (when used with "unless"). The |
| 8954 | first such rule executed ends the rules evaluation. The main purpose of |
| 8955 | this action is to force a connection to be finished between a client |
| 8956 | and a server after an exchange when the application protocol expects |
| 8957 | some long time outs to elapse first. The goal is to eliminate idle |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 8958 | connections which take significant resources on servers with certain |
Willy Tarreau | cc1e04b | 2013-09-11 23:20:29 +0200 | [diff] [blame] | 8959 | protocols. |
| 8960 | |
Emeric Brun | 0a3b67f | 2010-09-24 15:34:53 +0200 | [diff] [blame] | 8961 | - reject : |
| 8962 | rejects the response if the condition is true (when used with "if") |
| 8963 | or false (when used with "unless"). The first such rule executed ends |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 8964 | the rules evaluation. Rejected session are immediately closed. |
Emeric Brun | 0a3b67f | 2010-09-24 15:34:53 +0200 | [diff] [blame] | 8965 | |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 8966 | - set-var(<var-name>) <expr> |
| 8967 | Sets a variable. |
| 8968 | |
Thierry FOURNIER | e0627bd | 2015-08-04 08:20:33 +0200 | [diff] [blame] | 8969 | - sc-inc-gpc0(<sc-id>): |
| 8970 | This action increments the GPC0 counter according to the sticky |
| 8971 | counter designated by <sc-id>. If an error occurs, this action fails |
| 8972 | silently and the actions evaluation continues. |
| 8973 | |
Thierry FOURNIER | 236657b | 2015-08-19 08:25:14 +0200 | [diff] [blame] | 8974 | - sc-set-gpt0(<sc-id>) <int> : |
| 8975 | This action sets the GPT0 tag according to the sticky counter designated |
| 8976 | by <sc-id> and the value of <int>. The expected result is a boolean. If |
| 8977 | an error occurs, this action silently fails and the actions evaluation |
| 8978 | continues. |
| 8979 | |
Willy Tarreau | 2d392c2 | 2015-08-24 01:43:45 +0200 | [diff] [blame] | 8980 | - "silent-drop" : |
| 8981 | This stops the evaluation of the rules and makes the client-facing |
| 8982 | connection suddenly disappear using a system-dependant way that tries |
| 8983 | to prevent the client from being notified. The effect it then that the |
| 8984 | client still sees an established connection while there's none on |
| 8985 | HAProxy. The purpose is to achieve a comparable effect to "tarpit" |
| 8986 | except that it doesn't use any local resource at all on the machine |
| 8987 | running HAProxy. It can resist much higher loads than "tarpit", and |
| 8988 | slow down stronger attackers. It is important to undestand the impact |
| 8989 | of using this mechanism. All stateful equipments placed between the |
| 8990 | client and HAProxy (firewalls, proxies, load balancers) will also keep |
| 8991 | the established connection for a long time and may suffer from this |
| 8992 | action. On modern Linux systems running with enough privileges, the |
| 8993 | TCP_REPAIR socket option is used to block the emission of a TCP |
| 8994 | reset. On other systems, the socket's TTL is reduced to 1 so that the |
| 8995 | TCP reset doesn't pass the first router, though it's still delivered to |
| 8996 | local networks. Do not use it unless you fully understand how it works. |
| 8997 | |
Emeric Brun | 0a3b67f | 2010-09-24 15:34:53 +0200 | [diff] [blame] | 8998 | Note that the "if/unless" condition is optional. If no condition is set on |
| 8999 | the action, it is simply performed unconditionally. That can be useful for |
| 9000 | for changing the default action to a reject. |
| 9001 | |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 9002 | It is perfectly possible to match layer 7 contents with "tcp-response |
| 9003 | content" rules, but then it is important to ensure that a full response has |
| 9004 | been buffered, otherwise no contents will match. In order to achieve this, |
| 9005 | the best solution involves detecting the HTTP protocol during the inspection |
Emeric Brun | 0a3b67f | 2010-09-24 15:34:53 +0200 | [diff] [blame] | 9006 | period. |
| 9007 | |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 9008 | The "set-var" is used to set the content of a variable. The variable is |
| 9009 | declared inline. |
| 9010 | |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 9011 | <var-name> The name of the variable starts with an indication about |
| 9012 | its scope. The scopes allowed are: |
| 9013 | "sess" : the variable is shared with the whole session |
| 9014 | "txn" : the variable is shared with the transaction |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 9015 | (request and response) |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 9016 | "req" : the variable is shared only during request |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 9017 | processing |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 9018 | "res" : the variable is shared only during response |
| 9019 | processing |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 9020 | This prefix is followed by a name. The separator is a '.'. |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 9021 | The name may only contain characters 'a-z', 'A-Z', '0-9' |
| 9022 | and '_'. |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 9023 | |
| 9024 | <expr> Is a standard HAProxy expression formed by a sample-fetch |
| 9025 | followed by some converters. |
| 9026 | |
| 9027 | Example: |
| 9028 | |
| 9029 | tcp-request content set-var(sess.my_var) src |
| 9030 | |
Emeric Brun | 0a3b67f | 2010-09-24 15:34:53 +0200 | [diff] [blame] | 9031 | See section 7 about ACL usage. |
| 9032 | |
| 9033 | See also : "tcp-request content", "tcp-response inspect-delay" |
| 9034 | |
| 9035 | |
| 9036 | tcp-response inspect-delay <timeout> |
| 9037 | Set the maximum allowed time to wait for a response during content inspection |
| 9038 | May be used in sections : defaults | frontend | listen | backend |
| 9039 | no | no | yes | yes |
| 9040 | Arguments : |
| 9041 | <timeout> is the timeout value specified in milliseconds by default, but |
| 9042 | can be in any other unit if the number is suffixed by the unit, |
| 9043 | as explained at the top of this document. |
| 9044 | |
| 9045 | See also : "tcp-response content", "tcp-request inspect-delay". |
| 9046 | |
| 9047 | |
Krzysztof Piotr Oledzki | 5259dfe | 2008-01-21 01:54:06 +0100 | [diff] [blame] | 9048 | timeout check <timeout> |
| 9049 | Set additional check timeout, but only after a connection has been already |
| 9050 | established. |
| 9051 | |
| 9052 | May be used in sections: defaults | frontend | listen | backend |
| 9053 | yes | no | yes | yes |
| 9054 | Arguments: |
| 9055 | <timeout> is the timeout value specified in milliseconds by default, but |
| 9056 | can be in any other unit if the number is suffixed by the unit, |
| 9057 | as explained at the top of this document. |
| 9058 | |
| 9059 | If set, haproxy uses min("timeout connect", "inter") as a connect timeout |
| 9060 | for check and "timeout check" as an additional read timeout. The "min" is |
| 9061 | used so that people running with *very* long "timeout connect" (eg. those |
| 9062 | who needed this due to the queue or tarpit) do not slow down their checks. |
Willy Tarreau | d7550a2 | 2010-02-10 05:10:19 +0100 | [diff] [blame] | 9063 | (Please also note that there is no valid reason to have such long connect |
| 9064 | timeouts, because "timeout queue" and "timeout tarpit" can always be used to |
| 9065 | avoid that). |
Krzysztof Piotr Oledzki | 5259dfe | 2008-01-21 01:54:06 +0100 | [diff] [blame] | 9066 | |
| 9067 | If "timeout check" is not set haproxy uses "inter" for complete check |
| 9068 | timeout (connect + read) exactly like all <1.3.15 version. |
| 9069 | |
| 9070 | In most cases check request is much simpler and faster to handle than normal |
| 9071 | requests and people may want to kick out laggy servers so this timeout should |
Willy Tarreau | 41a340d | 2008-01-22 12:25:31 +0100 | [diff] [blame] | 9072 | be smaller than "timeout server". |
Krzysztof Piotr Oledzki | 5259dfe | 2008-01-21 01:54:06 +0100 | [diff] [blame] | 9073 | |
| 9074 | This parameter is specific to backends, but can be specified once for all in |
| 9075 | "defaults" sections. This is in fact one of the easiest solutions not to |
| 9076 | forget about it. |
| 9077 | |
Willy Tarreau | 41a340d | 2008-01-22 12:25:31 +0100 | [diff] [blame] | 9078 | See also: "timeout connect", "timeout queue", "timeout server", |
| 9079 | "timeout tarpit". |
Krzysztof Piotr Oledzki | 5259dfe | 2008-01-21 01:54:06 +0100 | [diff] [blame] | 9080 | |
| 9081 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 9082 | timeout client <timeout> |
| 9083 | timeout clitimeout <timeout> (deprecated) |
| 9084 | Set the maximum inactivity time on the client side. |
| 9085 | May be used in sections : defaults | frontend | listen | backend |
| 9086 | yes | yes | yes | no |
| 9087 | Arguments : |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9088 | <timeout> is the timeout value specified in milliseconds by default, but |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 9089 | can be in any other unit if the number is suffixed by the unit, |
| 9090 | as explained at the top of this document. |
| 9091 | |
| 9092 | The inactivity timeout applies when the client is expected to acknowledge or |
| 9093 | send data. In HTTP mode, this timeout is particularly important to consider |
| 9094 | during the first phase, when the client sends the request, and during the |
Baptiste Assmann | 2e1941e | 2016-03-06 23:24:12 +0100 | [diff] [blame] | 9095 | response while it is reading data sent by the server. That said, for the |
| 9096 | first phase, it is preferable to set the "timeout http-request" to better |
| 9097 | protect HAProxy from Slowloris like attacks. The value is specified in |
| 9098 | milliseconds by default, but can be in any other unit if the number is |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 9099 | suffixed by the unit, as specified at the top of this document. In TCP mode |
| 9100 | (and to a lesser extent, in HTTP mode), it is highly recommended that the |
| 9101 | client timeout remains equal to the server timeout in order to avoid complex |
Willy Tarreau | d2a4aa2 | 2008-01-31 15:28:22 +0100 | [diff] [blame] | 9102 | situations to debug. It is a good practice to cover one or several TCP packet |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 9103 | losses by specifying timeouts that are slightly above multiples of 3 seconds |
Willy Tarreau | ce887fd | 2012-05-12 12:50:00 +0200 | [diff] [blame] | 9104 | (eg: 4 or 5 seconds). If some long-lived sessions are mixed with short-lived |
| 9105 | sessions (eg: WebSocket and HTTP), it's worth considering "timeout tunnel", |
Willy Tarreau | 05cdd96 | 2014-05-10 14:30:07 +0200 | [diff] [blame] | 9106 | which overrides "timeout client" and "timeout server" for tunnels, as well as |
| 9107 | "timeout client-fin" for half-closed connections. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 9108 | |
| 9109 | This parameter is specific to frontends, but can be specified once for all in |
| 9110 | "defaults" sections. This is in fact one of the easiest solutions not to |
| 9111 | forget about it. An unspecified timeout results in an infinite timeout, which |
| 9112 | is not recommended. Such a usage is accepted and works but reports a warning |
| 9113 | during startup because it may results in accumulation of expired sessions in |
| 9114 | the system if the system's timeouts are not configured either. |
| 9115 | |
| 9116 | This parameter replaces the old, deprecated "clitimeout". It is recommended |
| 9117 | to use it to write new configurations. The form "timeout clitimeout" is |
| 9118 | provided only by backwards compatibility but its use is strongly discouraged. |
| 9119 | |
Baptiste Assmann | 2e1941e | 2016-03-06 23:24:12 +0100 | [diff] [blame] | 9120 | See also : "clitimeout", "timeout server", "timeout tunnel", |
| 9121 | "timeout http-request". |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 9122 | |
| 9123 | |
Willy Tarreau | 05cdd96 | 2014-05-10 14:30:07 +0200 | [diff] [blame] | 9124 | timeout client-fin <timeout> |
| 9125 | Set the inactivity timeout on the client side for half-closed connections. |
| 9126 | May be used in sections : defaults | frontend | listen | backend |
| 9127 | yes | yes | yes | no |
| 9128 | Arguments : |
| 9129 | <timeout> is the timeout value specified in milliseconds by default, but |
| 9130 | can be in any other unit if the number is suffixed by the unit, |
| 9131 | as explained at the top of this document. |
| 9132 | |
| 9133 | The inactivity timeout applies when the client is expected to acknowledge or |
| 9134 | send data while one direction is already shut down. This timeout is different |
| 9135 | from "timeout client" in that it only applies to connections which are closed |
| 9136 | in one direction. This is particularly useful to avoid keeping connections in |
| 9137 | FIN_WAIT state for too long when clients do not disconnect cleanly. This |
| 9138 | problem is particularly common long connections such as RDP or WebSocket. |
| 9139 | Note that this timeout can override "timeout tunnel" when a connection shuts |
| 9140 | down in one direction. |
| 9141 | |
| 9142 | This parameter is specific to frontends, but can be specified once for all in |
| 9143 | "defaults" sections. By default it is not set, so half-closed connections |
| 9144 | will use the other timeouts (timeout.client or timeout.tunnel). |
| 9145 | |
| 9146 | See also : "timeout client", "timeout server-fin", and "timeout tunnel". |
| 9147 | |
| 9148 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 9149 | timeout connect <timeout> |
| 9150 | timeout contimeout <timeout> (deprecated) |
| 9151 | Set the maximum time to wait for a connection attempt to a server to succeed. |
| 9152 | May be used in sections : defaults | frontend | listen | backend |
| 9153 | yes | no | yes | yes |
| 9154 | Arguments : |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9155 | <timeout> is the timeout value specified in milliseconds by default, but |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 9156 | can be in any other unit if the number is suffixed by the unit, |
| 9157 | as explained at the top of this document. |
| 9158 | |
| 9159 | If the server is located on the same LAN as haproxy, the connection should be |
Willy Tarreau | d2a4aa2 | 2008-01-31 15:28:22 +0100 | [diff] [blame] | 9160 | immediate (less than a few milliseconds). Anyway, it is a good practice to |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 9161 | cover one or several TCP packet losses by specifying timeouts that are |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 9162 | slightly above multiples of 3 seconds (eg: 4 or 5 seconds). By default, the |
Krzysztof Piotr Oledzki | 5259dfe | 2008-01-21 01:54:06 +0100 | [diff] [blame] | 9163 | connect timeout also presets both queue and tarpit timeouts to the same value |
| 9164 | if these have not been specified. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 9165 | |
| 9166 | This parameter is specific to backends, but can be specified once for all in |
| 9167 | "defaults" sections. This is in fact one of the easiest solutions not to |
| 9168 | forget about it. An unspecified timeout results in an infinite timeout, which |
| 9169 | is not recommended. Such a usage is accepted and works but reports a warning |
| 9170 | during startup because it may results in accumulation of failed sessions in |
| 9171 | the system if the system's timeouts are not configured either. |
| 9172 | |
| 9173 | This parameter replaces the old, deprecated "contimeout". It is recommended |
| 9174 | to use it to write new configurations. The form "timeout contimeout" is |
| 9175 | provided only by backwards compatibility but its use is strongly discouraged. |
| 9176 | |
Willy Tarreau | 41a340d | 2008-01-22 12:25:31 +0100 | [diff] [blame] | 9177 | See also: "timeout check", "timeout queue", "timeout server", "contimeout", |
| 9178 | "timeout tarpit". |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 9179 | |
| 9180 | |
Willy Tarreau | b16a574 | 2010-01-10 14:46:16 +0100 | [diff] [blame] | 9181 | timeout http-keep-alive <timeout> |
| 9182 | Set the maximum allowed time to wait for a new HTTP request to appear |
| 9183 | May be used in sections : defaults | frontend | listen | backend |
| 9184 | yes | yes | yes | yes |
| 9185 | Arguments : |
| 9186 | <timeout> is the timeout value specified in milliseconds by default, but |
| 9187 | can be in any other unit if the number is suffixed by the unit, |
| 9188 | as explained at the top of this document. |
| 9189 | |
| 9190 | By default, the time to wait for a new request in case of keep-alive is set |
| 9191 | by "timeout http-request". However this is not always convenient because some |
| 9192 | people want very short keep-alive timeouts in order to release connections |
| 9193 | faster, and others prefer to have larger ones but still have short timeouts |
| 9194 | once the request has started to present itself. |
| 9195 | |
| 9196 | The "http-keep-alive" timeout covers these needs. It will define how long to |
| 9197 | wait for a new HTTP request to start coming after a response was sent. Once |
| 9198 | the first byte of request has been seen, the "http-request" timeout is used |
| 9199 | to wait for the complete request to come. Note that empty lines prior to a |
| 9200 | new request do not refresh the timeout and are not counted as a new request. |
| 9201 | |
| 9202 | There is also another difference between the two timeouts : when a connection |
| 9203 | expires during timeout http-keep-alive, no error is returned, the connection |
| 9204 | just closes. If the connection expires in "http-request" while waiting for a |
| 9205 | connection to complete, a HTTP 408 error is returned. |
| 9206 | |
| 9207 | In general it is optimal to set this value to a few tens to hundreds of |
| 9208 | milliseconds, to allow users to fetch all objects of a page at once but |
| 9209 | without waiting for further clicks. Also, if set to a very small value (eg: |
| 9210 | 1 millisecond) it will probably only accept pipelined requests but not the |
| 9211 | non-pipelined ones. It may be a nice trade-off for very large sites running |
Patrick Mézard | 2382ad6 | 2010-05-09 10:43:32 +0200 | [diff] [blame] | 9212 | with tens to hundreds of thousands of clients. |
Willy Tarreau | b16a574 | 2010-01-10 14:46:16 +0100 | [diff] [blame] | 9213 | |
| 9214 | If this parameter is not set, the "http-request" timeout applies, and if both |
| 9215 | are not set, "timeout client" still applies at the lower level. It should be |
| 9216 | set in the frontend to take effect, unless the frontend is in TCP mode, in |
| 9217 | which case the HTTP backend's timeout will be used. |
| 9218 | |
| 9219 | See also : "timeout http-request", "timeout client". |
| 9220 | |
| 9221 | |
Willy Tarreau | 036fae0 | 2008-01-06 13:24:40 +0100 | [diff] [blame] | 9222 | timeout http-request <timeout> |
| 9223 | Set the maximum allowed time to wait for a complete HTTP request |
| 9224 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | cd7afc0 | 2009-07-12 10:03:17 +0200 | [diff] [blame] | 9225 | yes | yes | yes | yes |
Willy Tarreau | 036fae0 | 2008-01-06 13:24:40 +0100 | [diff] [blame] | 9226 | Arguments : |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9227 | <timeout> is the timeout value specified in milliseconds by default, but |
Willy Tarreau | 036fae0 | 2008-01-06 13:24:40 +0100 | [diff] [blame] | 9228 | can be in any other unit if the number is suffixed by the unit, |
| 9229 | as explained at the top of this document. |
| 9230 | |
| 9231 | In order to offer DoS protection, it may be required to lower the maximum |
| 9232 | accepted time to receive a complete HTTP request without affecting the client |
| 9233 | timeout. This helps protecting against established connections on which |
| 9234 | nothing is sent. The client timeout cannot offer a good protection against |
| 9235 | this abuse because it is an inactivity timeout, which means that if the |
| 9236 | attacker sends one character every now and then, the timeout will not |
| 9237 | trigger. With the HTTP request timeout, no matter what speed the client |
Willy Tarreau | 2705a61 | 2014-05-23 17:38:34 +0200 | [diff] [blame] | 9238 | types, the request will be aborted if it does not complete in time. When the |
| 9239 | timeout expires, an HTTP 408 response is sent to the client to inform it |
| 9240 | about the problem, and the connection is closed. The logs will report |
| 9241 | termination codes "cR". Some recent browsers are having problems with this |
| 9242 | standard, well-documented behaviour, so it might be needed to hide the 408 |
Willy Tarreau | 0f228a0 | 2015-05-01 15:37:53 +0200 | [diff] [blame] | 9243 | code using "option http-ignore-probes" or "errorfile 408 /dev/null". See |
| 9244 | more details in the explanations of the "cR" termination code in section 8.5. |
Willy Tarreau | 036fae0 | 2008-01-06 13:24:40 +0100 | [diff] [blame] | 9245 | |
Baptiste Assmann | eccdf43 | 2015-10-28 13:49:01 +0100 | [diff] [blame] | 9246 | By default, this timeout only applies to the header part of the request, |
| 9247 | and not to any data. As soon as the empty line is received, this timeout is |
| 9248 | not used anymore. When combined with "option http-buffer-request", this |
| 9249 | timeout also applies to the body of the request.. |
| 9250 | It is used again on keep-alive connections to wait for a second |
Willy Tarreau | b16a574 | 2010-01-10 14:46:16 +0100 | [diff] [blame] | 9251 | request if "timeout http-keep-alive" is not set. |
Willy Tarreau | 036fae0 | 2008-01-06 13:24:40 +0100 | [diff] [blame] | 9252 | |
| 9253 | Generally it is enough to set it to a few seconds, as most clients send the |
| 9254 | full request immediately upon connection. Add 3 or more seconds to cover TCP |
| 9255 | retransmits but that's all. Setting it to very low values (eg: 50 ms) will |
| 9256 | generally work on local networks as long as there are no packet losses. This |
| 9257 | will prevent people from sending bare HTTP requests using telnet. |
| 9258 | |
| 9259 | If this parameter is not set, the client timeout still applies between each |
Willy Tarreau | cd7afc0 | 2009-07-12 10:03:17 +0200 | [diff] [blame] | 9260 | chunk of the incoming request. It should be set in the frontend to take |
| 9261 | effect, unless the frontend is in TCP mode, in which case the HTTP backend's |
| 9262 | timeout will be used. |
Willy Tarreau | 036fae0 | 2008-01-06 13:24:40 +0100 | [diff] [blame] | 9263 | |
Willy Tarreau | 0f228a0 | 2015-05-01 15:37:53 +0200 | [diff] [blame] | 9264 | See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and |
Baptiste Assmann | eccdf43 | 2015-10-28 13:49:01 +0100 | [diff] [blame] | 9265 | "timeout client", "option http-buffer-request". |
Willy Tarreau | 036fae0 | 2008-01-06 13:24:40 +0100 | [diff] [blame] | 9266 | |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9267 | |
| 9268 | timeout queue <timeout> |
| 9269 | Set the maximum time to wait in the queue for a connection slot to be free |
| 9270 | May be used in sections : defaults | frontend | listen | backend |
| 9271 | yes | no | yes | yes |
| 9272 | Arguments : |
| 9273 | <timeout> is the timeout value specified in milliseconds by default, but |
| 9274 | can be in any other unit if the number is suffixed by the unit, |
| 9275 | as explained at the top of this document. |
| 9276 | |
| 9277 | When a server's maxconn is reached, connections are left pending in a queue |
| 9278 | which may be server-specific or global to the backend. In order not to wait |
| 9279 | indefinitely, a timeout is applied to requests pending in the queue. If the |
| 9280 | timeout is reached, it is considered that the request will almost never be |
| 9281 | served, so it is dropped and a 503 error is returned to the client. |
| 9282 | |
| 9283 | The "timeout queue" statement allows to fix the maximum time for a request to |
| 9284 | be left pending in a queue. If unspecified, the same value as the backend's |
| 9285 | connection timeout ("timeout connect") is used, for backwards compatibility |
| 9286 | with older versions with no "timeout queue" parameter. |
| 9287 | |
| 9288 | See also : "timeout connect", "contimeout". |
| 9289 | |
| 9290 | |
| 9291 | timeout server <timeout> |
| 9292 | timeout srvtimeout <timeout> (deprecated) |
| 9293 | Set the maximum inactivity time on the server side. |
| 9294 | May be used in sections : defaults | frontend | listen | backend |
| 9295 | yes | no | yes | yes |
| 9296 | Arguments : |
| 9297 | <timeout> is the timeout value specified in milliseconds by default, but |
| 9298 | can be in any other unit if the number is suffixed by the unit, |
| 9299 | as explained at the top of this document. |
| 9300 | |
| 9301 | The inactivity timeout applies when the server is expected to acknowledge or |
| 9302 | send data. In HTTP mode, this timeout is particularly important to consider |
| 9303 | during the first phase of the server's response, when it has to send the |
| 9304 | headers, as it directly represents the server's processing time for the |
| 9305 | request. To find out what value to put there, it's often good to start with |
| 9306 | what would be considered as unacceptable response times, then check the logs |
| 9307 | to observe the response time distribution, and adjust the value accordingly. |
| 9308 | |
| 9309 | The value is specified in milliseconds by default, but can be in any other |
| 9310 | unit if the number is suffixed by the unit, as specified at the top of this |
| 9311 | document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly |
| 9312 | recommended that the client timeout remains equal to the server timeout in |
| 9313 | order to avoid complex situations to debug. Whatever the expected server |
Willy Tarreau | d2a4aa2 | 2008-01-31 15:28:22 +0100 | [diff] [blame] | 9314 | response times, it is a good practice to cover at least one or several TCP |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9315 | packet losses by specifying timeouts that are slightly above multiples of 3 |
Willy Tarreau | ce887fd | 2012-05-12 12:50:00 +0200 | [diff] [blame] | 9316 | seconds (eg: 4 or 5 seconds minimum). If some long-lived sessions are mixed |
| 9317 | with short-lived sessions (eg: WebSocket and HTTP), it's worth considering |
| 9318 | "timeout tunnel", which overrides "timeout client" and "timeout server" for |
| 9319 | tunnels. |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9320 | |
| 9321 | This parameter is specific to backends, but can be specified once for all in |
| 9322 | "defaults" sections. This is in fact one of the easiest solutions not to |
| 9323 | forget about it. An unspecified timeout results in an infinite timeout, which |
| 9324 | is not recommended. Such a usage is accepted and works but reports a warning |
| 9325 | during startup because it may results in accumulation of expired sessions in |
| 9326 | the system if the system's timeouts are not configured either. |
| 9327 | |
| 9328 | This parameter replaces the old, deprecated "srvtimeout". It is recommended |
| 9329 | to use it to write new configurations. The form "timeout srvtimeout" is |
| 9330 | provided only by backwards compatibility but its use is strongly discouraged. |
| 9331 | |
Willy Tarreau | ce887fd | 2012-05-12 12:50:00 +0200 | [diff] [blame] | 9332 | See also : "srvtimeout", "timeout client" and "timeout tunnel". |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9333 | |
Willy Tarreau | 05cdd96 | 2014-05-10 14:30:07 +0200 | [diff] [blame] | 9334 | |
| 9335 | timeout server-fin <timeout> |
| 9336 | Set the inactivity timeout on the server side for half-closed connections. |
| 9337 | May be used in sections : defaults | frontend | listen | backend |
| 9338 | yes | no | yes | yes |
| 9339 | Arguments : |
| 9340 | <timeout> is the timeout value specified in milliseconds by default, but |
| 9341 | can be in any other unit if the number is suffixed by the unit, |
| 9342 | as explained at the top of this document. |
| 9343 | |
| 9344 | The inactivity timeout applies when the server is expected to acknowledge or |
| 9345 | send data while one direction is already shut down. This timeout is different |
| 9346 | from "timeout server" in that it only applies to connections which are closed |
| 9347 | in one direction. This is particularly useful to avoid keeping connections in |
| 9348 | FIN_WAIT state for too long when a remote server does not disconnect cleanly. |
| 9349 | This problem is particularly common long connections such as RDP or WebSocket. |
| 9350 | Note that this timeout can override "timeout tunnel" when a connection shuts |
| 9351 | down in one direction. This setting was provided for completeness, but in most |
| 9352 | situations, it should not be needed. |
| 9353 | |
| 9354 | This parameter is specific to backends, but can be specified once for all in |
| 9355 | "defaults" sections. By default it is not set, so half-closed connections |
| 9356 | will use the other timeouts (timeout.server or timeout.tunnel). |
| 9357 | |
| 9358 | See also : "timeout client-fin", "timeout server", and "timeout tunnel". |
| 9359 | |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9360 | |
| 9361 | timeout tarpit <timeout> |
Cyril Bonté | 78caf84 | 2010-03-10 22:41:43 +0100 | [diff] [blame] | 9362 | Set the duration for which tarpitted connections will be maintained |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9363 | May be used in sections : defaults | frontend | listen | backend |
| 9364 | yes | yes | yes | yes |
| 9365 | Arguments : |
| 9366 | <timeout> is the tarpit duration specified in milliseconds by default, but |
| 9367 | can be in any other unit if the number is suffixed by the unit, |
| 9368 | as explained at the top of this document. |
| 9369 | |
| 9370 | When a connection is tarpitted using "reqtarpit", it is maintained open with |
| 9371 | no activity for a certain amount of time, then closed. "timeout tarpit" |
| 9372 | defines how long it will be maintained open. |
| 9373 | |
| 9374 | The value is specified in milliseconds by default, but can be in any other |
| 9375 | unit if the number is suffixed by the unit, as specified at the top of this |
| 9376 | document. If unspecified, the same value as the backend's connection timeout |
| 9377 | ("timeout connect") is used, for backwards compatibility with older versions |
Cyril Bonté | 78caf84 | 2010-03-10 22:41:43 +0100 | [diff] [blame] | 9378 | with no "timeout tarpit" parameter. |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9379 | |
| 9380 | See also : "timeout connect", "contimeout". |
| 9381 | |
| 9382 | |
Willy Tarreau | ce887fd | 2012-05-12 12:50:00 +0200 | [diff] [blame] | 9383 | timeout tunnel <timeout> |
| 9384 | Set the maximum inactivity time on the client and server side for tunnels. |
| 9385 | May be used in sections : defaults | frontend | listen | backend |
| 9386 | yes | no | yes | yes |
| 9387 | Arguments : |
| 9388 | <timeout> is the timeout value specified in milliseconds by default, but |
| 9389 | can be in any other unit if the number is suffixed by the unit, |
| 9390 | as explained at the top of this document. |
| 9391 | |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 9392 | The tunnel timeout applies when a bidirectional connection is established |
Willy Tarreau | ce887fd | 2012-05-12 12:50:00 +0200 | [diff] [blame] | 9393 | between a client and a server, and the connection remains inactive in both |
| 9394 | directions. This timeout supersedes both the client and server timeouts once |
| 9395 | the connection becomes a tunnel. In TCP, this timeout is used as soon as no |
| 9396 | analyser remains attached to either connection (eg: tcp content rules are |
| 9397 | accepted). In HTTP, this timeout is used when a connection is upgraded (eg: |
| 9398 | when switching to the WebSocket protocol, or forwarding a CONNECT request |
| 9399 | to a proxy), or after the first response when no keepalive/close option is |
| 9400 | specified. |
| 9401 | |
Willy Tarreau | 05cdd96 | 2014-05-10 14:30:07 +0200 | [diff] [blame] | 9402 | Since this timeout is usually used in conjunction with long-lived connections, |
| 9403 | it usually is a good idea to also set "timeout client-fin" to handle the |
| 9404 | situation where a client suddenly disappears from the net and does not |
| 9405 | acknowledge a close, or sends a shutdown and does not acknowledge pending |
| 9406 | data anymore. This can happen in lossy networks where firewalls are present, |
| 9407 | and is detected by the presence of large amounts of sessions in a FIN_WAIT |
| 9408 | state. |
| 9409 | |
Willy Tarreau | ce887fd | 2012-05-12 12:50:00 +0200 | [diff] [blame] | 9410 | The value is specified in milliseconds by default, but can be in any other |
| 9411 | unit if the number is suffixed by the unit, as specified at the top of this |
| 9412 | document. Whatever the expected normal idle time, it is a good practice to |
| 9413 | cover at least one or several TCP packet losses by specifying timeouts that |
| 9414 | are slightly above multiples of 3 seconds (eg: 4 or 5 seconds minimum). |
| 9415 | |
| 9416 | This parameter is specific to backends, but can be specified once for all in |
| 9417 | "defaults" sections. This is in fact one of the easiest solutions not to |
| 9418 | forget about it. |
| 9419 | |
| 9420 | Example : |
| 9421 | defaults http |
| 9422 | option http-server-close |
| 9423 | timeout connect 5s |
| 9424 | timeout client 30s |
Willy Tarreau | 05cdd96 | 2014-05-10 14:30:07 +0200 | [diff] [blame] | 9425 | timeout client-fin 30s |
Willy Tarreau | ce887fd | 2012-05-12 12:50:00 +0200 | [diff] [blame] | 9426 | timeout server 30s |
| 9427 | timeout tunnel 1h # timeout to use with WebSocket and CONNECT |
| 9428 | |
Willy Tarreau | 05cdd96 | 2014-05-10 14:30:07 +0200 | [diff] [blame] | 9429 | See also : "timeout client", "timeout client-fin", "timeout server". |
Willy Tarreau | ce887fd | 2012-05-12 12:50:00 +0200 | [diff] [blame] | 9430 | |
| 9431 | |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9432 | transparent (deprecated) |
| 9433 | Enable client-side transparent proxying |
| 9434 | May be used in sections : defaults | frontend | listen | backend |
Willy Tarreau | 4b1f859 | 2008-12-23 23:13:55 +0100 | [diff] [blame] | 9435 | yes | no | yes | yes |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9436 | Arguments : none |
| 9437 | |
| 9438 | This keyword was introduced in order to provide layer 7 persistence to layer |
| 9439 | 3 load balancers. The idea is to use the OS's ability to redirect an incoming |
| 9440 | connection for a remote address to a local process (here HAProxy), and let |
| 9441 | this process know what address was initially requested. When this option is |
| 9442 | used, sessions without cookies will be forwarded to the original destination |
| 9443 | IP address of the incoming request (which should match that of another |
| 9444 | equipment), while requests with cookies will still be forwarded to the |
| 9445 | appropriate server. |
| 9446 | |
| 9447 | The "transparent" keyword is deprecated, use "option transparent" instead. |
| 9448 | |
| 9449 | Note that contrary to a common belief, this option does NOT make HAProxy |
| 9450 | present the client's IP to the server when establishing the connection. |
| 9451 | |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9452 | See also: "option transparent" |
| 9453 | |
William Lallemand | a73203e | 2012-03-12 12:48:57 +0100 | [diff] [blame] | 9454 | unique-id-format <string> |
| 9455 | Generate a unique ID for each request. |
| 9456 | May be used in sections : defaults | frontend | listen | backend |
| 9457 | yes | yes | yes | no |
| 9458 | Arguments : |
| 9459 | <string> is a log-format string. |
| 9460 | |
Cyril Bonté | 108cf6e | 2012-04-21 23:30:29 +0200 | [diff] [blame] | 9461 | This keyword creates a ID for each request using the custom log format. A |
| 9462 | unique ID is useful to trace a request passing through many components of |
| 9463 | a complex infrastructure. The newly created ID may also be logged using the |
| 9464 | %ID tag the log-format string. |
William Lallemand | a73203e | 2012-03-12 12:48:57 +0100 | [diff] [blame] | 9465 | |
Cyril Bonté | 108cf6e | 2012-04-21 23:30:29 +0200 | [diff] [blame] | 9466 | The format should be composed from elements that are guaranteed to be |
| 9467 | unique when combined together. For instance, if multiple haproxy instances |
| 9468 | are involved, it might be important to include the node name. It is often |
| 9469 | needed to log the incoming connection's source and destination addresses |
| 9470 | and ports. Note that since multiple requests may be performed over the same |
| 9471 | connection, including a request counter may help differentiate them. |
| 9472 | Similarly, a timestamp may protect against a rollover of the counter. |
| 9473 | Logging the process ID will avoid collisions after a service restart. |
William Lallemand | a73203e | 2012-03-12 12:48:57 +0100 | [diff] [blame] | 9474 | |
Cyril Bonté | 108cf6e | 2012-04-21 23:30:29 +0200 | [diff] [blame] | 9475 | It is recommended to use hexadecimal notation for many fields since it |
| 9476 | makes them more compact and saves space in logs. |
William Lallemand | a73203e | 2012-03-12 12:48:57 +0100 | [diff] [blame] | 9477 | |
Cyril Bonté | 108cf6e | 2012-04-21 23:30:29 +0200 | [diff] [blame] | 9478 | Example: |
William Lallemand | a73203e | 2012-03-12 12:48:57 +0100 | [diff] [blame] | 9479 | |
Julien Vehent | f21be32 | 2014-03-07 08:27:34 -0500 | [diff] [blame] | 9480 | unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid |
William Lallemand | a73203e | 2012-03-12 12:48:57 +0100 | [diff] [blame] | 9481 | |
| 9482 | will generate: |
| 9483 | |
| 9484 | 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A |
| 9485 | |
| 9486 | See also: "unique-id-header" |
| 9487 | |
| 9488 | unique-id-header <name> |
| 9489 | Add a unique ID header in the HTTP request. |
| 9490 | May be used in sections : defaults | frontend | listen | backend |
| 9491 | yes | yes | yes | no |
| 9492 | Arguments : |
| 9493 | <name> is the name of the header. |
| 9494 | |
Cyril Bonté | 108cf6e | 2012-04-21 23:30:29 +0200 | [diff] [blame] | 9495 | Add a unique-id header in the HTTP request sent to the server, using the |
| 9496 | unique-id-format. It can't work if the unique-id-format doesn't exist. |
William Lallemand | a73203e | 2012-03-12 12:48:57 +0100 | [diff] [blame] | 9497 | |
Cyril Bonté | 108cf6e | 2012-04-21 23:30:29 +0200 | [diff] [blame] | 9498 | Example: |
William Lallemand | a73203e | 2012-03-12 12:48:57 +0100 | [diff] [blame] | 9499 | |
Julien Vehent | f21be32 | 2014-03-07 08:27:34 -0500 | [diff] [blame] | 9500 | unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid |
William Lallemand | a73203e | 2012-03-12 12:48:57 +0100 | [diff] [blame] | 9501 | unique-id-header X-Unique-ID |
| 9502 | |
| 9503 | will generate: |
| 9504 | |
| 9505 | X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A |
| 9506 | |
| 9507 | See also: "unique-id-format" |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9508 | |
Willy Tarreau | f51658d | 2014-04-23 01:21:56 +0200 | [diff] [blame] | 9509 | use_backend <backend> [{if | unless} <condition>] |
Willy Tarreau | 1d0dfb1 | 2009-07-07 15:10:31 +0200 | [diff] [blame] | 9510 | Switch to a specific backend if/unless an ACL-based condition is matched. |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9511 | May be used in sections : defaults | frontend | listen | backend |
| 9512 | no | yes | yes | no |
| 9513 | Arguments : |
Bertrand Jacquin | 702d44f | 2013-11-19 11:43:06 +0100 | [diff] [blame] | 9514 | <backend> is the name of a valid backend or "listen" section, or a |
| 9515 | "log-format" string resolving to a backend name. |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9516 | |
Willy Tarreau | f51658d | 2014-04-23 01:21:56 +0200 | [diff] [blame] | 9517 | <condition> is a condition composed of ACLs, as described in section 7. If |
| 9518 | it is omitted, the rule is unconditionally applied. |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9519 | |
| 9520 | When doing content-switching, connections arrive on a frontend and are then |
| 9521 | dispatched to various backends depending on a number of conditions. The |
| 9522 | relation between the conditions and the backends is described with the |
Willy Tarreau | 1d0dfb1 | 2009-07-07 15:10:31 +0200 | [diff] [blame] | 9523 | "use_backend" keyword. While it is normally used with HTTP processing, it can |
| 9524 | also be used in pure TCP, either without content using stateless ACLs (eg: |
| 9525 | source address validation) or combined with a "tcp-request" rule to wait for |
| 9526 | some payload. |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 9527 | |
| 9528 | There may be as many "use_backend" rules as desired. All of these rules are |
| 9529 | evaluated in their declaration order, and the first one which matches will |
| 9530 | assign the backend. |
| 9531 | |
| 9532 | In the first form, the backend will be used if the condition is met. In the |
| 9533 | second form, the backend will be used if the condition is not met. If no |
| 9534 | condition is valid, the backend defined with "default_backend" will be used. |
| 9535 | If no default backend is defined, either the servers in the same section are |
| 9536 | used (in case of a "listen" section) or, in case of a frontend, no server is |
| 9537 | used and a 503 service unavailable response is returned. |
| 9538 | |
Willy Tarreau | 51aecc7 | 2009-07-12 09:47:04 +0200 | [diff] [blame] | 9539 | Note that it is possible to switch from a TCP frontend to an HTTP backend. In |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 9540 | this case, either the frontend has already checked that the protocol is HTTP, |
Willy Tarreau | 51aecc7 | 2009-07-12 09:47:04 +0200 | [diff] [blame] | 9541 | and backend processing will immediately follow, or the backend will wait for |
| 9542 | a complete HTTP request to get in. This feature is useful when a frontend |
| 9543 | must decode several protocols on a unique port, one of them being HTTP. |
| 9544 | |
Bertrand Jacquin | 702d44f | 2013-11-19 11:43:06 +0100 | [diff] [blame] | 9545 | When <backend> is a simple name, it is resolved at configuration time, and an |
| 9546 | error is reported if the specified backend does not exist. If <backend> is |
| 9547 | a log-format string instead, no check may be done at configuration time, so |
| 9548 | the backend name is resolved dynamically at run time. If the resulting |
| 9549 | backend name does not correspond to any valid backend, no other rule is |
| 9550 | evaluated, and the default_backend directive is applied instead. Note that |
| 9551 | when using dynamic backend names, it is highly recommended to use a prefix |
| 9552 | that no other backend uses in order to ensure that an unauthorized backend |
| 9553 | cannot be forced from the request. |
| 9554 | |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 9555 | It is worth mentioning that "use_backend" rules with an explicit name are |
Bertrand Jacquin | 702d44f | 2013-11-19 11:43:06 +0100 | [diff] [blame] | 9556 | used to detect the association between frontends and backends to compute the |
| 9557 | backend's "fullconn" setting. This cannot be done for dynamic names. |
| 9558 | |
| 9559 | See also: "default_backend", "tcp-request", "fullconn", "log-format", and |
| 9560 | section 7 about ACLs. |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 9561 | |
Willy Tarreau | 036fae0 | 2008-01-06 13:24:40 +0100 | [diff] [blame] | 9562 | |
Willy Tarreau | 4a5cade | 2012-04-05 21:09:48 +0200 | [diff] [blame] | 9563 | use-server <server> if <condition> |
| 9564 | use-server <server> unless <condition> |
| 9565 | Only use a specific server if/unless an ACL-based condition is matched. |
| 9566 | May be used in sections : defaults | frontend | listen | backend |
| 9567 | no | no | yes | yes |
| 9568 | Arguments : |
Cyril Bonté | 108cf6e | 2012-04-21 23:30:29 +0200 | [diff] [blame] | 9569 | <server> is the name of a valid server in the same backend section. |
Willy Tarreau | 4a5cade | 2012-04-05 21:09:48 +0200 | [diff] [blame] | 9570 | |
| 9571 | <condition> is a condition composed of ACLs, as described in section 7. |
| 9572 | |
| 9573 | By default, connections which arrive to a backend are load-balanced across |
| 9574 | the available servers according to the configured algorithm, unless a |
| 9575 | persistence mechanism such as a cookie is used and found in the request. |
| 9576 | |
| 9577 | Sometimes it is desirable to forward a particular request to a specific |
| 9578 | server without having to declare a dedicated backend for this server. This |
| 9579 | can be achieved using the "use-server" rules. These rules are evaluated after |
| 9580 | the "redirect" rules and before evaluating cookies, and they have precedence |
| 9581 | on them. There may be as many "use-server" rules as desired. All of these |
| 9582 | rules are evaluated in their declaration order, and the first one which |
| 9583 | matches will assign the server. |
| 9584 | |
| 9585 | If a rule designates a server which is down, and "option persist" is not used |
| 9586 | and no force-persist rule was validated, it is ignored and evaluation goes on |
| 9587 | with the next rules until one matches. |
| 9588 | |
| 9589 | In the first form, the server will be used if the condition is met. In the |
| 9590 | second form, the server will be used if the condition is not met. If no |
| 9591 | condition is valid, the processing continues and the server will be assigned |
| 9592 | according to other persistence mechanisms. |
| 9593 | |
| 9594 | Note that even if a rule is matched, cookie processing is still performed but |
| 9595 | does not assign the server. This allows prefixed cookies to have their prefix |
| 9596 | stripped. |
| 9597 | |
| 9598 | The "use-server" statement works both in HTTP and TCP mode. This makes it |
| 9599 | suitable for use with content-based inspection. For instance, a server could |
| 9600 | be selected in a farm according to the TLS SNI field. And if these servers |
| 9601 | have their weight set to zero, they will not be used for other traffic. |
| 9602 | |
| 9603 | Example : |
| 9604 | # intercept incoming TLS requests based on the SNI field |
| 9605 | use-server www if { req_ssl_sni -i www.example.com } |
| 9606 | server www 192.168.0.1:443 weight 0 |
| 9607 | use-server mail if { req_ssl_sni -i mail.example.com } |
| 9608 | server mail 192.168.0.1:587 weight 0 |
| 9609 | use-server imap if { req_ssl_sni -i imap.example.com } |
| 9610 | server mail 192.168.0.1:993 weight 0 |
| 9611 | # all the rest is forwarded to this server |
| 9612 | server default 192.168.0.2:443 check |
| 9613 | |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 9614 | See also: "use_backend", section 5 about server and section 7 about ACLs. |
Willy Tarreau | 4a5cade | 2012-04-05 21:09:48 +0200 | [diff] [blame] | 9615 | |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9616 | |
| 9617 | 5. Bind and Server options |
| 9618 | -------------------------- |
| 9619 | |
| 9620 | The "bind", "server" and "default-server" keywords support a number of settings |
| 9621 | depending on some build options and on the system HAProxy was built on. These |
| 9622 | settings generally each consist in one word sometimes followed by a value, |
| 9623 | written on the same line as the "bind" or "server" line. All these options are |
| 9624 | described in this section. |
| 9625 | |
| 9626 | |
| 9627 | 5.1. Bind options |
| 9628 | ----------------- |
| 9629 | |
| 9630 | The "bind" keyword supports a certain number of settings which are all passed |
| 9631 | as arguments on the same line. The order in which those arguments appear makes |
| 9632 | no importance, provided that they appear after the bind address. All of these |
| 9633 | parameters are optional. Some of them consist in a single words (booleans), |
| 9634 | while other ones expect a value after them. In this case, the value must be |
| 9635 | provided immediately after the setting name. |
| 9636 | |
| 9637 | The currently supported settings are the following ones. |
| 9638 | |
| 9639 | accept-proxy |
| 9640 | Enforces the use of the PROXY protocol over any connection accepted by any of |
Willy Tarreau | 7799267 | 2014-06-14 11:06:17 +0200 | [diff] [blame] | 9641 | the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol |
| 9642 | are supported and correctly detected. The PROXY protocol dictates the layer |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9643 | 3/4 addresses of the incoming connection to be used everywhere an address is |
| 9644 | used, with the only exception of "tcp-request connection" rules which will |
| 9645 | only see the real connection address. Logs will reflect the addresses |
| 9646 | indicated in the protocol, unless it is violated, in which case the real |
| 9647 | address will still be used. This keyword combined with support from external |
| 9648 | components can be used as an efficient and reliable alternative to the |
| 9649 | X-Forwarded-For mechanism which is not always reliable and not even always |
Willy Tarreau | 4f0d919 | 2013-06-11 20:40:55 +0200 | [diff] [blame] | 9650 | usable. See also "tcp-request connection expect-proxy" for a finer-grained |
| 9651 | setting of which client is allowed to use the protocol. |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9652 | |
Willy Tarreau | ab861d3 | 2013-04-02 02:30:41 +0200 | [diff] [blame] | 9653 | alpn <protocols> |
| 9654 | This enables the TLS ALPN extension and advertises the specified protocol |
| 9655 | list as supported on top of ALPN. The protocol list consists in a comma- |
| 9656 | delimited list of protocol names, for instance: "http/1.1,http/1.0" (without |
| 9657 | quotes). This requires that the SSL library is build with support for TLS |
| 9658 | extensions enabled (check with haproxy -vv). The ALPN extension replaces the |
| 9659 | initial NPN extension. |
| 9660 | |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9661 | backlog <backlog> |
| 9662 | Sets the socket's backlog to this value. If unspecified, the frontend's |
| 9663 | backlog is used instead, which generally defaults to the maxconn value. |
| 9664 | |
Emeric Brun | 7fb3442 | 2012-09-28 15:26:15 +0200 | [diff] [blame] | 9665 | ecdhe <named curve> |
| 9666 | This setting is only available when support for OpenSSL was built in. It sets |
Emeric Brun | 6924ef8 | 2013-03-06 14:08:53 +0100 | [diff] [blame] | 9667 | the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default, |
| 9668 | used named curve is prime256v1. |
Emeric Brun | 7fb3442 | 2012-09-28 15:26:15 +0200 | [diff] [blame] | 9669 | |
Emeric Brun | fd33a26 | 2012-10-11 16:28:27 +0200 | [diff] [blame] | 9670 | ca-file <cafile> |
Emeric Brun | 1a073b4 | 2012-09-28 17:07:34 +0200 | [diff] [blame] | 9671 | This setting is only available when support for OpenSSL was built in. It |
| 9672 | designates a PEM file from which to load CA certificates used to verify |
| 9673 | client's certificate. |
| 9674 | |
Emeric Brun | b6dc934 | 2012-09-28 17:55:37 +0200 | [diff] [blame] | 9675 | ca-ignore-err [all|<errorID>,...] |
| 9676 | This setting is only available when support for OpenSSL was built in. |
| 9677 | Sets a comma separated list of errorIDs to ignore during verify at depth > 0. |
| 9678 | If set to 'all', all errors are ignored. SSL handshake is not aborted if an |
| 9679 | error is ignored. |
| 9680 | |
Christopher Faulet | 31af49d | 2015-06-09 17:29:50 +0200 | [diff] [blame] | 9681 | ca-sign-file <cafile> |
| 9682 | This setting is only available when support for OpenSSL was built in. It |
| 9683 | designates a PEM file containing both the CA certificate and the CA private |
| 9684 | key used to create and sign server's certificates. This is a mandatory |
| 9685 | setting when the dynamic generation of certificates is enabled. See |
| 9686 | 'generate-certificates' for details. |
| 9687 | |
| 9688 | ca-sign-passphrase <passphrase> |
| 9689 | This setting is only available when support for OpenSSL was built in. It is |
| 9690 | the CA private key passphrase. This setting is optional and used only when |
| 9691 | the dynamic generation of certificates is enabled. See |
| 9692 | 'generate-certificates' for details. |
| 9693 | |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9694 | ciphers <ciphers> |
| 9695 | This setting is only available when support for OpenSSL was built in. It sets |
| 9696 | the string describing the list of cipher algorithms ("cipher suite") that are |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 9697 | negotiated during the SSL/TLS handshake. The format of the string is defined |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9698 | in "man 1 ciphers" from OpenSSL man pages, and can be for instance a string |
| 9699 | such as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without quotes). |
| 9700 | |
Emeric Brun | fd33a26 | 2012-10-11 16:28:27 +0200 | [diff] [blame] | 9701 | crl-file <crlfile> |
Emeric Brun | 1a073b4 | 2012-09-28 17:07:34 +0200 | [diff] [blame] | 9702 | This setting is only available when support for OpenSSL was built in. It |
| 9703 | designates a PEM file from which to load certificate revocation list used |
| 9704 | to verify client's certificate. |
| 9705 | |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9706 | crt <cert> |
Alex Davies | 0fbf016 | 2013-03-02 16:04:50 +0000 | [diff] [blame] | 9707 | This setting is only available when support for OpenSSL was built in. It |
| 9708 | designates a PEM file containing both the required certificates and any |
| 9709 | associated private keys. This file can be built by concatenating multiple |
| 9710 | PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA |
| 9711 | requires an intermediate certificate, this can also be concatenated into this |
| 9712 | file. |
| 9713 | |
| 9714 | If the OpenSSL used supports Diffie-Hellman, parameters present in this file |
| 9715 | are loaded. |
| 9716 | |
| 9717 | If a directory name is used instead of a PEM file, then all files found in |
Cyril Bonté | 3180f7b | 2015-01-25 00:16:08 +0100 | [diff] [blame] | 9718 | that directory will be loaded in alphabetic order unless their name ends with |
Janusz Dziemidowicz | 2c701b5 | 2015-03-07 23:03:59 +0100 | [diff] [blame] | 9719 | '.issuer', '.ocsp' or '.sctl' (reserved extensions). This directive may be |
| 9720 | specified multiple times in order to load certificates from multiple files or |
| 9721 | directories. The certificates will be presented to clients who provide a |
| 9722 | valid TLS Server Name Indication field matching one of their CN or alt |
| 9723 | subjects. Wildcards are supported, where a wildcard character '*' is used |
| 9724 | instead of the first hostname component (eg: *.example.org matches |
| 9725 | www.example.org but not www.sub.example.org). |
Alex Davies | 0fbf016 | 2013-03-02 16:04:50 +0000 | [diff] [blame] | 9726 | |
| 9727 | If no SNI is provided by the client or if the SSL library does not support |
| 9728 | TLS extensions, or if the client provides an SNI hostname which does not |
| 9729 | match any certificate, then the first loaded certificate will be presented. |
| 9730 | This means that when loading certificates from a directory, it is highly |
Cyril Bonté | 3180f7b | 2015-01-25 00:16:08 +0100 | [diff] [blame] | 9731 | recommended to load the default one first as a file or to ensure that it will |
| 9732 | always be the first one in the directory. |
Alex Davies | 0fbf016 | 2013-03-02 16:04:50 +0000 | [diff] [blame] | 9733 | |
Emeric Brun | e032bfa | 2012-09-28 13:01:45 +0200 | [diff] [blame] | 9734 | Note that the same cert may be loaded multiple times without side effects. |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9735 | |
Alex Davies | 0fbf016 | 2013-03-02 16:04:50 +0000 | [diff] [blame] | 9736 | Some CAs (such as Godaddy) offer a drop down list of server types that do not |
| 9737 | include HAProxy when obtaining a certificate. If this happens be sure to |
Godbach | 8bf60a1 | 2014-04-21 21:42:41 +0800 | [diff] [blame] | 9738 | choose a webserver that the CA believes requires an intermediate CA (for |
Alex Davies | 0fbf016 | 2013-03-02 16:04:50 +0000 | [diff] [blame] | 9739 | Godaddy, selection Apache Tomcat will get the correct bundle, but many |
| 9740 | others, e.g. nginx, result in a wrong bundle that will not work for some |
| 9741 | clients). |
| 9742 | |
Emeric Brun | 4147b2e | 2014-06-16 18:36:30 +0200 | [diff] [blame] | 9743 | For each PEM file, haproxy checks for the presence of file at the same path |
| 9744 | suffixed by ".ocsp". If such file is found, support for the TLS Certificate |
| 9745 | Status Request extension (also known as "OCSP stapling") is automatically |
| 9746 | enabled. The content of this file is optional. If not empty, it must contain |
| 9747 | a valid OCSP Response in DER format. In order to be valid an OCSP Response |
| 9748 | must comply with the following rules: it has to indicate a good status, |
| 9749 | it has to be a single response for the certificate of the PEM file, and it |
| 9750 | has to be valid at the moment of addition. If these rules are not respected |
| 9751 | the OCSP Response is ignored and a warning is emitted. In order to identify |
| 9752 | which certificate an OCSP Response applies to, the issuer's certificate is |
| 9753 | necessary. If the issuer's certificate is not found in the PEM file, it will |
| 9754 | be loaded from a file at the same path as the PEM file suffixed by ".issuer" |
| 9755 | if it exists otherwise it will fail with an error. |
| 9756 | |
Janusz Dziemidowicz | 2c701b5 | 2015-03-07 23:03:59 +0100 | [diff] [blame] | 9757 | For each PEM file, haproxy also checks for the presence of file at the same |
| 9758 | path suffixed by ".sctl". If such file is found, support for Certificate |
| 9759 | Transparency (RFC6962) TLS extension is enabled. The file must contain a |
| 9760 | valid Signed Certificate Timestamp List, as described in RFC. File is parsed |
| 9761 | to check basic syntax, but no signatures are verified. |
| 9762 | |
yanbzhu | 6c25e9e | 2016-01-05 12:52:02 -0500 | [diff] [blame] | 9763 | There are cases where it is desirable to support multiple key types, e.g. RSA |
| 9764 | and ECDSA in the cipher suites offered to the clients. This allows clients |
| 9765 | that support EC certificates to be able to use EC ciphers, while |
| 9766 | simultaneously supporting older, RSA only clients. |
yanbzhu | d19630c | 2015-12-14 15:10:25 -0500 | [diff] [blame] | 9767 | |
| 9768 | In order to provide this functionality, multiple PEM files, each with a |
| 9769 | different key type, are required. To associate these PEM files into a |
| 9770 | "cert bundle" that is recognized by haproxy, they must be named in the |
| 9771 | following way: All PEM files that are to be bundled must have the same base |
| 9772 | name, with a suffix indicating the key type. Currently, three suffixes are |
| 9773 | supported: rsa, dsa and ecdsa. For example, if www.example.com has two PEM |
| 9774 | files, an RSA file and an ECDSA file, they must be named: "example.pem.rsa" |
| 9775 | and "example.pem.ecdsa". The first part of the filename is arbitrary; only the |
| 9776 | suffix matters. To load this bundle into haproxy, specify the base name only: |
| 9777 | |
| 9778 | Example : bind :8443 ssl crt example.pem |
| 9779 | |
yanbzhu | 6c25e9e | 2016-01-05 12:52:02 -0500 | [diff] [blame] | 9780 | Note that the suffix is not given to haproxy; this tells haproxy to look for |
yanbzhu | d19630c | 2015-12-14 15:10:25 -0500 | [diff] [blame] | 9781 | a cert bundle. |
| 9782 | |
| 9783 | Haproxy will load all PEM files in the bundle at the same time to try to |
| 9784 | support multiple key types. PEM files are combined based on Common Name |
| 9785 | (CN) and Subject Alternative Name (SAN) to support SNI lookups. This means |
| 9786 | that even if you give haproxy a cert bundle, if there are no shared CN/SAN |
| 9787 | entries in the certificates in that bundle, haproxy will not be able to |
| 9788 | provide multi-cert support. |
| 9789 | |
| 9790 | Assuming bundle in the example above contained the following: |
| 9791 | |
| 9792 | Filename | CN | SAN |
| 9793 | -------------------+-----------------+------------------- |
| 9794 | example.pem.rsa | www.example.com | rsa.example.com |
yanbzhu | 6c25e9e | 2016-01-05 12:52:02 -0500 | [diff] [blame] | 9795 | -------------------+-----------------+------------------- |
yanbzhu | d19630c | 2015-12-14 15:10:25 -0500 | [diff] [blame] | 9796 | example.pem.ecdsa | www.example.com | ecdsa.example.com |
| 9797 | -------------------+-----------------+------------------- |
| 9798 | |
| 9799 | Users connecting with an SNI of "www.example.com" will be able |
| 9800 | to use both RSA and ECDSA cipher suites. Users connecting with an SNI of |
| 9801 | "rsa.example.com" will only be able to use RSA cipher suites, and users |
| 9802 | connecting with "ecdsa.example.com" will only be able to use ECDSA cipher |
| 9803 | suites. |
| 9804 | |
| 9805 | If a directory name is given as the <cert> argument, haproxy will |
| 9806 | automatically search and load bundled files in that directory. |
| 9807 | |
| 9808 | OSCP files (.ocsp) and issuer files (.issuer) are supported with multi-cert |
| 9809 | bundling. Each certificate can have its own .ocsp and .issuer file. At this |
| 9810 | time, sctl is not supported in multi-certificate bundling. |
| 9811 | |
Emeric Brun | b6dc934 | 2012-09-28 17:55:37 +0200 | [diff] [blame] | 9812 | crt-ignore-err <errors> |
Alex Davies | 0fbf016 | 2013-03-02 16:04:50 +0000 | [diff] [blame] | 9813 | This setting is only available when support for OpenSSL was built in. Sets a |
| 9814 | comma separated list of errorIDs to ignore during verify at depth == 0. If |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 9815 | set to 'all', all errors are ignored. SSL handshake is not aborted if an error |
Alex Davies | 0fbf016 | 2013-03-02 16:04:50 +0000 | [diff] [blame] | 9816 | is ignored. |
Emeric Brun | b6dc934 | 2012-09-28 17:55:37 +0200 | [diff] [blame] | 9817 | |
Emmanuel Hocdet | fe61656 | 2013-01-22 15:31:15 +0100 | [diff] [blame] | 9818 | crt-list <file> |
| 9819 | This setting is only available when support for OpenSSL was built in. It |
Emmanuel Hocdet | 7c41a1b | 2013-05-07 20:20:06 +0200 | [diff] [blame] | 9820 | designates a list of PEM file with an optional list of SNI filter per |
| 9821 | certificate, with the following format for each line : |
Emmanuel Hocdet | fe61656 | 2013-01-22 15:31:15 +0100 | [diff] [blame] | 9822 | |
Emmanuel Hocdet | 7c41a1b | 2013-05-07 20:20:06 +0200 | [diff] [blame] | 9823 | <crtfile> [[!]<snifilter> ...] |
Emmanuel Hocdet | fe61656 | 2013-01-22 15:31:15 +0100 | [diff] [blame] | 9824 | |
Emmanuel Hocdet | 7c41a1b | 2013-05-07 20:20:06 +0200 | [diff] [blame] | 9825 | Wildcards are supported in the SNI filter. Negative filter are also supported, |
| 9826 | only useful in combination with a wildcard filter to exclude a particular SNI. |
| 9827 | The certificates will be presented to clients who provide a valid TLS Server |
| 9828 | Name Indication field matching one of the SNI filters. If no SNI filter is |
| 9829 | specified, the CN and alt subjects are used. This directive may be specified |
| 9830 | multiple times. See the "crt" option for more information. The default |
| 9831 | certificate is still needed to meet OpenSSL expectations. If it is not used, |
| 9832 | the 'strict-sni' option may be used. |
Emmanuel Hocdet | fe61656 | 2013-01-22 15:31:15 +0100 | [diff] [blame] | 9833 | |
yanbzhu | 6c25e9e | 2016-01-05 12:52:02 -0500 | [diff] [blame] | 9834 | Multi-cert bundling (see "crt") is supported with crt-list, as long as only |
| 9835 | the base name is given in the crt-list. Due to the nature of bundling, all SNI |
yanbzhu | d19630c | 2015-12-14 15:10:25 -0500 | [diff] [blame] | 9836 | filters given to a multi-cert bundle entry are ignored. |
| 9837 | |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9838 | defer-accept |
| 9839 | Is an optional keyword which is supported only on certain Linux kernels. It |
| 9840 | states that a connection will only be accepted once some data arrive on it, |
| 9841 | or at worst after the first retransmit. This should be used only on protocols |
| 9842 | for which the client talks first (eg: HTTP). It can slightly improve |
| 9843 | performance by ensuring that most of the request is already available when |
| 9844 | the connection is accepted. On the other hand, it will not be able to detect |
| 9845 | connections which don't talk. It is important to note that this option is |
| 9846 | broken in all kernels up to 2.6.31, as the connection is never accepted until |
| 9847 | the client talks. This can cause issues with front firewalls which would see |
| 9848 | an established connection while the proxy will only see it in SYN_RECV. This |
| 9849 | option is only supported on TCPv4/TCPv6 sockets and ignored by other ones. |
| 9850 | |
Emeric Brun | 2cb7ae5 | 2012-10-05 14:14:21 +0200 | [diff] [blame] | 9851 | force-sslv3 |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 9852 | This option enforces use of SSLv3 only on SSL connections instantiated from |
Emeric Brun | 2cb7ae5 | 2012-10-05 14:14:21 +0200 | [diff] [blame] | 9853 | this listener. SSLv3 is generally less expensive than the TLS counterparts |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 9854 | for high connection rates. This option is also available on global statement |
| 9855 | "ssl-default-bind-options". See also "no-tlsv*" and "no-sslv3". |
Emeric Brun | 2cb7ae5 | 2012-10-05 14:14:21 +0200 | [diff] [blame] | 9856 | |
| 9857 | force-tlsv10 |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 9858 | This option enforces use of TLSv1.0 only on SSL connections instantiated from |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 9859 | this listener. This option is also available on global statement |
| 9860 | "ssl-default-bind-options". See also "no-tlsv*" and "no-sslv3". |
Emeric Brun | 2cb7ae5 | 2012-10-05 14:14:21 +0200 | [diff] [blame] | 9861 | |
| 9862 | force-tlsv11 |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 9863 | This option enforces use of TLSv1.1 only on SSL connections instantiated from |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 9864 | this listener. This option is also available on global statement |
| 9865 | "ssl-default-bind-options". See also "no-tlsv*", and "no-sslv3". |
Emeric Brun | 2cb7ae5 | 2012-10-05 14:14:21 +0200 | [diff] [blame] | 9866 | |
| 9867 | force-tlsv12 |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 9868 | This option enforces use of TLSv1.2 only on SSL connections instantiated from |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 9869 | this listener. This option is also available on global statement |
| 9870 | "ssl-default-bind-options". See also "no-tlsv*", and "no-sslv3". |
Emeric Brun | 2cb7ae5 | 2012-10-05 14:14:21 +0200 | [diff] [blame] | 9871 | |
Christopher Faulet | 31af49d | 2015-06-09 17:29:50 +0200 | [diff] [blame] | 9872 | generate-certificates |
| 9873 | This setting is only available when support for OpenSSL was built in. It |
| 9874 | enables the dynamic SSL certificates generation. A CA certificate and its |
| 9875 | private key are necessary (see 'ca-sign-file'). When HAProxy is configured as |
| 9876 | a transparent forward proxy, SSL requests generate errors because of a common |
| 9877 | name mismatch on the certificate presented to the client. With this option |
| 9878 | enabled, HAProxy will try to forge a certificate using the SNI hostname |
| 9879 | indicated by the client. This is done only if no certificate matches the SNI |
| 9880 | hostname (see 'crt-list'). If an error occurs, the default certificate is |
| 9881 | used, else the 'strict-sni' option is set. |
| 9882 | It can also be used when HAProxy is configured as a reverse proxy to ease the |
| 9883 | deployment of an architecture with many backends. |
| 9884 | |
| 9885 | Creating a SSL certificate is an expensive operation, so a LRU cache is used |
| 9886 | to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It |
| 9887 | increases the HAProxy's memroy footprint to reduce latency when the same |
| 9888 | certificate is used many times. |
| 9889 | |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9890 | gid <gid> |
| 9891 | Sets the group of the UNIX sockets to the designated system gid. It can also |
| 9892 | be set by default in the global section's "unix-bind" statement. Note that |
| 9893 | some platforms simply ignore this. This setting is equivalent to the "group" |
| 9894 | setting except that the group ID is used instead of its name. This setting is |
| 9895 | ignored by non UNIX sockets. |
| 9896 | |
| 9897 | group <group> |
| 9898 | Sets the group of the UNIX sockets to the designated system group. It can |
| 9899 | also be set by default in the global section's "unix-bind" statement. Note |
| 9900 | that some platforms simply ignore this. This setting is equivalent to the |
| 9901 | "gid" setting except that the group name is used instead of its gid. This |
| 9902 | setting is ignored by non UNIX sockets. |
| 9903 | |
| 9904 | id <id> |
| 9905 | Fixes the socket ID. By default, socket IDs are automatically assigned, but |
| 9906 | sometimes it is more convenient to fix them to ease monitoring. This value |
| 9907 | must be strictly positive and unique within the listener/frontend. This |
| 9908 | option can only be used when defining only a single socket. |
| 9909 | |
| 9910 | interface <interface> |
Lukas Tribus | fce2e96 | 2013-02-12 22:13:19 +0100 | [diff] [blame] | 9911 | Restricts the socket to a specific interface. When specified, only packets |
| 9912 | received from that particular interface are processed by the socket. This is |
| 9913 | currently only supported on Linux. The interface must be a primary system |
| 9914 | interface, not an aliased interface. It is also possible to bind multiple |
| 9915 | frontends to the same address if they are bound to different interfaces. Note |
| 9916 | that binding to a network interface requires root privileges. This parameter |
| 9917 | is only compatible with TCPv4/TCPv6 sockets. |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9918 | |
Willy Tarreau | abb175f | 2012-09-24 12:43:26 +0200 | [diff] [blame] | 9919 | level <level> |
| 9920 | This setting is used with the stats sockets only to restrict the nature of |
| 9921 | the commands that can be issued on the socket. It is ignored by other |
| 9922 | sockets. <level> can be one of : |
| 9923 | - "user" is the least privileged level ; only non-sensitive stats can be |
| 9924 | read, and no change is allowed. It would make sense on systems where it |
| 9925 | is not easy to restrict access to the socket. |
| 9926 | - "operator" is the default level and fits most common uses. All data can |
| 9927 | be read, and only non-sensitive changes are permitted (eg: clear max |
| 9928 | counters). |
| 9929 | - "admin" should be used with care, as everything is permitted (eg: clear |
| 9930 | all counters). |
| 9931 | |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9932 | maxconn <maxconn> |
| 9933 | Limits the sockets to this number of concurrent connections. Extraneous |
| 9934 | connections will remain in the system's backlog until a connection is |
| 9935 | released. If unspecified, the limit will be the same as the frontend's |
| 9936 | maxconn. Note that in case of port ranges or multiple addresses, the same |
| 9937 | value will be applied to each socket. This setting enables different |
| 9938 | limitations on expensive sockets, for instance SSL entries which may easily |
| 9939 | eat all memory. |
| 9940 | |
| 9941 | mode <mode> |
| 9942 | Sets the octal mode used to define access permissions on the UNIX socket. It |
| 9943 | can also be set by default in the global section's "unix-bind" statement. |
| 9944 | Note that some platforms simply ignore this. This setting is ignored by non |
| 9945 | UNIX sockets. |
| 9946 | |
| 9947 | mss <maxseg> |
| 9948 | Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming |
| 9949 | connections. This can be used to force a lower MSS for certain specific |
| 9950 | ports, for instance for connections passing through a VPN. Note that this |
| 9951 | relies on a kernel feature which is theoretically supported under Linux but |
| 9952 | was buggy in all versions prior to 2.6.28. It may or may not work on other |
| 9953 | operating systems. It may also not change the advertised value but change the |
| 9954 | effective size of outgoing segments. The commonly advertised value for TCPv4 |
| 9955 | over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is |
| 9956 | positive, it will be used as the advertised MSS. If it is negative, it will |
| 9957 | indicate by how much to reduce the incoming connection's advertised MSS for |
| 9958 | outgoing segments. This parameter is only compatible with TCP v4/v6 sockets. |
| 9959 | |
| 9960 | name <name> |
| 9961 | Sets an optional name for these sockets, which will be reported on the stats |
| 9962 | page. |
| 9963 | |
Willy Tarreau | d72f0f3 | 2015-10-13 14:50:22 +0200 | [diff] [blame] | 9964 | namespace <name> |
| 9965 | On Linux, it is possible to specify which network namespace a socket will |
| 9966 | belong to. This directive makes it possible to explicitly bind a listener to |
| 9967 | a namespace different from the default one. Please refer to your operating |
| 9968 | system's documentation to find more details about network namespaces. |
| 9969 | |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9970 | nice <nice> |
| 9971 | Sets the 'niceness' of connections initiated from the socket. Value must be |
| 9972 | in the range -1024..1024 inclusive, and defaults to zero. Positive values |
| 9973 | means that such connections are more friendly to others and easily offer |
| 9974 | their place in the scheduler. On the opposite, negative values mean that |
| 9975 | connections want to run with a higher priority than others. The difference |
| 9976 | only happens under high loads when the system is close to saturation. |
| 9977 | Negative values are appropriate for low-latency or administration services, |
| 9978 | and high values are generally recommended for CPU intensive tasks such as SSL |
| 9979 | processing or bulk transfers which are less sensible to latency. For example, |
| 9980 | it may make sense to use a positive value for an SMTP socket and a negative |
| 9981 | one for an RDP socket. |
| 9982 | |
Emeric Brun | 9b3009b | 2012-10-05 11:55:06 +0200 | [diff] [blame] | 9983 | no-sslv3 |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9984 | This setting is only available when support for OpenSSL was built in. It |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 9985 | disables support for SSLv3 on any sockets instantiated from the listener when |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9986 | SSL is supported. Note that SSLv2 is forced disabled in the code and cannot |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 9987 | be enabled using any configuration option. This option is also available on |
| 9988 | global statement "ssl-default-bind-options". See also "force-tls*", |
Emeric Brun | 2cb7ae5 | 2012-10-05 14:14:21 +0200 | [diff] [blame] | 9989 | and "force-sslv3". |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9990 | |
Emeric Brun | 90ad872 | 2012-10-02 14:00:59 +0200 | [diff] [blame] | 9991 | no-tls-tickets |
| 9992 | This setting is only available when support for OpenSSL was built in. It |
| 9993 | disables the stateless session resumption (RFC 5077 TLS Ticket |
| 9994 | extension) and force to use stateful session resumption. Stateless |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 9995 | session resumption is more expensive in CPU usage. This option is also |
| 9996 | available on global statement "ssl-default-bind-options". |
Emeric Brun | 90ad872 | 2012-10-02 14:00:59 +0200 | [diff] [blame] | 9997 | |
Emeric Brun | 9b3009b | 2012-10-05 11:55:06 +0200 | [diff] [blame] | 9998 | no-tlsv10 |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 9999 | This setting is only available when support for OpenSSL was built in. It |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 10000 | disables support for TLSv1.0 on any sockets instantiated from the listener |
Emeric Brun | 2cb7ae5 | 2012-10-05 14:14:21 +0200 | [diff] [blame] | 10001 | when SSL is supported. Note that SSLv2 is forced disabled in the code and |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 10002 | cannot be enabled using any configuration option. This option is also |
| 10003 | available on global statement "ssl-default-bind-options". See also |
| 10004 | "force-tlsv*", and "force-sslv3". |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 10005 | |
Emeric Brun | 9b3009b | 2012-10-05 11:55:06 +0200 | [diff] [blame] | 10006 | no-tlsv11 |
Emeric Brun | f5da493 | 2012-09-28 19:42:54 +0200 | [diff] [blame] | 10007 | This setting is only available when support for OpenSSL was built in. It |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 10008 | disables support for TLSv1.1 on any sockets instantiated from the listener |
Emeric Brun | 2cb7ae5 | 2012-10-05 14:14:21 +0200 | [diff] [blame] | 10009 | when SSL is supported. Note that SSLv2 is forced disabled in the code and |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 10010 | cannot be enabled using any configuration option. This option is also |
| 10011 | available on global statement "ssl-default-bind-options". See also |
| 10012 | "force-tlsv*", and "force-sslv3". |
Emeric Brun | f5da493 | 2012-09-28 19:42:54 +0200 | [diff] [blame] | 10013 | |
Emeric Brun | 9b3009b | 2012-10-05 11:55:06 +0200 | [diff] [blame] | 10014 | no-tlsv12 |
Emeric Brun | f5da493 | 2012-09-28 19:42:54 +0200 | [diff] [blame] | 10015 | This setting is only available when support for OpenSSL was built in. It |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 10016 | disables support for TLSv1.2 on any sockets instantiated from the listener |
Emeric Brun | 2cb7ae5 | 2012-10-05 14:14:21 +0200 | [diff] [blame] | 10017 | when SSL is supported. Note that SSLv2 is forced disabled in the code and |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 10018 | cannot be enabled using any configuration option. This option is also |
| 10019 | available on global statement "ssl-default-bind-options". See also |
| 10020 | "force-tlsv*", and "force-sslv3". |
Emeric Brun | f5da493 | 2012-09-28 19:42:54 +0200 | [diff] [blame] | 10021 | |
Willy Tarreau | 6c9a3d5 | 2012-10-18 18:57:14 +0200 | [diff] [blame] | 10022 | npn <protocols> |
| 10023 | This enables the NPN TLS extension and advertises the specified protocol list |
| 10024 | as supported on top of NPN. The protocol list consists in a comma-delimited |
| 10025 | list of protocol names, for instance: "http/1.1,http/1.0" (without quotes). |
| 10026 | This requires that the SSL library is build with support for TLS extensions |
Willy Tarreau | ab861d3 | 2013-04-02 02:30:41 +0200 | [diff] [blame] | 10027 | enabled (check with haproxy -vv). Note that the NPN extension has been |
| 10028 | replaced with the ALPN extension (see the "alpn" keyword). |
Willy Tarreau | 6c9a3d5 | 2012-10-18 18:57:14 +0200 | [diff] [blame] | 10029 | |
Willy Tarreau | 6ae1ba6 | 2014-05-07 19:01:58 +0200 | [diff] [blame] | 10030 | process [ all | odd | even | <number 1-64>[-<number 1-64>] ] |
| 10031 | This restricts the list of processes on which this listener is allowed to |
| 10032 | run. It does not enforce any process but eliminates those which do not match. |
| 10033 | If the frontend uses a "bind-process" setting, the intersection between the |
| 10034 | two is applied. If in the end the listener is not allowed to run on any |
| 10035 | remaining process, a warning is emitted, and the listener will either run on |
| 10036 | the first process of the listener if a single process was specified, or on |
| 10037 | all of its processes if multiple processes were specified. For the unlikely |
Willy Tarreau | ae30253 | 2014-05-07 19:22:24 +0200 | [diff] [blame] | 10038 | case where several ranges are needed, this directive may be repeated. The |
| 10039 | main purpose of this directive is to be used with the stats sockets and have |
| 10040 | one different socket per process. The second purpose is to have multiple bind |
| 10041 | lines sharing the same IP:port but not the same process in a listener, so |
| 10042 | that the system can distribute the incoming connections into multiple queues |
| 10043 | and allow a smoother inter-process load balancing. Currently Linux 3.9 and |
| 10044 | above is known for supporting this. See also "bind-process" and "nbproc". |
Willy Tarreau | 6ae1ba6 | 2014-05-07 19:01:58 +0200 | [diff] [blame] | 10045 | |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 10046 | ssl |
| 10047 | This setting is only available when support for OpenSSL was built in. It |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 10048 | enables SSL deciphering on connections instantiated from this listener. A |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 10049 | certificate is necessary (see "crt" above). All contents in the buffers will |
| 10050 | appear in clear text, so that ACLs and HTTP processing will only have access |
| 10051 | to deciphered contents. |
| 10052 | |
Emmanuel Hocdet | 6562337 | 2013-01-24 17:17:15 +0100 | [diff] [blame] | 10053 | strict-sni |
| 10054 | This setting is only available when support for OpenSSL was built in. The |
| 10055 | SSL/TLS negotiation is allow only if the client provided an SNI which match |
| 10056 | a certificate. The default certificate is not used. |
| 10057 | See the "crt" option for more information. |
| 10058 | |
Willy Tarreau | 2af207a | 2015-02-04 00:45:58 +0100 | [diff] [blame] | 10059 | tcp-ut <delay> |
| 10060 | Sets the TCP User Timeout for all incoming connections instanciated from this |
| 10061 | listening socket. This option is available on Linux since version 2.6.37. It |
| 10062 | allows haproxy to configure a timeout for sockets which contain data not |
| 10063 | receiving an acknoledgement for the configured delay. This is especially |
| 10064 | useful on long-lived connections experiencing long idle periods such as |
| 10065 | remote terminals or database connection pools, where the client and server |
| 10066 | timeouts must remain high to allow a long period of idle, but where it is |
| 10067 | important to detect that the client has disappeared in order to release all |
| 10068 | resources associated with its connection (and the server's session). The |
| 10069 | argument is a delay expressed in milliseconds by default. This only works |
| 10070 | for regular TCP connections, and is ignored for other protocols. |
| 10071 | |
Willy Tarreau | 1c862c5 | 2012-10-05 16:21:00 +0200 | [diff] [blame] | 10072 | tfo |
Lukas Tribus | 0defb90 | 2013-02-13 23:35:39 +0100 | [diff] [blame] | 10073 | Is an optional keyword which is supported only on Linux kernels >= 3.7. It |
Willy Tarreau | 1c862c5 | 2012-10-05 16:21:00 +0200 | [diff] [blame] | 10074 | enables TCP Fast Open on the listening socket, which means that clients which |
| 10075 | support this feature will be able to send a request and receive a response |
| 10076 | during the 3-way handshake starting from second connection, thus saving one |
| 10077 | round-trip after the first connection. This only makes sense with protocols |
| 10078 | that use high connection rates and where each round trip matters. This can |
| 10079 | possibly cause issues with many firewalls which do not accept data on SYN |
| 10080 | packets, so this option should only be enabled once well tested. This option |
Lukas Tribus | 0999f76 | 2013-04-02 16:43:24 +0200 | [diff] [blame] | 10081 | is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may |
| 10082 | need to build HAProxy with USE_TFO=1 if your libc doesn't define |
| 10083 | TCP_FASTOPEN. |
Willy Tarreau | 1c862c5 | 2012-10-05 16:21:00 +0200 | [diff] [blame] | 10084 | |
Nenad Merdanovic | 188ad3e | 2015-02-27 19:56:50 +0100 | [diff] [blame] | 10085 | tls-ticket-keys <keyfile> |
| 10086 | Sets the TLS ticket keys file to load the keys from. The keys need to be 48 |
| 10087 | bytes long, encoded with base64 (ex. openssl rand -base64 48). Number of keys |
| 10088 | is specified by the TLS_TICKETS_NO build option (default 3) and at least as |
| 10089 | many keys need to be present in the file. Last TLS_TICKETS_NO keys will be |
| 10090 | used for decryption and the penultimate one for encryption. This enables easy |
| 10091 | key rotation by just appending new key to the file and reloading the process. |
| 10092 | Keys must be periodically rotated (ex. every 12h) or Perfect Forward Secrecy |
| 10093 | is compromised. It is also a good idea to keep the keys off any permanent |
| 10094 | storage such as hard drives (hint: use tmpfs and don't swap those files). |
| 10095 | Lifetime hint can be changed using tune.ssl.timeout. |
| 10096 | |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 10097 | transparent |
| 10098 | Is an optional keyword which is supported only on certain Linux kernels. It |
| 10099 | indicates that the addresses will be bound even if they do not belong to the |
| 10100 | local machine, and that packets targeting any of these addresses will be |
| 10101 | intercepted just as if the addresses were locally configured. This normally |
| 10102 | requires that IP forwarding is enabled. Caution! do not use this with the |
| 10103 | default address '*', as it would redirect any traffic for the specified port. |
| 10104 | This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1. |
| 10105 | This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on |
| 10106 | kernel version. Some distribution kernels include backports of the feature, |
| 10107 | so check for support with your vendor. |
| 10108 | |
Willy Tarreau | 77e3af9 | 2012-11-24 15:07:23 +0100 | [diff] [blame] | 10109 | v4v6 |
| 10110 | Is an optional keyword which is supported only on most recent systems |
| 10111 | including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4 |
| 10112 | and IPv6 when it uses the default address. Doing so is sometimes necessary |
| 10113 | on systems which bind to IPv6 only by default. It has no effect on non-IPv6 |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 10114 | sockets, and is overridden by the "v6only" option. |
Willy Tarreau | 77e3af9 | 2012-11-24 15:07:23 +0100 | [diff] [blame] | 10115 | |
Willy Tarreau | 9b6700f | 2012-11-24 11:55:28 +0100 | [diff] [blame] | 10116 | v6only |
| 10117 | Is an optional keyword which is supported only on most recent systems |
| 10118 | including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only |
| 10119 | when it uses the default address. Doing so is sometimes preferred to doing it |
Willy Tarreau | 77e3af9 | 2012-11-24 15:07:23 +0100 | [diff] [blame] | 10120 | system-wide as it is per-listener. It has no effect on non-IPv6 sockets and |
| 10121 | has precedence over the "v4v6" option. |
Willy Tarreau | 9b6700f | 2012-11-24 11:55:28 +0100 | [diff] [blame] | 10122 | |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 10123 | uid <uid> |
| 10124 | Sets the owner of the UNIX sockets to the designated system uid. It can also |
| 10125 | be set by default in the global section's "unix-bind" statement. Note that |
| 10126 | some platforms simply ignore this. This setting is equivalent to the "user" |
| 10127 | setting except that the user numeric ID is used instead of its name. This |
| 10128 | setting is ignored by non UNIX sockets. |
| 10129 | |
| 10130 | user <user> |
| 10131 | Sets the owner of the UNIX sockets to the designated system user. It can also |
| 10132 | be set by default in the global section's "unix-bind" statement. Note that |
| 10133 | some platforms simply ignore this. This setting is equivalent to the "uid" |
| 10134 | setting except that the user name is used instead of its uid. This setting is |
| 10135 | ignored by non UNIX sockets. |
| 10136 | |
Emeric Brun | 1a073b4 | 2012-09-28 17:07:34 +0200 | [diff] [blame] | 10137 | verify [none|optional|required] |
| 10138 | This setting is only available when support for OpenSSL was built in. If set |
| 10139 | to 'none', client certificate is not requested. This is the default. In other |
| 10140 | cases, a client certificate is requested. If the client does not provide a |
| 10141 | certificate after the request and if 'verify' is set to 'required', then the |
| 10142 | handshake is aborted, while it would have succeeded if set to 'optional'. The |
Emeric Brun | fd33a26 | 2012-10-11 16:28:27 +0200 | [diff] [blame] | 10143 | certificate provided by the client is always verified using CAs from |
| 10144 | 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake |
| 10145 | is aborted, regardless of the 'verify' option, unless the error code exactly |
| 10146 | matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'. |
Willy Tarreau | 4a5cade | 2012-04-05 21:09:48 +0200 | [diff] [blame] | 10147 | |
Willy Tarreau | b6205fd | 2012-09-24 12:27:33 +0200 | [diff] [blame] | 10148 | 5.2. Server and default-server options |
Cyril Bonté | f0c6061 | 2010-02-06 14:44:47 +0100 | [diff] [blame] | 10149 | ------------------------------------ |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 10150 | |
Krzysztof Piotr Oledzki | c6df066 | 2010-01-05 16:38:49 +0100 | [diff] [blame] | 10151 | The "server" and "default-server" keywords support a certain number of settings |
| 10152 | which are all passed as arguments on the server line. The order in which those |
| 10153 | arguments appear does not count, and they are all optional. Some of those |
| 10154 | settings are single words (booleans) while others expect one or several values |
| 10155 | after them. In this case, the values must immediately follow the setting name. |
| 10156 | Except default-server, all those settings must be specified after the server's |
| 10157 | address if they are used: |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 10158 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10159 | server <name> <address>[:port] [settings ...] |
Krzysztof Piotr Oledzki | c6df066 | 2010-01-05 16:38:49 +0100 | [diff] [blame] | 10160 | default-server [settings ...] |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 10161 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10162 | The currently supported settings are the following ones. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 10163 | |
Willy Tarreau | ceb4ac9 | 2012-04-28 00:41:46 +0200 | [diff] [blame] | 10164 | addr <ipv4|ipv6> |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10165 | Using the "addr" parameter, it becomes possible to use a different IP address |
Baptiste Assmann | 13f8353 | 2016-03-06 23:14:36 +0100 | [diff] [blame] | 10166 | to send health-checks or to probe the agent-check. On some servers, it may be |
| 10167 | desirable to dedicate an IP address to specific component able to perform |
| 10168 | complex tests which are more suitable to health-checks than the application. |
| 10169 | This parameter is ignored if the "check" parameter is not set. See also the |
| 10170 | "port" parameter. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 10171 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10172 | Supported in default-server: No |
| 10173 | |
Simon Horman | d60d691 | 2013-11-25 10:46:36 +0900 | [diff] [blame] | 10174 | agent-check |
| 10175 | Enable an auxiliary agent check which is run independently of a regular |
Willy Tarreau | 81f5d94 | 2013-12-09 20:51:51 +0100 | [diff] [blame] | 10176 | health check. An agent health check is performed by making a TCP connection |
| 10177 | to the port set by the "agent-port" parameter and reading an ASCII string. |
| 10178 | The string is made of a series of words delimited by spaces, tabs or commas |
| 10179 | in any order, optionally terminated by '\r' and/or '\n', each consisting of : |
Simon Horman | d60d691 | 2013-11-25 10:46:36 +0900 | [diff] [blame] | 10180 | |
Willy Tarreau | 81f5d94 | 2013-12-09 20:51:51 +0100 | [diff] [blame] | 10181 | - An ASCII representation of a positive integer percentage, e.g. "75%". |
Simon Horman | d60d691 | 2013-11-25 10:46:36 +0900 | [diff] [blame] | 10182 | Values in this format will set the weight proportional to the initial |
Willy Tarreau | c5af3a6 | 2014-10-07 15:27:33 +0200 | [diff] [blame] | 10183 | weight of a server as configured when haproxy starts. Note that a zero |
| 10184 | weight is reported on the stats page as "DRAIN" since it has the same |
| 10185 | effect on the server (it's removed from the LB farm). |
Simon Horman | d60d691 | 2013-11-25 10:46:36 +0900 | [diff] [blame] | 10186 | |
Nenad Merdanovic | 174dd37 | 2016-04-24 23:10:06 +0200 | [diff] [blame] | 10187 | - The string "maxconn:" followed by an integer (no space between). Values in |
| 10188 | this format will set the maxconn of a server. The maximum number of |
| 10189 | connections advertised needs to be multipled by the number of load balancers |
| 10190 | and different backends that use this health check to get the total number |
| 10191 | of connections the server might receive. Example: maxconn:30 |
| 10192 | |
Willy Tarreau | 81f5d94 | 2013-12-09 20:51:51 +0100 | [diff] [blame] | 10193 | - The word "ready". This will turn the server's administrative state to the |
| 10194 | READY mode, thus cancelling any DRAIN or MAINT state |
Simon Horman | d60d691 | 2013-11-25 10:46:36 +0900 | [diff] [blame] | 10195 | |
Willy Tarreau | 81f5d94 | 2013-12-09 20:51:51 +0100 | [diff] [blame] | 10196 | - The word "drain". This will turn the server's administrative state to the |
| 10197 | DRAIN mode, thus it will not accept any new connections other than those |
| 10198 | that are accepted via persistence. |
Simon Horman | d60d691 | 2013-11-25 10:46:36 +0900 | [diff] [blame] | 10199 | |
Willy Tarreau | 81f5d94 | 2013-12-09 20:51:51 +0100 | [diff] [blame] | 10200 | - The word "maint". This will turn the server's administrative state to the |
| 10201 | MAINT mode, thus it will not accept any new connections at all, and health |
| 10202 | checks will be stopped. |
Simon Horman | d60d691 | 2013-11-25 10:46:36 +0900 | [diff] [blame] | 10203 | |
Willy Tarreau | 81f5d94 | 2013-12-09 20:51:51 +0100 | [diff] [blame] | 10204 | - The words "down", "failed", or "stopped", optionally followed by a |
| 10205 | description string after a sharp ('#'). All of these mark the server's |
| 10206 | operating state as DOWN, but since the word itself is reported on the stats |
| 10207 | page, the difference allows an administrator to know if the situation was |
| 10208 | expected or not : the service may intentionally be stopped, may appear up |
| 10209 | but fail some validity tests, or may be seen as down (eg: missing process, |
| 10210 | or port not responding). |
Simon Horman | d60d691 | 2013-11-25 10:46:36 +0900 | [diff] [blame] | 10211 | |
Willy Tarreau | 81f5d94 | 2013-12-09 20:51:51 +0100 | [diff] [blame] | 10212 | - The word "up" sets back the server's operating state as UP if health checks |
| 10213 | also report that the service is accessible. |
Simon Horman | d60d691 | 2013-11-25 10:46:36 +0900 | [diff] [blame] | 10214 | |
Willy Tarreau | 81f5d94 | 2013-12-09 20:51:51 +0100 | [diff] [blame] | 10215 | Parameters which are not advertised by the agent are not changed. For |
| 10216 | example, an agent might be designed to monitor CPU usage and only report a |
| 10217 | relative weight and never interact with the operating status. Similarly, an |
| 10218 | agent could be designed as an end-user interface with 3 radio buttons |
| 10219 | allowing an administrator to change only the administrative state. However, |
| 10220 | it is important to consider that only the agent may revert its own actions, |
| 10221 | so if a server is set to DRAIN mode or to DOWN state using the agent, the |
| 10222 | agent must implement the other equivalent actions to bring the service into |
| 10223 | operations again. |
Simon Horman | d60d691 | 2013-11-25 10:46:36 +0900 | [diff] [blame] | 10224 | |
Simon Horman | 2f1f955 | 2013-11-25 10:46:37 +0900 | [diff] [blame] | 10225 | Failure to connect to the agent is not considered an error as connectivity |
| 10226 | is tested by the regular health check which is enabled by the "check" |
Willy Tarreau | 81f5d94 | 2013-12-09 20:51:51 +0100 | [diff] [blame] | 10227 | parameter. Warning though, it is not a good idea to stop an agent after it |
| 10228 | reports "down", since only an agent reporting "up" will be able to turn the |
| 10229 | server up again. Note that the CLI on the Unix stats socket is also able to |
Willy Tarreau | 989222a | 2016-01-15 10:26:26 +0100 | [diff] [blame] | 10230 | force an agent's result in order to work around a bogus agent if needed. |
Simon Horman | 2f1f955 | 2013-11-25 10:46:37 +0900 | [diff] [blame] | 10231 | |
Willy Tarreau | 81f5d94 | 2013-12-09 20:51:51 +0100 | [diff] [blame] | 10232 | Requires the "agent-port" parameter to be set. See also the "agent-inter" |
| 10233 | parameter. |
Simon Horman | d60d691 | 2013-11-25 10:46:36 +0900 | [diff] [blame] | 10234 | |
| 10235 | Supported in default-server: No |
| 10236 | |
James Brown | 55f9ff1 | 2015-10-21 18:19:05 -0700 | [diff] [blame] | 10237 | agent-send <string> |
| 10238 | If this option is specified, haproxy will send the given string (verbatim) |
| 10239 | to the agent server upon connection. You could, for example, encode |
| 10240 | the backend name into this string, which would enable your agent to send |
| 10241 | different responses based on the backend. Make sure to include a '\n' if |
| 10242 | you want to terminate your request with a newline. |
| 10243 | |
Simon Horman | d60d691 | 2013-11-25 10:46:36 +0900 | [diff] [blame] | 10244 | agent-inter <delay> |
| 10245 | The "agent-inter" parameter sets the interval between two agent checks |
| 10246 | to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms. |
| 10247 | |
| 10248 | Just as with every other time-based parameter, it may be entered in any |
| 10249 | other explicit unit among { us, ms, s, m, h, d }. The "agent-inter" |
| 10250 | parameter also serves as a timeout for agent checks "timeout check" is |
| 10251 | not set. In order to reduce "resonance" effects when multiple servers are |
| 10252 | hosted on the same hardware, the agent and health checks of all servers |
| 10253 | are started with a small time offset between them. It is also possible to |
| 10254 | add some random noise in the agent and health checks interval using the |
| 10255 | global "spread-checks" keyword. This makes sense for instance when a lot |
| 10256 | of backends use the same servers. |
| 10257 | |
| 10258 | See also the "agent-check" and "agent-port" parameters. |
| 10259 | |
| 10260 | Supported in default-server: Yes |
| 10261 | |
| 10262 | agent-port <port> |
| 10263 | The "agent-port" parameter sets the TCP port used for agent checks. |
| 10264 | |
| 10265 | See also the "agent-check" and "agent-inter" parameters. |
| 10266 | |
| 10267 | Supported in default-server: Yes |
| 10268 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10269 | backup |
| 10270 | When "backup" is present on a server line, the server is only used in load |
| 10271 | balancing when all other non-backup servers are unavailable. Requests coming |
| 10272 | with a persistence cookie referencing the server will always be served |
| 10273 | though. By default, only the first operational backup server is used, unless |
| 10274 | the "allbackups" option is set in the backend. See also the "allbackups" |
| 10275 | option. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 10276 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10277 | Supported in default-server: No |
| 10278 | |
Emeric Brun | ef42d92 | 2012-10-11 16:11:36 +0200 | [diff] [blame] | 10279 | ca-file <cafile> |
| 10280 | This setting is only available when support for OpenSSL was built in. It |
| 10281 | designates a PEM file from which to load CA certificates used to verify |
| 10282 | server's certificate. |
| 10283 | |
| 10284 | Supported in default-server: No |
| 10285 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10286 | check |
| 10287 | This option enables health checks on the server. By default, a server is |
Patrick Mézard | b7aeec6 | 2012-01-22 16:01:22 +0100 | [diff] [blame] | 10288 | always considered available. If "check" is set, the server is available when |
| 10289 | accepting periodic TCP connections, to ensure that it is really able to serve |
| 10290 | requests. The default address and port to send the tests to are those of the |
| 10291 | server, and the default source is the same as the one defined in the |
| 10292 | backend. It is possible to change the address using the "addr" parameter, the |
| 10293 | port using the "port" parameter, the source address using the "source" |
| 10294 | address, and the interval and timers using the "inter", "rise" and "fall" |
Simon Horman | afc47ee | 2013-11-25 10:46:35 +0900 | [diff] [blame] | 10295 | parameters. The request method is define in the backend using the "httpchk", |
| 10296 | "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please |
| 10297 | refer to those options and parameters for more information. |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10298 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10299 | Supported in default-server: No |
| 10300 | |
Willy Tarreau | 6c16adc | 2012-10-05 00:04:16 +0200 | [diff] [blame] | 10301 | check-send-proxy |
| 10302 | This option forces emission of a PROXY protocol line with outgoing health |
| 10303 | checks, regardless of whether the server uses send-proxy or not for the |
| 10304 | normal traffic. By default, the PROXY protocol is enabled for health checks |
| 10305 | if it is already enabled for normal traffic and if no "port" nor "addr" |
| 10306 | directive is present. However, if such a directive is present, the |
| 10307 | "check-send-proxy" option needs to be used to force the use of the |
| 10308 | protocol. See also the "send-proxy" option for more information. |
| 10309 | |
| 10310 | Supported in default-server: No |
| 10311 | |
Willy Tarreau | 763a95b | 2012-10-04 23:15:39 +0200 | [diff] [blame] | 10312 | check-ssl |
| 10313 | This option forces encryption of all health checks over SSL, regardless of |
| 10314 | whether the server uses SSL or not for the normal traffic. This is generally |
| 10315 | used when an explicit "port" or "addr" directive is specified and SSL health |
| 10316 | checks are not inherited. It is important to understand that this option |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 10317 | inserts an SSL transport layer below the checks, so that a simple TCP connect |
Willy Tarreau | 763a95b | 2012-10-04 23:15:39 +0200 | [diff] [blame] | 10318 | check becomes an SSL connect, which replaces the old ssl-hello-chk. The most |
| 10319 | common use is to send HTTPS checks by combining "httpchk" with SSL checks. |
| 10320 | All SSL settings are common to health checks and traffic (eg: ciphers). |
| 10321 | See the "ssl" option for more information. |
| 10322 | |
| 10323 | Supported in default-server: No |
| 10324 | |
Willy Tarreau | a0ee1d0 | 2012-09-10 09:01:23 +0200 | [diff] [blame] | 10325 | ciphers <ciphers> |
| 10326 | This option sets the string describing the list of cipher algorithms that is |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 10327 | is negotiated during the SSL/TLS handshake with the server. The format of the |
Willy Tarreau | a0ee1d0 | 2012-09-10 09:01:23 +0200 | [diff] [blame] | 10328 | string is defined in "man 1 ciphers". When SSL is used to communicate with |
| 10329 | servers on the local network, it is common to see a weaker set of algorithms |
| 10330 | than what is used over the internet. Doing so reduces CPU usage on both the |
| 10331 | server and haproxy while still keeping it compatible with deployed software. |
| 10332 | Some algorithms such as RC4-SHA1 are reasonably cheap. If no security at all |
| 10333 | is needed and just connectivity, using DES can be appropriate. |
| 10334 | |
Willy Tarreau | 763a95b | 2012-10-04 23:15:39 +0200 | [diff] [blame] | 10335 | Supported in default-server: No |
| 10336 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10337 | cookie <value> |
| 10338 | The "cookie" parameter sets the cookie value assigned to the server to |
| 10339 | <value>. This value will be checked in incoming requests, and the first |
| 10340 | operational server possessing the same value will be selected. In return, in |
| 10341 | cookie insertion or rewrite modes, this value will be assigned to the cookie |
| 10342 | sent to the client. There is nothing wrong in having several servers sharing |
| 10343 | the same cookie value, and it is in fact somewhat common between normal and |
| 10344 | backup servers. See also the "cookie" keyword in backend section. |
| 10345 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10346 | Supported in default-server: No |
| 10347 | |
Emeric Brun | ef42d92 | 2012-10-11 16:11:36 +0200 | [diff] [blame] | 10348 | crl-file <crlfile> |
| 10349 | This setting is only available when support for OpenSSL was built in. It |
| 10350 | designates a PEM file from which to load certificate revocation list used |
| 10351 | to verify server's certificate. |
| 10352 | |
| 10353 | Supported in default-server: No |
| 10354 | |
Emeric Brun | a7aa309 | 2012-10-26 12:58:00 +0200 | [diff] [blame] | 10355 | crt <cert> |
| 10356 | This setting is only available when support for OpenSSL was built in. |
| 10357 | It designates a PEM file from which to load both a certificate and the |
| 10358 | associated private key. This file can be built by concatenating both PEM |
| 10359 | files into one. This certificate will be sent if the server send a client |
| 10360 | certificate request. |
| 10361 | |
| 10362 | Supported in default-server: No |
| 10363 | |
Willy Tarreau | 9683909 | 2010-03-29 10:02:24 +0200 | [diff] [blame] | 10364 | disabled |
| 10365 | The "disabled" keyword starts the server in the "disabled" state. That means |
| 10366 | that it is marked down in maintenance mode, and no connection other than the |
| 10367 | ones allowed by persist mode will reach it. It is very well suited to setup |
| 10368 | new servers, because normal traffic will never reach them, while it is still |
| 10369 | possible to test the service by making use of the force-persist mechanism. |
| 10370 | |
| 10371 | Supported in default-server: No |
| 10372 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10373 | error-limit <count> |
Willy Tarreau | 983e01e | 2010-01-11 18:42:06 +0100 | [diff] [blame] | 10374 | If health observing is enabled, the "error-limit" parameter specifies the |
| 10375 | number of consecutive errors that triggers event selected by the "on-error" |
| 10376 | option. By default it is set to 10 consecutive errors. |
Krzysztof Piotr Oledzki | 97f07b8 | 2009-12-15 22:31:24 +0100 | [diff] [blame] | 10377 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10378 | Supported in default-server: Yes |
| 10379 | |
| 10380 | See also the "check", "error-limit" and "on-error". |
Krzysztof Piotr Oledzki | 97f07b8 | 2009-12-15 22:31:24 +0100 | [diff] [blame] | 10381 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10382 | fall <count> |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10383 | The "fall" parameter states that a server will be considered as dead after |
| 10384 | <count> consecutive unsuccessful health checks. This value defaults to 3 if |
| 10385 | unspecified. See also the "check", "inter" and "rise" parameters. |
| 10386 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10387 | Supported in default-server: Yes |
| 10388 | |
Emeric Brun | 8694b9a | 2012-10-05 14:39:07 +0200 | [diff] [blame] | 10389 | force-sslv3 |
| 10390 | This option enforces use of SSLv3 only when SSL is used to communicate with |
| 10391 | the server. SSLv3 is generally less expensive than the TLS counterparts for |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 10392 | high connection rates. This option is also available on global statement |
| 10393 | "ssl-default-server-options". See also "no-tlsv*", "no-sslv3". |
Emeric Brun | 8694b9a | 2012-10-05 14:39:07 +0200 | [diff] [blame] | 10394 | |
| 10395 | Supported in default-server: No |
| 10396 | |
| 10397 | force-tlsv10 |
| 10398 | This option enforces use of TLSv1.0 only when SSL is used to communicate with |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 10399 | the server. This option is also available on global statement |
| 10400 | "ssl-default-server-options". See also "no-tlsv*", "no-sslv3". |
Emeric Brun | 8694b9a | 2012-10-05 14:39:07 +0200 | [diff] [blame] | 10401 | |
| 10402 | Supported in default-server: No |
| 10403 | |
| 10404 | force-tlsv11 |
| 10405 | This option enforces use of TLSv1.1 only when SSL is used to communicate with |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 10406 | the server. This option is also available on global statement |
| 10407 | "ssl-default-server-options". See also "no-tlsv*", "no-sslv3". |
Emeric Brun | 8694b9a | 2012-10-05 14:39:07 +0200 | [diff] [blame] | 10408 | |
| 10409 | Supported in default-server: No |
| 10410 | |
| 10411 | force-tlsv12 |
| 10412 | This option enforces use of TLSv1.2 only when SSL is used to communicate with |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 10413 | the server. This option is also available on global statement |
| 10414 | "ssl-default-server-options". See also "no-tlsv*", "no-sslv3". |
Emeric Brun | 8694b9a | 2012-10-05 14:39:07 +0200 | [diff] [blame] | 10415 | |
| 10416 | Supported in default-server: No |
| 10417 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10418 | id <value> |
Willy Tarreau | 53fb4ae | 2009-10-04 23:04:08 +0200 | [diff] [blame] | 10419 | Set a persistent ID for the server. This ID must be positive and unique for |
| 10420 | the proxy. An unused ID will automatically be assigned if unset. The first |
| 10421 | assigned value will be 1. This ID is currently only returned in statistics. |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10422 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10423 | Supported in default-server: No |
| 10424 | |
| 10425 | inter <delay> |
| 10426 | fastinter <delay> |
| 10427 | downinter <delay> |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10428 | The "inter" parameter sets the interval between two consecutive health checks |
| 10429 | to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms. |
| 10430 | It is also possible to use "fastinter" and "downinter" to optimize delays |
| 10431 | between checks depending on the server state : |
| 10432 | |
Pieter Baauw | 44fc9df | 2015-09-17 21:30:46 +0200 | [diff] [blame] | 10433 | Server state | Interval used |
| 10434 | ----------------------------------------+---------------------------------- |
| 10435 | UP 100% (non-transitional) | "inter" |
| 10436 | ----------------------------------------+---------------------------------- |
| 10437 | Transitionally UP (going down "fall"), | "fastinter" if set, |
| 10438 | Transitionally DOWN (going up "rise"), | "inter" otherwise. |
| 10439 | or yet unchecked. | |
| 10440 | ----------------------------------------+---------------------------------- |
| 10441 | DOWN 100% (non-transitional) | "downinter" if set, |
| 10442 | | "inter" otherwise. |
| 10443 | ----------------------------------------+---------------------------------- |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 10444 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10445 | Just as with every other time-based parameter, they can be entered in any |
| 10446 | other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also |
| 10447 | serves as a timeout for health checks sent to servers if "timeout check" is |
| 10448 | not set. In order to reduce "resonance" effects when multiple servers are |
Simon Horman | d60d691 | 2013-11-25 10:46:36 +0900 | [diff] [blame] | 10449 | hosted on the same hardware, the agent and health checks of all servers |
| 10450 | are started with a small time offset between them. It is also possible to |
| 10451 | add some random noise in the agent and health checks interval using the |
| 10452 | global "spread-checks" keyword. This makes sense for instance when a lot |
| 10453 | of backends use the same servers. |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10454 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10455 | Supported in default-server: Yes |
| 10456 | |
| 10457 | maxconn <maxconn> |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10458 | The "maxconn" parameter specifies the maximal number of concurrent |
| 10459 | connections that will be sent to this server. If the number of incoming |
| 10460 | concurrent requests goes higher than this value, they will be queued, waiting |
| 10461 | for a connection to be released. This parameter is very important as it can |
| 10462 | save fragile servers from going down under extreme loads. If a "minconn" |
| 10463 | parameter is specified, the limit becomes dynamic. The default value is "0" |
| 10464 | which means unlimited. See also the "minconn" and "maxqueue" parameters, and |
| 10465 | the backend's "fullconn" keyword. |
| 10466 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10467 | Supported in default-server: Yes |
| 10468 | |
| 10469 | maxqueue <maxqueue> |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10470 | The "maxqueue" parameter specifies the maximal number of connections which |
| 10471 | will wait in the queue for this server. If this limit is reached, next |
| 10472 | requests will be redispatched to other servers instead of indefinitely |
| 10473 | waiting to be served. This will break persistence but may allow people to |
| 10474 | quickly re-log in when the server they try to connect to is dying. The |
| 10475 | default value is "0" which means the queue is unlimited. See also the |
| 10476 | "maxconn" and "minconn" parameters. |
| 10477 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10478 | Supported in default-server: Yes |
| 10479 | |
| 10480 | minconn <minconn> |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10481 | When the "minconn" parameter is set, the maxconn limit becomes a dynamic |
| 10482 | limit following the backend's load. The server will always accept at least |
| 10483 | <minconn> connections, never more than <maxconn>, and the limit will be on |
| 10484 | the ramp between both values when the backend has less than <fullconn> |
| 10485 | concurrent connections. This makes it possible to limit the load on the |
| 10486 | server during normal loads, but push it further for important loads without |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 10487 | overloading the server during exceptional loads. See also the "maxconn" |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10488 | and "maxqueue" parameters, as well as the "fullconn" backend keyword. |
Krzysztof Piotr Oledzki | 97f07b8 | 2009-12-15 22:31:24 +0100 | [diff] [blame] | 10489 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10490 | Supported in default-server: Yes |
| 10491 | |
Willy Tarreau | d72f0f3 | 2015-10-13 14:50:22 +0200 | [diff] [blame] | 10492 | namespace <name> |
| 10493 | On Linux, it is possible to specify which network namespace a socket will |
| 10494 | belong to. This directive makes it possible to explicitly bind a server to |
| 10495 | a namespace different from the default one. Please refer to your operating |
| 10496 | system's documentation to find more details about network namespaces. |
| 10497 | |
Willy Tarreau | 2a3fb1c | 2015-02-05 16:47:07 +0100 | [diff] [blame] | 10498 | no-ssl-reuse |
| 10499 | This option disables SSL session reuse when SSL is used to communicate with |
| 10500 | the server. It will force the server to perform a full handshake for every |
| 10501 | new connection. It's probably only useful for benchmarking, troubleshooting, |
| 10502 | and for paranoid users. |
| 10503 | |
| 10504 | Supported in default-server: No |
| 10505 | |
Emeric Brun | 9b3009b | 2012-10-05 11:55:06 +0200 | [diff] [blame] | 10506 | no-sslv3 |
Willy Tarreau | a0ee1d0 | 2012-09-10 09:01:23 +0200 | [diff] [blame] | 10507 | This option disables support for SSLv3 when SSL is used to communicate with |
| 10508 | the server. Note that SSLv2 is disabled in the code and cannot be enabled |
Emeric Brun | 8694b9a | 2012-10-05 14:39:07 +0200 | [diff] [blame] | 10509 | using any configuration option. See also "force-sslv3", "force-tlsv*". |
Willy Tarreau | a0ee1d0 | 2012-09-10 09:01:23 +0200 | [diff] [blame] | 10510 | |
Willy Tarreau | 763a95b | 2012-10-04 23:15:39 +0200 | [diff] [blame] | 10511 | Supported in default-server: No |
| 10512 | |
Emeric Brun | f9c5c47 | 2012-10-11 15:28:34 +0200 | [diff] [blame] | 10513 | no-tls-tickets |
| 10514 | This setting is only available when support for OpenSSL was built in. It |
| 10515 | disables the stateless session resumption (RFC 5077 TLS Ticket |
| 10516 | extension) and force to use stateful session resumption. Stateless |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 10517 | session resumption is more expensive in CPU usage for servers. This option |
| 10518 | is also available on global statement "ssl-default-server-options". |
Emeric Brun | f9c5c47 | 2012-10-11 15:28:34 +0200 | [diff] [blame] | 10519 | |
| 10520 | Supported in default-server: No |
| 10521 | |
Emeric Brun | 9b3009b | 2012-10-05 11:55:06 +0200 | [diff] [blame] | 10522 | no-tlsv10 |
Emeric Brun | 8694b9a | 2012-10-05 14:39:07 +0200 | [diff] [blame] | 10523 | This option disables support for TLSv1.0 when SSL is used to communicate with |
Emeric Brun | f5da493 | 2012-09-28 19:42:54 +0200 | [diff] [blame] | 10524 | the server. Note that SSLv2 is disabled in the code and cannot be enabled |
| 10525 | using any configuration option. TLSv1 is more expensive than SSLv3 so it |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 10526 | often makes sense to disable it when communicating with local servers. This |
| 10527 | option is also available on global statement "ssl-default-server-options". |
| 10528 | See also "force-sslv3", "force-tlsv*". |
Emeric Brun | f5da493 | 2012-09-28 19:42:54 +0200 | [diff] [blame] | 10529 | |
Willy Tarreau | 763a95b | 2012-10-04 23:15:39 +0200 | [diff] [blame] | 10530 | Supported in default-server: No |
| 10531 | |
Emeric Brun | 9b3009b | 2012-10-05 11:55:06 +0200 | [diff] [blame] | 10532 | no-tlsv11 |
Emeric Brun | 8694b9a | 2012-10-05 14:39:07 +0200 | [diff] [blame] | 10533 | This option disables support for TLSv1.1 when SSL is used to communicate with |
Emeric Brun | f5da493 | 2012-09-28 19:42:54 +0200 | [diff] [blame] | 10534 | the server. Note that SSLv2 is disabled in the code and cannot be enabled |
| 10535 | using any configuration option. TLSv1 is more expensive than SSLv3 so it |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 10536 | often makes sense to disable it when communicating with local servers. This |
| 10537 | option is also available on global statement "ssl-default-server-options". |
| 10538 | See also "force-sslv3", "force-tlsv*". |
Emeric Brun | f5da493 | 2012-09-28 19:42:54 +0200 | [diff] [blame] | 10539 | |
Willy Tarreau | 763a95b | 2012-10-04 23:15:39 +0200 | [diff] [blame] | 10540 | Supported in default-server: No |
| 10541 | |
Emeric Brun | 9b3009b | 2012-10-05 11:55:06 +0200 | [diff] [blame] | 10542 | no-tlsv12 |
Emeric Brun | 8694b9a | 2012-10-05 14:39:07 +0200 | [diff] [blame] | 10543 | This option disables support for TLSv1.2 when SSL is used to communicate with |
Willy Tarreau | a0ee1d0 | 2012-09-10 09:01:23 +0200 | [diff] [blame] | 10544 | the server. Note that SSLv2 is disabled in the code and cannot be enabled |
| 10545 | using any configuration option. TLSv1 is more expensive than SSLv3 so it |
Emeric Brun | 2c86cbf | 2014-10-30 15:56:50 +0100 | [diff] [blame] | 10546 | often makes sense to disable it when communicating with local servers. This |
| 10547 | option is also available on global statement "ssl-default-server-options". |
| 10548 | See also "force-sslv3", "force-tlsv*". |
Willy Tarreau | a0ee1d0 | 2012-09-10 09:01:23 +0200 | [diff] [blame] | 10549 | |
Willy Tarreau | 763a95b | 2012-10-04 23:15:39 +0200 | [diff] [blame] | 10550 | Supported in default-server: No |
| 10551 | |
Simon Horman | fa46168 | 2011-06-25 09:39:49 +0900 | [diff] [blame] | 10552 | non-stick |
| 10553 | Never add connections allocated to this sever to a stick-table. |
| 10554 | This may be used in conjunction with backup to ensure that |
| 10555 | stick-table persistence is disabled for backup servers. |
| 10556 | |
Willy Tarreau | 763a95b | 2012-10-04 23:15:39 +0200 | [diff] [blame] | 10557 | Supported in default-server: No |
| 10558 | |
Krzysztof Piotr Oledzki | 97f07b8 | 2009-12-15 22:31:24 +0100 | [diff] [blame] | 10559 | observe <mode> |
| 10560 | This option enables health adjusting based on observing communication with |
| 10561 | the server. By default this functionality is disabled and enabling it also |
| 10562 | requires to enable health checks. There are two supported modes: "layer4" and |
| 10563 | "layer7". In layer4 mode, only successful/unsuccessful tcp connections are |
| 10564 | significant. In layer7, which is only allowed for http proxies, responses |
| 10565 | received from server are verified, like valid/wrong http code, unparsable |
Willy Tarreau | 150d146 | 2012-03-10 08:19:02 +0100 | [diff] [blame] | 10566 | headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505. |
Krzysztof Piotr Oledzki | 97f07b8 | 2009-12-15 22:31:24 +0100 | [diff] [blame] | 10567 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10568 | Supported in default-server: No |
| 10569 | |
Krzysztof Piotr Oledzki | 97f07b8 | 2009-12-15 22:31:24 +0100 | [diff] [blame] | 10570 | See also the "check", "on-error" and "error-limit". |
| 10571 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10572 | on-error <mode> |
Krzysztof Piotr Oledzki | 97f07b8 | 2009-12-15 22:31:24 +0100 | [diff] [blame] | 10573 | Select what should happen when enough consecutive errors are detected. |
| 10574 | Currently, four modes are available: |
| 10575 | - fastinter: force fastinter |
| 10576 | - fail-check: simulate a failed check, also forces fastinter (default) |
| 10577 | - sudden-death: simulate a pre-fatal failed health check, one more failed |
| 10578 | check will mark a server down, forces fastinter |
| 10579 | - mark-down: mark the server immediately down and force fastinter |
| 10580 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10581 | Supported in default-server: Yes |
| 10582 | |
Krzysztof Piotr Oledzki | 97f07b8 | 2009-12-15 22:31:24 +0100 | [diff] [blame] | 10583 | See also the "check", "observe" and "error-limit". |
| 10584 | |
Simon Horman | e0d1bfb | 2011-06-21 14:34:58 +0900 | [diff] [blame] | 10585 | on-marked-down <action> |
| 10586 | Modify what occurs when a server is marked down. |
| 10587 | Currently one action is available: |
Justin Karneges | eb2c24a | 2012-05-24 15:28:52 -0700 | [diff] [blame] | 10588 | - shutdown-sessions: Shutdown peer sessions. When this setting is enabled, |
| 10589 | all connections to the server are immediately terminated when the server |
| 10590 | goes down. It might be used if the health check detects more complex cases |
| 10591 | than a simple connection status, and long timeouts would cause the service |
| 10592 | to remain unresponsive for too long a time. For instance, a health check |
| 10593 | might detect that a database is stuck and that there's no chance to reuse |
| 10594 | existing connections anymore. Connections killed this way are logged with |
| 10595 | a 'D' termination code (for "Down"). |
Simon Horman | e0d1bfb | 2011-06-21 14:34:58 +0900 | [diff] [blame] | 10596 | |
| 10597 | Actions are disabled by default |
| 10598 | |
| 10599 | Supported in default-server: Yes |
| 10600 | |
Justin Karneges | eb2c24a | 2012-05-24 15:28:52 -0700 | [diff] [blame] | 10601 | on-marked-up <action> |
| 10602 | Modify what occurs when a server is marked up. |
| 10603 | Currently one action is available: |
| 10604 | - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is |
| 10605 | done only if the server is not in backup state and if it is not disabled |
| 10606 | (it must have an effective weight > 0). This can be used sometimes to force |
| 10607 | an active server to take all the traffic back after recovery when dealing |
| 10608 | with long sessions (eg: LDAP, SQL, ...). Doing this can cause more trouble |
| 10609 | than it tries to solve (eg: incomplete transactions), so use this feature |
| 10610 | with extreme care. Sessions killed because a server comes up are logged |
| 10611 | with an 'U' termination code (for "Up"). |
| 10612 | |
| 10613 | Actions are disabled by default |
| 10614 | |
| 10615 | Supported in default-server: Yes |
| 10616 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10617 | port <port> |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10618 | Using the "port" parameter, it becomes possible to use a different port to |
| 10619 | send health-checks. On some servers, it may be desirable to dedicate a port |
| 10620 | to a specific component able to perform complex tests which are more suitable |
| 10621 | to health-checks than the application. It is common to run a simple script in |
| 10622 | inetd for instance. This parameter is ignored if the "check" parameter is not |
| 10623 | set. See also the "addr" parameter. |
| 10624 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10625 | Supported in default-server: Yes |
| 10626 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10627 | redir <prefix> |
| 10628 | The "redir" parameter enables the redirection mode for all GET and HEAD |
| 10629 | requests addressing this server. This means that instead of having HAProxy |
| 10630 | forward the request to the server, it will send an "HTTP 302" response with |
| 10631 | the "Location" header composed of this prefix immediately followed by the |
| 10632 | requested URI beginning at the leading '/' of the path component. That means |
| 10633 | that no trailing slash should be used after <prefix>. All invalid requests |
| 10634 | will be rejected, and all non-GET or HEAD requests will be normally served by |
| 10635 | the server. Note that since the response is completely forged, no header |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 10636 | mangling nor cookie insertion is possible in the response. However, cookies in |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10637 | requests are still analysed, making this solution completely usable to direct |
| 10638 | users to a remote location in case of local disaster. Main use consists in |
| 10639 | increasing bandwidth for static servers by having the clients directly |
| 10640 | connect to them. Note: never use a relative location here, it would cause a |
| 10641 | loop between the client and HAProxy! |
| 10642 | |
| 10643 | Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check |
| 10644 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10645 | Supported in default-server: No |
| 10646 | |
| 10647 | rise <count> |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10648 | The "rise" parameter states that a server will be considered as operational |
| 10649 | after <count> consecutive successful health checks. This value defaults to 2 |
| 10650 | if unspecified. See also the "check", "inter" and "fall" parameters. |
| 10651 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10652 | Supported in default-server: Yes |
| 10653 | |
Baptiste Assmann | 1fa6666 | 2015-04-14 00:28:47 +0200 | [diff] [blame] | 10654 | resolve-prefer <family> |
| 10655 | When DNS resolution is enabled for a server and multiple IP addresses from |
| 10656 | different families are returned, HAProxy will prefer using an IP address |
| 10657 | from the family mentioned in the "resolve-prefer" parameter. |
| 10658 | Available families: "ipv4" and "ipv6" |
| 10659 | |
Baptiste Assmann | c4aabae | 2015-08-04 22:43:06 +0200 | [diff] [blame] | 10660 | Default value: ipv6 |
| 10661 | |
| 10662 | Supported in default-server: Yes |
Baptiste Assmann | 1fa6666 | 2015-04-14 00:28:47 +0200 | [diff] [blame] | 10663 | |
| 10664 | Example: server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6 |
| 10665 | |
Thierry Fournier | ac88cfe | 2016-02-17 22:05:30 +0100 | [diff] [blame] | 10666 | resolve-net <network>[,<network[,...]] |
| 10667 | This options prioritize th choice of an ip address matching a network. This is |
| 10668 | useful with clouds to prefer a local ip. In some cases, a cloud high |
| 10669 | avalailibility service can be announced with many ip addresses on many |
| 10670 | differents datacenters. The latency between datacenter is not negligible, so |
| 10671 | this patch permitsto prefers a local datacenter. If none address matchs the |
| 10672 | configured network, another address is selected. |
| 10673 | |
| 10674 | Supported in default-server: Yes |
| 10675 | |
| 10676 | Example: server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8 |
| 10677 | |
Baptiste Assmann | 1fa6666 | 2015-04-14 00:28:47 +0200 | [diff] [blame] | 10678 | resolvers <id> |
| 10679 | Points to an existing "resolvers" section to resolve current server's |
| 10680 | hostname. |
Baptiste Assmann | 62b75b4 | 2015-09-09 01:11:36 +0200 | [diff] [blame] | 10681 | In order to be operational, DNS resolution requires that health check is |
| 10682 | enabled on the server. Actually, health checks triggers the DNS resolution. |
| 10683 | You must precise one 'resolvers' parameter on each server line where DNS |
| 10684 | resolution is required. |
Baptiste Assmann | 1fa6666 | 2015-04-14 00:28:47 +0200 | [diff] [blame] | 10685 | |
Baptiste Assmann | 62b75b4 | 2015-09-09 01:11:36 +0200 | [diff] [blame] | 10686 | Supported in default-server: No |
| 10687 | |
| 10688 | Example: server s1 app1.domain.com:80 check resolvers mydns |
Baptiste Assmann | 1fa6666 | 2015-04-14 00:28:47 +0200 | [diff] [blame] | 10689 | |
| 10690 | See also chapter 5.3 |
| 10691 | |
Willy Tarreau | 5ab04ec | 2011-03-20 10:32:26 +0100 | [diff] [blame] | 10692 | send-proxy |
| 10693 | The "send-proxy" parameter enforces use of the PROXY protocol over any |
| 10694 | connection established to this server. The PROXY protocol informs the other |
| 10695 | end about the layer 3/4 addresses of the incoming connection, so that it can |
| 10696 | know the client's address or the public address it accessed to, whatever the |
| 10697 | upper layer protocol. For connections accepted by an "accept-proxy" listener, |
| 10698 | the advertised address will be used. Only TCPv4 and TCPv6 address families |
| 10699 | are supported. Other families such as Unix sockets, will report an UNKNOWN |
| 10700 | family. Servers using this option can fully be chained to another instance of |
| 10701 | haproxy listening with an "accept-proxy" setting. This setting must not be |
Willy Tarreau | 6c16adc | 2012-10-05 00:04:16 +0200 | [diff] [blame] | 10702 | used if the server isn't aware of the protocol. When health checks are sent |
| 10703 | to the server, the PROXY protocol is automatically used when this option is |
| 10704 | set, unless there is an explicit "port" or "addr" directive, in which case an |
| 10705 | explicit "check-send-proxy" directive would also be needed to use the PROXY |
| 10706 | protocol. See also the "accept-proxy" option of the "bind" keyword. |
Willy Tarreau | 5ab04ec | 2011-03-20 10:32:26 +0100 | [diff] [blame] | 10707 | |
| 10708 | Supported in default-server: No |
| 10709 | |
David S | afb7683 | 2014-05-08 23:42:08 -0400 | [diff] [blame] | 10710 | send-proxy-v2 |
| 10711 | The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2 |
| 10712 | over any connection established to this server. The PROXY protocol informs |
| 10713 | the other end about the layer 3/4 addresses of the incoming connection, so |
| 10714 | that it can know the client's address or the public address it accessed to, |
| 10715 | whatever the upper layer protocol. This setting must not be used if the |
| 10716 | server isn't aware of this version of the protocol. See also the "send-proxy" |
| 10717 | option of the "bind" keyword. |
| 10718 | |
| 10719 | Supported in default-server: No |
| 10720 | |
| 10721 | send-proxy-v2-ssl |
| 10722 | The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version |
| 10723 | 2 over any connection established to this server. The PROXY protocol informs |
| 10724 | the other end about the layer 3/4 addresses of the incoming connection, so |
| 10725 | that it can know the client's address or the public address it accessed to, |
| 10726 | whatever the upper layer protocol. In addition, the SSL information extension |
| 10727 | of the PROXY protocol is added to the PROXY protocol header. This setting |
| 10728 | must not be used if the server isn't aware of this version of the protocol. |
| 10729 | See also the "send-proxy-v2" option of the "bind" keyword. |
| 10730 | |
| 10731 | Supported in default-server: No |
| 10732 | |
| 10733 | send-proxy-v2-ssl-cn |
| 10734 | The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version |
| 10735 | 2 over any connection established to this server. The PROXY protocol informs |
| 10736 | the other end about the layer 3/4 addresses of the incoming connection, so |
| 10737 | that it can know the client's address or the public address it accessed to, |
| 10738 | whatever the upper layer protocol. In addition, the SSL information extension |
| 10739 | of the PROXY protocol, along along with the Common Name from the subject of |
| 10740 | the client certificate (if any), is added to the PROXY protocol header. This |
| 10741 | setting must not be used if the server isn't aware of this version of the |
| 10742 | protocol. See also the "send-proxy-v2" option of the "bind" keyword. |
| 10743 | |
| 10744 | Supported in default-server: No |
| 10745 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10746 | slowstart <start_time_in_ms> |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10747 | The "slowstart" parameter for a server accepts a value in milliseconds which |
| 10748 | indicates after how long a server which has just come back up will run at |
| 10749 | full speed. Just as with every other time-based parameter, it can be entered |
| 10750 | in any other explicit unit among { us, ms, s, m, h, d }. The speed grows |
| 10751 | linearly from 0 to 100% during this time. The limitation applies to two |
| 10752 | parameters : |
| 10753 | |
| 10754 | - maxconn: the number of connections accepted by the server will grow from 1 |
| 10755 | to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn). |
| 10756 | |
| 10757 | - weight: when the backend uses a dynamic weighted algorithm, the weight |
| 10758 | grows linearly from 1 to 100%. In this case, the weight is updated at every |
| 10759 | health-check. For this reason, it is important that the "inter" parameter |
| 10760 | is smaller than the "slowstart", in order to maximize the number of steps. |
| 10761 | |
| 10762 | The slowstart never applies when haproxy starts, otherwise it would cause |
| 10763 | trouble to running servers. It only applies when a server has been previously |
| 10764 | seen as failed. |
| 10765 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10766 | Supported in default-server: Yes |
| 10767 | |
Willy Tarreau | 732eac4 | 2015-07-09 11:40:25 +0200 | [diff] [blame] | 10768 | sni <expression> |
| 10769 | The "sni" parameter evaluates the sample fetch expression, converts it to a |
| 10770 | string and uses the result as the host name sent in the SNI TLS extension to |
| 10771 | the server. A typical use case is to send the SNI received from the client in |
| 10772 | a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the |
| 10773 | expression, though alternatives such as req.hdr(host) can also make sense. |
| 10774 | |
| 10775 | Supported in default-server: no |
| 10776 | |
Willy Tarreau | c6f4ce8 | 2009-06-10 11:09:37 +0200 | [diff] [blame] | 10777 | source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ] |
Willy Tarreau | bce7088 | 2009-09-07 11:51:47 +0200 | [diff] [blame] | 10778 | source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ] |
Willy Tarreau | c6f4ce8 | 2009-06-10 11:09:37 +0200 | [diff] [blame] | 10779 | source <addr>[:<pl>[-<ph>]] [interface <name>] ... |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10780 | The "source" parameter sets the source address which will be used when |
| 10781 | connecting to the server. It follows the exact same parameters and principle |
| 10782 | as the backend "source" keyword, except that it only applies to the server |
| 10783 | referencing it. Please consult the "source" keyword for details. |
| 10784 | |
Willy Tarreau | c6f4ce8 | 2009-06-10 11:09:37 +0200 | [diff] [blame] | 10785 | Additionally, the "source" statement on a server line allows one to specify a |
| 10786 | source port range by indicating the lower and higher bounds delimited by a |
| 10787 | dash ('-'). Some operating systems might require a valid IP address when a |
| 10788 | source port range is specified. It is permitted to have the same IP/range for |
| 10789 | several servers. Doing so makes it possible to bypass the maximum of 64k |
| 10790 | total concurrent connections. The limit will then reach 64k connections per |
| 10791 | server. |
| 10792 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10793 | Supported in default-server: No |
| 10794 | |
Willy Tarreau | a0ee1d0 | 2012-09-10 09:01:23 +0200 | [diff] [blame] | 10795 | ssl |
Willy Tarreau | 44f6539 | 2013-06-25 07:56:20 +0200 | [diff] [blame] | 10796 | This option enables SSL ciphering on outgoing connections to the server. It |
| 10797 | is critical to verify server certificates using "verify" when using SSL to |
| 10798 | connect to servers, otherwise the communication is prone to trivial man in |
| 10799 | the-middle attacks rendering SSL useless. When this option is used, health |
| 10800 | checks are automatically sent in SSL too unless there is a "port" or an |
| 10801 | "addr" directive indicating the check should be sent to a different location. |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 10802 | See the "check-ssl" option to force SSL health checks. |
Willy Tarreau | 763a95b | 2012-10-04 23:15:39 +0200 | [diff] [blame] | 10803 | |
| 10804 | Supported in default-server: No |
Willy Tarreau | a0ee1d0 | 2012-09-10 09:01:23 +0200 | [diff] [blame] | 10805 | |
Willy Tarreau | 163d462 | 2015-10-13 16:16:41 +0200 | [diff] [blame] | 10806 | tcp-ut <delay> |
| 10807 | Sets the TCP User Timeout for all outgoing connections to this server. This |
| 10808 | option is available on Linux since version 2.6.37. It allows haproxy to |
| 10809 | configure a timeout for sockets which contain data not receiving an |
| 10810 | acknoledgement for the configured delay. This is especially useful on |
| 10811 | long-lived connections experiencing long idle periods such as remote |
| 10812 | terminals or database connection pools, where the client and server timeouts |
| 10813 | must remain high to allow a long period of idle, but where it is important to |
| 10814 | detect that the server has disappeared in order to release all resources |
| 10815 | associated with its connection (and the client's session). One typical use |
| 10816 | case is also to force dead server connections to die when health checks are |
| 10817 | too slow or during a soft reload since health checks are then disabled. The |
| 10818 | argument is a delay expressed in milliseconds by default. This only works for |
| 10819 | regular TCP connections, and is ignored for other protocols. |
| 10820 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10821 | track [<proxy>/]<server> |
Willy Tarreau | 3209123 | 2014-05-16 13:52:00 +0200 | [diff] [blame] | 10822 | This option enables ability to set the current state of the server by tracking |
| 10823 | another one. It is possible to track a server which itself tracks another |
| 10824 | server, provided that at the end of the chain, a server has health checks |
| 10825 | enabled. If <proxy> is omitted the current one is used. If disable-on-404 is |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10826 | used, it has to be enabled on both proxies. |
| 10827 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10828 | Supported in default-server: No |
| 10829 | |
Emeric Brun | ef42d92 | 2012-10-11 16:11:36 +0200 | [diff] [blame] | 10830 | verify [none|required] |
| 10831 | This setting is only available when support for OpenSSL was built in. If set |
Emeric Brun | 850efd5 | 2014-01-29 12:24:34 +0100 | [diff] [blame] | 10832 | to 'none', server certificate is not verified. In the other case, The |
| 10833 | certificate provided by the server is verified using CAs from 'ca-file' |
| 10834 | and optional CRLs from 'crl-file'. If 'ssl_server_verify' is not specified |
| 10835 | in global section, this is the default. On verify failure the handshake |
Willy Tarreau | 44f6539 | 2013-06-25 07:56:20 +0200 | [diff] [blame] | 10836 | is aborted. It is critically important to verify server certificates when |
| 10837 | using SSL to connect to servers, otherwise the communication is prone to |
| 10838 | trivial man-in-the-middle attacks rendering SSL totally useless. |
Emeric Brun | ef42d92 | 2012-10-11 16:11:36 +0200 | [diff] [blame] | 10839 | |
| 10840 | Supported in default-server: No |
| 10841 | |
Evan Broder | be55431 | 2013-06-27 00:05:25 -0700 | [diff] [blame] | 10842 | verifyhost <hostname> |
| 10843 | This setting is only available when support for OpenSSL was built in, and |
| 10844 | only takes effect if 'verify required' is also specified. When set, the |
| 10845 | hostnames in the subject and subjectAlternateNames of the certificate |
| 10846 | provided by the server are checked. If none of the hostnames in the |
| 10847 | certificate match the specified hostname, the handshake is aborted. The |
| 10848 | hostnames in the server-provided certificate may include wildcards. |
| 10849 | |
| 10850 | Supported in default-server: No |
| 10851 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10852 | weight <weight> |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10853 | The "weight" parameter is used to adjust the server's weight relative to |
| 10854 | other servers. All servers will receive a load proportional to their weight |
| 10855 | relative to the sum of all weights, so the higher the weight, the higher the |
Willy Tarreau | 6704d67 | 2009-06-15 10:56:05 +0200 | [diff] [blame] | 10856 | load. The default weight is 1, and the maximal value is 256. A value of 0 |
| 10857 | means the server will not participate in load-balancing but will still accept |
| 10858 | persistent connections. If this parameter is used to distribute the load |
| 10859 | according to server's capacity, it is recommended to start with values which |
| 10860 | can both grow and shrink, for instance between 10 and 100 to leave enough |
| 10861 | room above and below for later adjustments. |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10862 | |
Krzysztof Piotr Oledzki | c53601c | 2010-01-06 10:50:42 +0100 | [diff] [blame] | 10863 | Supported in default-server: Yes |
| 10864 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 10865 | |
Cyril Bonté | 46175dd | 2015-07-02 22:45:32 +0200 | [diff] [blame] | 10866 | 5.3. Server IP address resolution using DNS |
| 10867 | ------------------------------------------- |
Baptiste Assmann | 1fa6666 | 2015-04-14 00:28:47 +0200 | [diff] [blame] | 10868 | |
Baptiste Assmann | 62b75b4 | 2015-09-09 01:11:36 +0200 | [diff] [blame] | 10869 | HAProxy allows using a host name on the server line to retrieve its IP address |
| 10870 | using name servers. By default, HAProxy resolves the name when parsing the |
| 10871 | configuration file, at startup and cache the result for the process' life. |
Baptiste Assmann | 1fa6666 | 2015-04-14 00:28:47 +0200 | [diff] [blame] | 10872 | This is not sufficient in some cases, such as in Amazon where a server's IP |
| 10873 | can change after a reboot or an ELB Virtual IP can change based on current |
| 10874 | workload. |
| 10875 | This chapter describes how HAProxy can be configured to process server's name |
| 10876 | resolution at run time. |
| 10877 | Whether run time server name resolution has been enable or not, HAProxy will |
| 10878 | carry on doing the first resolution when parsing the configuration. |
| 10879 | |
Baptiste Assmann | 62b75b4 | 2015-09-09 01:11:36 +0200 | [diff] [blame] | 10880 | Bear in mind that DNS resolution is triggered by health checks. This makes |
| 10881 | health checks mandatory to allow DNS resolution. |
| 10882 | |
Baptiste Assmann | 1fa6666 | 2015-04-14 00:28:47 +0200 | [diff] [blame] | 10883 | |
Cyril Bonté | 46175dd | 2015-07-02 22:45:32 +0200 | [diff] [blame] | 10884 | 5.3.1. Global overview |
| 10885 | ---------------------- |
Baptiste Assmann | 1fa6666 | 2015-04-14 00:28:47 +0200 | [diff] [blame] | 10886 | |
| 10887 | As we've seen in introduction, name resolution in HAProxy occurs at two |
| 10888 | different steps of the process life: |
| 10889 | |
| 10890 | 1. when starting up, HAProxy parses the server line definition and matches a |
| 10891 | host name. It uses libc functions to get the host name resolved. This |
| 10892 | resolution relies on /etc/resolv.conf file. |
| 10893 | |
| 10894 | 2. at run time, when HAProxy gets prepared to run a health check on a server, |
| 10895 | it verifies if the current name resolution is still considered as valid. |
| 10896 | If not, it processes a new resolution, in parallel of the health check. |
| 10897 | |
| 10898 | A few other events can trigger a name resolution at run time: |
| 10899 | - when a server's health check ends up in a connection timeout: this may be |
| 10900 | because the server has a new IP address. So we need to trigger a name |
| 10901 | resolution to know this new IP. |
| 10902 | |
| 10903 | A few things important to notice: |
| 10904 | - all the name servers are queried in the mean time. HAProxy will process the |
| 10905 | first valid response. |
| 10906 | |
| 10907 | - a resolution is considered as invalid (NX, timeout, refused), when all the |
| 10908 | servers return an error. |
| 10909 | |
| 10910 | |
Cyril Bonté | 46175dd | 2015-07-02 22:45:32 +0200 | [diff] [blame] | 10911 | 5.3.2. The resolvers section |
| 10912 | ---------------------------- |
Baptiste Assmann | 1fa6666 | 2015-04-14 00:28:47 +0200 | [diff] [blame] | 10913 | |
| 10914 | This section is dedicated to host information related to name resolution in |
| 10915 | HAProxy. |
| 10916 | There can be as many as resolvers section as needed. Each section can contain |
| 10917 | many name servers. |
| 10918 | |
Baptiste Assmann | 62b75b4 | 2015-09-09 01:11:36 +0200 | [diff] [blame] | 10919 | When multiple name servers are configured in a resolvers section, then HAProxy |
| 10920 | uses the first valid response. In case of invalid responses, only the last one |
| 10921 | is treated. Purpose is to give the chance to a slow server to deliver a valid |
| 10922 | answer after a fast faulty or outdated server. |
| 10923 | |
| 10924 | When each server returns a different error type, then only the last error is |
| 10925 | used by HAProxy to decide what type of behavior to apply. |
| 10926 | |
| 10927 | Two types of behavior can be applied: |
| 10928 | 1. stop DNS resolution |
| 10929 | 2. replay the DNS query with a new query type |
| 10930 | In such case, the following types are applied in this exact order: |
| 10931 | 1. ANY query type |
| 10932 | 2. query type corresponding to family pointed by resolve-prefer |
| 10933 | server's parameter |
| 10934 | 3. remaining family type |
| 10935 | |
| 10936 | HAProxy stops DNS resolution when the following errors occur: |
| 10937 | - invalid DNS response packet |
| 10938 | - wrong name in the query section of the response |
| 10939 | - NX domain |
| 10940 | - Query refused by server |
| 10941 | - CNAME not pointing to an IP address |
| 10942 | |
| 10943 | HAProxy tries a new query type when the following errors occur: |
| 10944 | - no Answer records in the response |
| 10945 | - DNS response truncated |
| 10946 | - Error in DNS response |
| 10947 | - No expected DNS records found in the response |
| 10948 | - name server timeout |
| 10949 | |
| 10950 | For example, with 2 name servers configured in a resolvers section: |
| 10951 | - first response is valid and is applied directly, second response is ignored |
| 10952 | - first response is invalid and second one is valid, then second response is |
| 10953 | applied; |
| 10954 | - first response is a NX domain and second one a truncated response, then |
| 10955 | HAProxy replays the query with a new type; |
| 10956 | - first response is truncated and second one is a NX Domain, then HAProxy |
| 10957 | stops resolution. |
| 10958 | |
| 10959 | |
Baptiste Assmann | 1fa6666 | 2015-04-14 00:28:47 +0200 | [diff] [blame] | 10960 | resolvers <resolvers id> |
| 10961 | Creates a new name server list labelled <resolvers id> |
| 10962 | |
| 10963 | A resolvers section accept the following parameters: |
| 10964 | |
| 10965 | nameserver <id> <ip>:<port> |
| 10966 | DNS server description: |
| 10967 | <id> : label of the server, should be unique |
| 10968 | <ip> : IP address of the server |
| 10969 | <port> : port where the DNS service actually runs |
| 10970 | |
| 10971 | hold <status> <period> |
| 10972 | Defines <period> during which the last name resolution should be kept based |
| 10973 | on last resolution <status> |
| 10974 | <status> : last name resolution status. Only "valid" is accepted for now. |
| 10975 | <period> : interval between two successive name resolution when the last |
| 10976 | answer was in <status>. It follows the HAProxy time format. |
| 10977 | <period> is in milliseconds by default. |
| 10978 | |
| 10979 | Default value is 10s for "valid". |
| 10980 | |
| 10981 | Note: since the name resolution is triggered by the health checks, a new |
| 10982 | resolution is triggered after <period> modulo the <inter> parameter of |
| 10983 | the healch check. |
| 10984 | |
| 10985 | resolve_retries <nb> |
| 10986 | Defines the number <nb> of queries to send to resolve a server name before |
| 10987 | giving up. |
| 10988 | Default value: 3 |
| 10989 | |
Baptiste Assmann | 62b75b4 | 2015-09-09 01:11:36 +0200 | [diff] [blame] | 10990 | A retry occurs on name server timeout or when the full sequence of DNS query |
| 10991 | type failover is over and we need to start up from the default ANY query |
| 10992 | type. |
| 10993 | |
Baptiste Assmann | 1fa6666 | 2015-04-14 00:28:47 +0200 | [diff] [blame] | 10994 | timeout <event> <time> |
| 10995 | Defines timeouts related to name resolution |
| 10996 | <event> : the event on which the <time> timeout period applies to. |
| 10997 | events available are: |
| 10998 | - retry: time between two DNS queries, when no response have |
| 10999 | been received. |
| 11000 | Default value: 1s |
| 11001 | <time> : time related to the event. It follows the HAProxy time format. |
| 11002 | <time> is expressed in milliseconds. |
| 11003 | |
| 11004 | Example of a resolvers section (with default values): |
| 11005 | |
| 11006 | resolvers mydns |
| 11007 | nameserver dns1 10.0.0.1:53 |
| 11008 | nameserver dns2 10.0.0.2:53 |
| 11009 | resolve_retries 3 |
| 11010 | timeout retry 1s |
| 11011 | hold valid 10s |
| 11012 | |
| 11013 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 11014 | 6. HTTP header manipulation |
| 11015 | --------------------------- |
| 11016 | |
| 11017 | In HTTP mode, it is possible to rewrite, add or delete some of the request and |
| 11018 | response headers based on regular expressions. It is also possible to block a |
| 11019 | request or a response if a particular header matches a regular expression, |
| 11020 | which is enough to stop most elementary protocol attacks, and to protect |
Willy Tarreau | 70dffda | 2014-01-30 03:07:23 +0100 | [diff] [blame] | 11021 | against information leak from the internal network. |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 11022 | |
Willy Tarreau | 70dffda | 2014-01-30 03:07:23 +0100 | [diff] [blame] | 11023 | If HAProxy encounters an "Informational Response" (status code 1xx), it is able |
| 11024 | to process all rsp* rules which can allow, deny, rewrite or delete a header, |
| 11025 | but it will refuse to add a header to any such messages as this is not |
| 11026 | HTTP-compliant. The reason for still processing headers in such responses is to |
| 11027 | stop and/or fix any possible information leak which may happen, for instance |
| 11028 | because another downstream equipment would unconditionally add a header, or if |
| 11029 | a server name appears there. When such messages are seen, normal processing |
| 11030 | still occurs on the next non-informational messages. |
Willy Tarreau | 816b979 | 2009-09-15 21:25:21 +0200 | [diff] [blame] | 11031 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 11032 | This section covers common usage of the following keywords, described in detail |
| 11033 | in section 4.2 : |
| 11034 | |
| 11035 | - reqadd <string> |
| 11036 | - reqallow <search> |
| 11037 | - reqiallow <search> |
| 11038 | - reqdel <search> |
| 11039 | - reqidel <search> |
| 11040 | - reqdeny <search> |
| 11041 | - reqideny <search> |
| 11042 | - reqpass <search> |
| 11043 | - reqipass <search> |
| 11044 | - reqrep <search> <replace> |
| 11045 | - reqirep <search> <replace> |
| 11046 | - reqtarpit <search> |
| 11047 | - reqitarpit <search> |
| 11048 | - rspadd <string> |
| 11049 | - rspdel <search> |
| 11050 | - rspidel <search> |
| 11051 | - rspdeny <search> |
| 11052 | - rspideny <search> |
| 11053 | - rsprep <search> <replace> |
| 11054 | - rspirep <search> <replace> |
| 11055 | |
| 11056 | With all these keywords, the same conventions are used. The <search> parameter |
| 11057 | is a POSIX extended regular expression (regex) which supports grouping through |
| 11058 | parenthesis (without the backslash). Spaces and other delimiters must be |
| 11059 | prefixed with a backslash ('\') to avoid confusion with a field delimiter. |
| 11060 | Other characters may be prefixed with a backslash to change their meaning : |
| 11061 | |
| 11062 | \t for a tab |
| 11063 | \r for a carriage return (CR) |
| 11064 | \n for a new line (LF) |
| 11065 | \ to mark a space and differentiate it from a delimiter |
| 11066 | \# to mark a sharp and differentiate it from a comment |
| 11067 | \\ to use a backslash in a regex |
| 11068 | \\\\ to use a backslash in the text (*2 for regex, *2 for haproxy) |
| 11069 | \xXX to write the ASCII hex code XX as in the C language |
| 11070 | |
| 11071 | The <replace> parameter contains the string to be used to replace the largest |
| 11072 | portion of text matching the regex. It can make use of the special characters |
| 11073 | above, and can reference a substring which is delimited by parenthesis in the |
| 11074 | regex, by writing a backslash ('\') immediately followed by one digit from 0 to |
| 11075 | 9 indicating the group position (0 designating the entire line). This practice |
| 11076 | is very common to users of the "sed" program. |
| 11077 | |
| 11078 | The <string> parameter represents the string which will systematically be added |
| 11079 | after the last header line. It can also use special character sequences above. |
| 11080 | |
| 11081 | Notes related to these keywords : |
| 11082 | --------------------------------- |
| 11083 | - these keywords are not always convenient to allow/deny based on header |
| 11084 | contents. It is strongly recommended to use ACLs with the "block" keyword |
| 11085 | instead, resulting in far more flexible and manageable rules. |
| 11086 | |
| 11087 | - lines are always considered as a whole. It is not possible to reference |
| 11088 | a header name only or a value only. This is important because of the way |
| 11089 | headers are written (notably the number of spaces after the colon). |
| 11090 | |
| 11091 | - the first line is always considered as a header, which makes it possible to |
| 11092 | rewrite or filter HTTP requests URIs or response codes, but in turn makes |
| 11093 | it harder to distinguish between headers and request line. The regex prefix |
| 11094 | ^[^\ \t]*[\ \t] matches any HTTP method followed by a space, and the prefix |
| 11095 | ^[^ \t:]*: matches any header name followed by a colon. |
| 11096 | |
| 11097 | - for performances reasons, the number of characters added to a request or to |
| 11098 | a response is limited at build time to values between 1 and 4 kB. This |
| 11099 | should normally be far more than enough for most usages. If it is too short |
| 11100 | on occasional usages, it is possible to gain some space by removing some |
| 11101 | useless headers before adding new ones. |
| 11102 | |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 11103 | - keywords beginning with "reqi" and "rspi" are the same as their counterpart |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 11104 | without the 'i' letter except that they ignore case when matching patterns. |
| 11105 | |
| 11106 | - when a request passes through a frontend then a backend, all req* rules |
| 11107 | from the frontend will be evaluated, then all req* rules from the backend |
| 11108 | will be evaluated. The reverse path is applied to responses. |
| 11109 | |
| 11110 | - req* statements are applied after "block" statements, so that "block" is |
| 11111 | always the first one, but before "use_backend" in order to permit rewriting |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 11112 | before switching. |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 11113 | |
| 11114 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11115 | 7. Using ACLs and fetching samples |
| 11116 | ---------------------------------- |
| 11117 | |
| 11118 | Haproxy is capable of extracting data from request or response streams, from |
| 11119 | client or server information, from tables, environmental information etc... |
| 11120 | The action of extracting such data is called fetching a sample. Once retrieved, |
| 11121 | these samples may be used for various purposes such as a key to a stick-table, |
| 11122 | but most common usages consist in matching them against predefined constant |
| 11123 | data called patterns. |
| 11124 | |
| 11125 | |
| 11126 | 7.1. ACL basics |
| 11127 | --------------- |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 11128 | |
| 11129 | The use of Access Control Lists (ACL) provides a flexible solution to perform |
| 11130 | content switching and generally to take decisions based on content extracted |
| 11131 | from the request, the response or any environmental status. The principle is |
| 11132 | simple : |
| 11133 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11134 | - extract a data sample from a stream, table or the environment |
Willy Tarreau | e6b11e4 | 2013-11-26 19:02:32 +0100 | [diff] [blame] | 11135 | - optionally apply some format conversion to the extracted sample |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11136 | - apply one or multiple pattern matching methods on this sample |
| 11137 | - perform actions only when a pattern matches the sample |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 11138 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11139 | The actions generally consist in blocking a request, selecting a backend, or |
| 11140 | adding a header. |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 11141 | |
| 11142 | In order to define a test, the "acl" keyword is used. The syntax is : |
| 11143 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11144 | acl <aclname> <criterion> [flags] [operator] [<value>] ... |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 11145 | |
| 11146 | This creates a new ACL <aclname> or completes an existing one with new tests. |
| 11147 | Those tests apply to the portion of request/response specified in <criterion> |
| 11148 | and may be adjusted with optional flags [flags]. Some criteria also support |
Willy Tarreau | e6b11e4 | 2013-11-26 19:02:32 +0100 | [diff] [blame] | 11149 | an operator which may be specified before the set of values. Optionally some |
| 11150 | conversion operators may be applied to the sample, and they will be specified |
| 11151 | as a comma-delimited list of keywords just after the first keyword. The values |
| 11152 | are of the type supported by the criterion, and are separated by spaces. |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 11153 | |
| 11154 | ACL names must be formed from upper and lower case letters, digits, '-' (dash), |
| 11155 | '_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive, |
| 11156 | which means that "my_acl" and "My_Acl" are two different ACLs. |
| 11157 | |
| 11158 | There is no enforced limit to the number of ACLs. The unused ones do not affect |
| 11159 | performance, they just consume a small amount of memory. |
| 11160 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11161 | The criterion generally is the name of a sample fetch method, or one of its ACL |
| 11162 | specific declinations. The default test method is implied by the output type of |
| 11163 | this sample fetch method. The ACL declinations can describe alternate matching |
Willy Tarreau | e6b11e4 | 2013-11-26 19:02:32 +0100 | [diff] [blame] | 11164 | methods of a same sample fetch method. The sample fetch methods are the only |
| 11165 | ones supporting a conversion. |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11166 | |
| 11167 | Sample fetch methods return data which can be of the following types : |
| 11168 | - boolean |
| 11169 | - integer (signed or unsigned) |
| 11170 | - IPv4 or IPv6 address |
| 11171 | - string |
| 11172 | - data block |
| 11173 | |
Willy Tarreau | e6b11e4 | 2013-11-26 19:02:32 +0100 | [diff] [blame] | 11174 | Converters transform any of these data into any of these. For example, some |
| 11175 | converters might convert a string to a lower-case string while other ones |
| 11176 | would turn a string to an IPv4 address, or apply a netmask to an IP address. |
| 11177 | The resulting sample is of the type of the last converter applied to the list, |
| 11178 | which defaults to the type of the sample fetch method. |
| 11179 | |
Thierry FOURNIER | 2a06e39 | 2014-05-11 15:49:55 +0200 | [diff] [blame] | 11180 | Each sample or converter returns data of a specific type, specified with its |
| 11181 | keyword in this documentation. When an ACL is declared using a standard sample |
| 11182 | fetch method, certain types automatically involved a default matching method |
| 11183 | which are summarized in the table below : |
| 11184 | |
| 11185 | +---------------------+-----------------+ |
| 11186 | | Sample or converter | Default | |
| 11187 | | output type | matching method | |
| 11188 | +---------------------+-----------------+ |
| 11189 | | boolean | bool | |
| 11190 | +---------------------+-----------------+ |
| 11191 | | integer | int | |
| 11192 | +---------------------+-----------------+ |
| 11193 | | ip | ip | |
| 11194 | +---------------------+-----------------+ |
| 11195 | | string | str | |
| 11196 | +---------------------+-----------------+ |
| 11197 | | binary | none, use "-m" | |
| 11198 | +---------------------+-----------------+ |
| 11199 | |
| 11200 | Note that in order to match a binary samples, it is mandatory to specify a |
| 11201 | matching method, see below. |
| 11202 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11203 | The ACL engine can match these types against patterns of the following types : |
| 11204 | - boolean |
| 11205 | - integer or integer range |
| 11206 | - IP address / network |
| 11207 | - string (exact, substring, suffix, prefix, subdir, domain) |
| 11208 | - regular expression |
| 11209 | - hex block |
| 11210 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 11211 | The following ACL flags are currently supported : |
| 11212 | |
Willy Tarreau | 2b5285d | 2010-05-09 23:45:24 +0200 | [diff] [blame] | 11213 | -i : ignore case during matching of all subsequent patterns. |
| 11214 | -f : load patterns from a file. |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11215 | -m : use a specific pattern matching method |
Thierry FOURNIER | b7729c9 | 2014-02-11 16:24:41 +0100 | [diff] [blame] | 11216 | -n : forbid the DNS resolutions |
Thierry FOURNIER | 9860c41 | 2014-01-29 14:23:29 +0100 | [diff] [blame] | 11217 | -M : load the file pointed by -f like a map file. |
Thierry FOURNIER | 3534d88 | 2014-01-20 17:01:44 +0100 | [diff] [blame] | 11218 | -u : force the unique id of the ACL |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11219 | -- : force end of flags. Useful when a string looks like one of the flags. |
| 11220 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11221 | The "-f" flag is followed by the name of a file from which all lines will be |
| 11222 | read as individual values. It is even possible to pass multiple "-f" arguments |
| 11223 | if the patterns are to be loaded from multiple files. Empty lines as well as |
| 11224 | lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs |
| 11225 | will be stripped. If it is absolutely necessary to insert a valid pattern |
| 11226 | beginning with a sharp, just prefix it with a space so that it is not taken for |
| 11227 | a comment. Depending on the data type and match method, haproxy may load the |
| 11228 | lines into a binary tree, allowing very fast lookups. This is true for IPv4 and |
| 11229 | exact string matching. In this case, duplicates will automatically be removed. |
| 11230 | |
Thierry FOURNIER | 9860c41 | 2014-01-29 14:23:29 +0100 | [diff] [blame] | 11231 | The "-M" flag allows an ACL to use a map file. If this flag is set, the file is |
| 11232 | parsed as two column file. The first column contains the patterns used by the |
| 11233 | ACL, and the second column contain the samples. The sample can be used later by |
| 11234 | a map. This can be useful in some rare cases where an ACL would just be used to |
| 11235 | check for the existence of a pattern in a map before a mapping is applied. |
| 11236 | |
Thierry FOURNIER | 3534d88 | 2014-01-20 17:01:44 +0100 | [diff] [blame] | 11237 | The "-u" flag forces the unique id of the ACL. This unique id is used with the |
| 11238 | socket interface to identify ACL and dynamically change its values. Note that a |
| 11239 | file is always identified by its name even if an id is set. |
| 11240 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11241 | Also, note that the "-i" flag applies to subsequent entries and not to entries |
| 11242 | loaded from files preceding it. For instance : |
| 11243 | |
| 11244 | acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test |
| 11245 | |
| 11246 | In this example, each line of "exact-ua.lst" will be exactly matched against |
| 11247 | the "user-agent" header of the request. Then each line of "generic-ua" will be |
| 11248 | case-insensitively matched. Then the word "test" will be insensitively matched |
| 11249 | as well. |
| 11250 | |
| 11251 | The "-m" flag is used to select a specific pattern matching method on the input |
| 11252 | sample. All ACL-specific criteria imply a pattern matching method and generally |
| 11253 | do not need this flag. However, this flag is useful with generic sample fetch |
| 11254 | methods to describe how they're going to be matched against the patterns. This |
| 11255 | is required for sample fetches which return data type for which there is no |
| 11256 | obvious matching method (eg: string or binary). When "-m" is specified and |
| 11257 | followed by a pattern matching method name, this method is used instead of the |
| 11258 | default one for the criterion. This makes it possible to match contents in ways |
| 11259 | that were not initially planned, or with sample fetch methods which return a |
| 11260 | string. The matching method also affects the way the patterns are parsed. |
| 11261 | |
Thierry FOURNIER | b7729c9 | 2014-02-11 16:24:41 +0100 | [diff] [blame] | 11262 | The "-n" flag forbids the dns resolutions. It is used with the load of ip files. |
| 11263 | By default, if the parser cannot parse ip address it considers that the parsed |
| 11264 | string is maybe a domain name and try dns resolution. The flag "-n" disable this |
| 11265 | resolution. It is useful for detecting malformed ip lists. Note that if the DNS |
| 11266 | server is not reachable, the haproxy configuration parsing may last many minutes |
| 11267 | waiting fir the timeout. During this time no error messages are displayed. The |
| 11268 | flag "-n" disable this behavior. Note also that during the runtime, this |
| 11269 | function is disabled for the dynamic acl modifications. |
| 11270 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11271 | There are some restrictions however. Not all methods can be used with all |
| 11272 | sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must |
| 11273 | be placed first. The pattern matching method must be one of the following : |
Willy Tarreau | 5adeda1 | 2013-03-31 22:13:34 +0200 | [diff] [blame] | 11274 | |
| 11275 | - "found" : only check if the requested sample could be found in the stream, |
| 11276 | but do not compare it against any pattern. It is recommended not |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11277 | to pass any pattern to avoid confusion. This matching method is |
| 11278 | particularly useful to detect presence of certain contents such |
| 11279 | as headers, cookies, etc... even if they are empty and without |
| 11280 | comparing them to anything nor counting them. |
Willy Tarreau | 5adeda1 | 2013-03-31 22:13:34 +0200 | [diff] [blame] | 11281 | |
| 11282 | - "bool" : check the value as a boolean. It can only be applied to fetches |
| 11283 | which return a boolean or integer value, and takes no pattern. |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11284 | Value zero or false does not match, all other values do match. |
Willy Tarreau | 5adeda1 | 2013-03-31 22:13:34 +0200 | [diff] [blame] | 11285 | |
| 11286 | - "int" : match the value as an integer. It can be used with integer and |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11287 | boolean samples. Boolean false is integer 0, true is integer 1. |
Willy Tarreau | 5adeda1 | 2013-03-31 22:13:34 +0200 | [diff] [blame] | 11288 | |
| 11289 | - "ip" : match the value as an IPv4 or IPv6 address. It is compatible |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11290 | with IP address samples only, so it is implied and never needed. |
Willy Tarreau | 5adeda1 | 2013-03-31 22:13:34 +0200 | [diff] [blame] | 11291 | |
| 11292 | - "bin" : match the contents against an hexadecimal string representing a |
| 11293 | binary sequence. This may be used with binary or string samples. |
| 11294 | |
| 11295 | - "len" : match the sample's length as an integer. This may be used with |
| 11296 | binary or string samples. |
| 11297 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11298 | - "str" : exact match : match the contents against a string. This may be |
| 11299 | used with binary or string samples. |
Willy Tarreau | 5adeda1 | 2013-03-31 22:13:34 +0200 | [diff] [blame] | 11300 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11301 | - "sub" : substring match : check that the contents contain at least one of |
| 11302 | the provided string patterns. This may be used with binary or |
| 11303 | string samples. |
Willy Tarreau | 5adeda1 | 2013-03-31 22:13:34 +0200 | [diff] [blame] | 11304 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11305 | - "reg" : regex match : match the contents against a list of regular |
| 11306 | expressions. This may be used with binary or string samples. |
Willy Tarreau | 5adeda1 | 2013-03-31 22:13:34 +0200 | [diff] [blame] | 11307 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11308 | - "beg" : prefix match : check that the contents begin like the provided |
| 11309 | string patterns. This may be used with binary or string samples. |
Willy Tarreau | 5adeda1 | 2013-03-31 22:13:34 +0200 | [diff] [blame] | 11310 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11311 | - "end" : suffix match : check that the contents end like the provided |
| 11312 | string patterns. This may be used with binary or string samples. |
Willy Tarreau | 5adeda1 | 2013-03-31 22:13:34 +0200 | [diff] [blame] | 11313 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11314 | - "dir" : subdir match : check that a slash-delimited portion of the |
| 11315 | contents exactly matches one of the provided string patterns. |
Willy Tarreau | 5adeda1 | 2013-03-31 22:13:34 +0200 | [diff] [blame] | 11316 | This may be used with binary or string samples. |
| 11317 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11318 | - "dom" : domain match : check that a dot-delimited portion of the contents |
| 11319 | exactly match one of the provided string patterns. This may be |
| 11320 | used with binary or string samples. |
Willy Tarreau | 5adeda1 | 2013-03-31 22:13:34 +0200 | [diff] [blame] | 11321 | |
| 11322 | For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP |
| 11323 | request, it is possible to do : |
| 11324 | |
| 11325 | acl jsess_present cook(JSESSIONID) -m found |
| 11326 | |
| 11327 | In order to apply a regular expression on the 500 first bytes of data in the |
| 11328 | buffer, one would use the following acl : |
| 11329 | |
| 11330 | acl script_tag payload(0,500) -m reg -i <script> |
| 11331 | |
Willy Tarreau | e6b11e4 | 2013-11-26 19:02:32 +0100 | [diff] [blame] | 11332 | On systems where the regex library is much slower when using "-i", it is |
| 11333 | possible to convert the sample to lowercase before matching, like this : |
| 11334 | |
| 11335 | acl script_tag payload(0,500),lower -m reg <script> |
| 11336 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11337 | All ACL-specific criteria imply a default matching method. Most often, these |
| 11338 | criteria are composed by concatenating the name of the original sample fetch |
| 11339 | method and the matching method. For example, "hdr_beg" applies the "beg" match |
| 11340 | to samples retrieved using the "hdr" fetch method. Since all ACL-specific |
| 11341 | criteria rely on a sample fetch method, it is always possible instead to use |
| 11342 | the original sample fetch method and the explicit matching method using "-m". |
Willy Tarreau | 2b5285d | 2010-05-09 23:45:24 +0200 | [diff] [blame] | 11343 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11344 | If an alternate match is specified using "-m" on an ACL-specific criterion, |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 11345 | the matching method is simply applied to the underlying sample fetch method. |
| 11346 | For example, all ACLs below are exact equivalent : |
Willy Tarreau | 2b5285d | 2010-05-09 23:45:24 +0200 | [diff] [blame] | 11347 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11348 | acl short_form hdr_beg(host) www. |
| 11349 | acl alternate1 hdr_beg(host) -m beg www. |
| 11350 | acl alternate2 hdr_dom(host) -m beg www. |
| 11351 | acl alternate3 hdr(host) -m beg www. |
Willy Tarreau | 2b5285d | 2010-05-09 23:45:24 +0200 | [diff] [blame] | 11352 | |
Willy Tarreau | 2b5285d | 2010-05-09 23:45:24 +0200 | [diff] [blame] | 11353 | |
Thierry FOURNIER | 2a06e39 | 2014-05-11 15:49:55 +0200 | [diff] [blame] | 11354 | The table below summarizes the compatibility matrix between sample or converter |
| 11355 | types and the pattern types to fetch against. It indicates for each compatible |
| 11356 | combination the name of the matching method to be used, surrounded with angle |
| 11357 | brackets ">" and "<" when the method is the default one and will work by |
| 11358 | default without "-m". |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 11359 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11360 | +-------------------------------------------------+ |
| 11361 | | Input sample type | |
| 11362 | +----------------------+---------+---------+---------+---------+---------+ |
Thierry FOURNIER | 2a06e39 | 2014-05-11 15:49:55 +0200 | [diff] [blame] | 11363 | | pattern type | boolean | integer | ip | string | binary | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11364 | +----------------------+---------+---------+---------+---------+---------+ |
| 11365 | | none (presence only) | found | found | found | found | found | |
| 11366 | +----------------------+---------+---------+---------+---------+---------+ |
Thierry FOURNIER | 2a06e39 | 2014-05-11 15:49:55 +0200 | [diff] [blame] | 11367 | | none (boolean value) |> bool <| bool | | bool | | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11368 | +----------------------+---------+---------+---------+---------+---------+ |
Thierry FOURNIER | 2a06e39 | 2014-05-11 15:49:55 +0200 | [diff] [blame] | 11369 | | integer (value) | int |> int <| int | int | | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11370 | +----------------------+---------+---------+---------+---------+---------+ |
Thierry FOURNIER | e3ded59 | 2013-12-06 15:36:54 +0100 | [diff] [blame] | 11371 | | integer (length) | len | len | len | len | len | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11372 | +----------------------+---------+---------+---------+---------+---------+ |
Thierry FOURNIER | 2a06e39 | 2014-05-11 15:49:55 +0200 | [diff] [blame] | 11373 | | IP address | | |> ip <| ip | ip | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11374 | +----------------------+---------+---------+---------+---------+---------+ |
Thierry FOURNIER | 2a06e39 | 2014-05-11 15:49:55 +0200 | [diff] [blame] | 11375 | | exact string | str | str | str |> str <| str | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11376 | +----------------------+---------+---------+---------+---------+---------+ |
Thierry FOURNIER | e3ded59 | 2013-12-06 15:36:54 +0100 | [diff] [blame] | 11377 | | prefix | beg | beg | beg | beg | beg | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11378 | +----------------------+---------+---------+---------+---------+---------+ |
Thierry FOURNIER | e3ded59 | 2013-12-06 15:36:54 +0100 | [diff] [blame] | 11379 | | suffix | end | end | end | end | end | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11380 | +----------------------+---------+---------+---------+---------+---------+ |
Thierry FOURNIER | e3ded59 | 2013-12-06 15:36:54 +0100 | [diff] [blame] | 11381 | | substring | sub | sub | sub | sub | sub | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11382 | +----------------------+---------+---------+---------+---------+---------+ |
Thierry FOURNIER | e3ded59 | 2013-12-06 15:36:54 +0100 | [diff] [blame] | 11383 | | subdir | dir | dir | dir | dir | dir | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11384 | +----------------------+---------+---------+---------+---------+---------+ |
Thierry FOURNIER | e3ded59 | 2013-12-06 15:36:54 +0100 | [diff] [blame] | 11385 | | domain | dom | dom | dom | dom | dom | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11386 | +----------------------+---------+---------+---------+---------+---------+ |
Thierry FOURNIER | e3ded59 | 2013-12-06 15:36:54 +0100 | [diff] [blame] | 11387 | | regex | reg | reg | reg | reg | reg | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11388 | +----------------------+---------+---------+---------+---------+---------+ |
| 11389 | | hex block | | | | bin | bin | |
| 11390 | +----------------------+---------+---------+---------+---------+---------+ |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11391 | |
| 11392 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11393 | 7.1.1. Matching booleans |
| 11394 | ------------------------ |
| 11395 | |
| 11396 | In order to match a boolean, no value is needed and all values are ignored. |
| 11397 | Boolean matching is used by default for all fetch methods of type "boolean". |
| 11398 | When boolean matching is used, the fetched value is returned as-is, which means |
| 11399 | that a boolean "true" will always match and a boolean "false" will never match. |
| 11400 | |
| 11401 | Boolean matching may also be enforced using "-m bool" on fetch methods which |
| 11402 | return an integer value. Then, integer value 0 is converted to the boolean |
| 11403 | "false" and all other values are converted to "true". |
| 11404 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11405 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11406 | 7.1.2. Matching integers |
| 11407 | ------------------------ |
| 11408 | |
| 11409 | Integer matching applies by default to integer fetch methods. It can also be |
| 11410 | enforced on boolean fetches using "-m int". In this case, "false" is converted |
| 11411 | to the integer 0, and "true" is converted to the integer 1. |
| 11412 | |
| 11413 | Integer matching also supports integer ranges and operators. Note that integer |
| 11414 | matching only applies to positive values. A range is a value expressed with a |
| 11415 | lower and an upper bound separated with a colon, both of which may be omitted. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11416 | |
| 11417 | For instance, "1024:65535" is a valid range to represent a range of |
| 11418 | unprivileged ports, and "1024:" would also work. "0:1023" is a valid |
| 11419 | representation of privileged ports, and ":1023" would also work. |
| 11420 | |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 11421 | As a special case, some ACL functions support decimal numbers which are in fact |
| 11422 | two integers separated by a dot. This is used with some version checks for |
| 11423 | instance. All integer properties apply to those decimal numbers, including |
| 11424 | ranges and operators. |
| 11425 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11426 | For an easier usage, comparison operators are also supported. Note that using |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 11427 | operators with ranges does not make much sense and is strongly discouraged. |
| 11428 | Similarly, it does not make much sense to perform order comparisons with a set |
| 11429 | of values. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11430 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 11431 | Available operators for integer matching are : |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11432 | |
| 11433 | eq : true if the tested value equals at least one value |
| 11434 | ge : true if the tested value is greater than or equal to at least one value |
| 11435 | gt : true if the tested value is greater than at least one value |
| 11436 | le : true if the tested value is less than or equal to at least one value |
| 11437 | lt : true if the tested value is less than at least one value |
| 11438 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 11439 | For instance, the following ACL matches any negative Content-Length header : |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11440 | |
| 11441 | acl negative-length hdr_val(content-length) lt 0 |
| 11442 | |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 11443 | This one matches SSL versions between 3.0 and 3.1 (inclusive) : |
| 11444 | |
| 11445 | acl sslv3 req_ssl_ver 3:3.1 |
| 11446 | |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11447 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11448 | 7.1.3. Matching strings |
| 11449 | ----------------------- |
| 11450 | |
| 11451 | String matching applies to string or binary fetch methods, and exists in 6 |
| 11452 | different forms : |
| 11453 | |
| 11454 | - exact match (-m str) : the extracted string must exactly match the |
| 11455 | patterns ; |
| 11456 | |
| 11457 | - substring match (-m sub) : the patterns are looked up inside the |
| 11458 | extracted string, and the ACL matches if any of them is found inside ; |
| 11459 | |
| 11460 | - prefix match (-m beg) : the patterns are compared with the beginning of |
| 11461 | the extracted string, and the ACL matches if any of them matches. |
| 11462 | |
| 11463 | - suffix match (-m end) : the patterns are compared with the end of the |
| 11464 | extracted string, and the ACL matches if any of them matches. |
| 11465 | |
Baptiste Assmann | 33db600 | 2016-03-06 23:32:10 +0100 | [diff] [blame] | 11466 | - subdir match (-m dir) : the patterns are looked up inside the extracted |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11467 | string, delimited with slashes ("/"), and the ACL matches if any of them |
| 11468 | matches. |
| 11469 | |
| 11470 | - domain match (-m dom) : the patterns are looked up inside the extracted |
| 11471 | string, delimited with dots ("."), and the ACL matches if any of them |
| 11472 | matches. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11473 | |
| 11474 | String matching applies to verbatim strings as they are passed, with the |
| 11475 | exception of the backslash ("\") which makes it possible to escape some |
| 11476 | characters such as the space. If the "-i" flag is passed before the first |
| 11477 | string, then the matching will be performed ignoring the case. In order |
| 11478 | to match the string "-i", either set it second, or pass the "--" flag |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 11479 | before the first string. Same applies of course to match the string "--". |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11480 | |
| 11481 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11482 | 7.1.4. Matching regular expressions (regexes) |
| 11483 | --------------------------------------------- |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11484 | |
| 11485 | Just like with string matching, regex matching applies to verbatim strings as |
| 11486 | they are passed, with the exception of the backslash ("\") which makes it |
| 11487 | possible to escape some characters such as the space. If the "-i" flag is |
| 11488 | passed before the first regex, then the matching will be performed ignoring |
| 11489 | the case. In order to match the string "-i", either set it second, or pass |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 11490 | the "--" flag before the first string. Same principle applies of course to |
| 11491 | match the string "--". |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11492 | |
| 11493 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11494 | 7.1.5. Matching arbitrary data blocks |
| 11495 | ------------------------------------- |
| 11496 | |
| 11497 | It is possible to match some extracted samples against a binary block which may |
| 11498 | not safely be represented as a string. For this, the patterns must be passed as |
| 11499 | a series of hexadecimal digits in an even number, when the match method is set |
| 11500 | to binary. Each sequence of two digits will represent a byte. The hexadecimal |
| 11501 | digits may be used upper or lower case. |
| 11502 | |
| 11503 | Example : |
| 11504 | # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a) |
| 11505 | acl hello payload(0,6) -m bin 48656c6c6f0a |
| 11506 | |
| 11507 | |
| 11508 | 7.1.6. Matching IPv4 and IPv6 addresses |
| 11509 | --------------------------------------- |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11510 | |
| 11511 | IPv4 addresses values can be specified either as plain addresses or with a |
| 11512 | netmask appended, in which case the IPv4 address matches whenever it is |
| 11513 | within the network. Plain addresses may also be replaced with a resolvable |
Willy Tarreau | d2a4aa2 | 2008-01-31 15:28:22 +0100 | [diff] [blame] | 11514 | host name, but this practice is generally discouraged as it makes it more |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 11515 | difficult to read and debug configurations. If hostnames are used, you should |
| 11516 | at least ensure that they are present in /etc/hosts so that the configuration |
| 11517 | does not depend on any random DNS match at the moment the configuration is |
| 11518 | parsed. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11519 | |
Daniel Schneller | eba5634 | 2016-04-13 00:26:52 +0200 | [diff] [blame] | 11520 | The dotted IPv4 address notation is supported in both regular as well as the |
| 11521 | abbreviated form with all-0-octets omitted: |
| 11522 | |
| 11523 | +------------------+------------------+------------------+ |
| 11524 | | Example 1 | Example 2 | Example 3 | |
| 11525 | +------------------+------------------+------------------+ |
| 11526 | | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 | |
| 11527 | | 192.168.1 | 10.12 | 127.1 | |
| 11528 | | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 | |
| 11529 | | 192.168.1/22 | 10.12/8 | 127.1/8 | |
| 11530 | +------------------+------------------+------------------+ |
| 11531 | |
| 11532 | Notice that this is different from RFC 4632 CIDR address notation in which |
| 11533 | 192.168.42/24 would be equivalent to 192.168.42.0/24. |
| 11534 | |
Willy Tarreau | ceb4ac9 | 2012-04-28 00:41:46 +0200 | [diff] [blame] | 11535 | IPv6 may be entered in their usual form, with or without a netmask appended. |
| 11536 | Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of |
| 11537 | trouble with randomly resolved IP addresses, host names are never allowed in |
| 11538 | IPv6 patterns. |
| 11539 | |
| 11540 | HAProxy is also able to match IPv4 addresses with IPv6 addresses in the |
| 11541 | following situations : |
| 11542 | - tested address is IPv4, pattern address is IPv4, the match applies |
| 11543 | in IPv4 using the supplied mask if any. |
| 11544 | - tested address is IPv6, pattern address is IPv6, the match applies |
| 11545 | in IPv6 using the supplied mask if any. |
| 11546 | - tested address is IPv6, pattern address is IPv4, the match applies in IPv4 |
| 11547 | using the pattern's mask if the IPv6 address matches with 2002:IPV4::, |
| 11548 | ::IPV4 or ::ffff:IPV4, otherwise it fails. |
| 11549 | - tested address is IPv4, pattern address is IPv6, the IPv4 address is first |
| 11550 | converted to IPv6 by prefixing ::ffff: in front of it, then the match is |
| 11551 | applied in IPv6 using the supplied IPv6 mask. |
| 11552 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11553 | |
| 11554 | 7.2. Using ACLs to form conditions |
| 11555 | ---------------------------------- |
| 11556 | |
| 11557 | Some actions are only performed upon a valid condition. A condition is a |
| 11558 | combination of ACLs with operators. 3 operators are supported : |
| 11559 | |
| 11560 | - AND (implicit) |
| 11561 | - OR (explicit with the "or" keyword or the "||" operator) |
| 11562 | - Negation with the exclamation mark ("!") |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11563 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11564 | A condition is formed as a disjunctive form: |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 11565 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11566 | [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ... |
Willy Tarreau | bef91e7 | 2013-03-31 23:14:46 +0200 | [diff] [blame] | 11567 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11568 | Such conditions are generally used after an "if" or "unless" statement, |
| 11569 | indicating when the condition will trigger the action. |
Willy Tarreau | bef91e7 | 2013-03-31 23:14:46 +0200 | [diff] [blame] | 11570 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11571 | For instance, to block HTTP requests to the "*" URL with methods other than |
| 11572 | "OPTIONS", as well as POST requests without content-length, and GET or HEAD |
| 11573 | requests with a content-length greater than 0, and finally every request which |
| 11574 | is not either GET/HEAD/POST/OPTIONS ! |
| 11575 | |
| 11576 | acl missing_cl hdr_cnt(Content-length) eq 0 |
| 11577 | block if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl |
| 11578 | block if METH_GET HTTP_CONTENT |
| 11579 | block unless METH_GET or METH_POST or METH_OPTIONS |
| 11580 | |
| 11581 | To select a different backend for requests to static contents on the "www" site |
| 11582 | and to every request on the "img", "video", "download" and "ftp" hosts : |
| 11583 | |
| 11584 | acl url_static path_beg /static /images /img /css |
| 11585 | acl url_static path_end .gif .png .jpg .css .js |
| 11586 | acl host_www hdr_beg(host) -i www |
| 11587 | acl host_static hdr_beg(host) -i img. video. download. ftp. |
| 11588 | |
| 11589 | # now use backend "static" for all static-only hosts, and for static urls |
| 11590 | # of host "www". Use backend "www" for the rest. |
| 11591 | use_backend static if host_static or host_www url_static |
| 11592 | use_backend www if host_www |
| 11593 | |
| 11594 | It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL |
| 11595 | expressions that are built on the fly without needing to be declared. They must |
| 11596 | be enclosed between braces, with a space before and after each brace (because |
| 11597 | the braces must be seen as independent words). Example : |
| 11598 | |
| 11599 | The following rule : |
| 11600 | |
| 11601 | acl missing_cl hdr_cnt(Content-length) eq 0 |
| 11602 | block if METH_POST missing_cl |
| 11603 | |
| 11604 | Can also be written that way : |
| 11605 | |
| 11606 | block if METH_POST { hdr_cnt(Content-length) eq 0 } |
| 11607 | |
| 11608 | It is generally not recommended to use this construct because it's a lot easier |
| 11609 | to leave errors in the configuration when written that way. However, for very |
| 11610 | simple rules matching only one source IP address for instance, it can make more |
| 11611 | sense to use them than to declare ACLs with random names. Another example of |
| 11612 | good use is the following : |
| 11613 | |
| 11614 | With named ACLs : |
| 11615 | |
| 11616 | acl site_dead nbsrv(dynamic) lt 2 |
| 11617 | acl site_dead nbsrv(static) lt 2 |
| 11618 | monitor fail if site_dead |
| 11619 | |
| 11620 | With anonymous ACLs : |
| 11621 | |
| 11622 | monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 } |
| 11623 | |
| 11624 | See section 4.2 for detailed help on the "block" and "use_backend" keywords. |
| 11625 | |
| 11626 | |
| 11627 | 7.3. Fetching samples |
| 11628 | --------------------- |
| 11629 | |
| 11630 | Historically, sample fetch methods were only used to retrieve data to match |
| 11631 | against patterns using ACLs. With the arrival of stick-tables, a new class of |
| 11632 | sample fetch methods was created, most often sharing the same syntax as their |
| 11633 | ACL counterpart. These sample fetch methods are also known as "fetches". As |
| 11634 | of now, ACLs and fetches have converged. All ACL fetch methods have been made |
| 11635 | available as fetch methods, and ACLs may use any sample fetch method as well. |
| 11636 | |
| 11637 | This section details all available sample fetch methods and their output type. |
| 11638 | Some sample fetch methods have deprecated aliases that are used to maintain |
| 11639 | compatibility with existing configurations. They are then explicitly marked as |
| 11640 | deprecated and should not be used in new setups. |
| 11641 | |
| 11642 | The ACL derivatives are also indicated when available, with their respective |
| 11643 | matching methods. These ones all have a well defined default pattern matching |
| 11644 | method, so it is never necessary (though allowed) to pass the "-m" option to |
| 11645 | indicate how the sample will be matched using ACLs. |
| 11646 | |
| 11647 | As indicated in the sample type versus matching compatibility matrix above, |
| 11648 | when using a generic sample fetch method in an ACL, the "-m" option is |
| 11649 | mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When |
| 11650 | the same keyword exists as an ACL keyword and as a standard fetch method, the |
| 11651 | ACL engine will automatically pick the ACL-only one by default. |
| 11652 | |
| 11653 | Some of these keywords support one or multiple mandatory arguments, and one or |
| 11654 | multiple optional arguments. These arguments are strongly typed and are checked |
| 11655 | when the configuration is parsed so that there is no risk of running with an |
| 11656 | incorrect argument (eg: an unresolved backend name). Fetch function arguments |
| 11657 | are passed between parenthesis and are delimited by commas. When an argument |
| 11658 | is optional, it will be indicated below between square brackets ('[ ]'). When |
| 11659 | all arguments are optional, the parenthesis may be omitted. |
| 11660 | |
| 11661 | Thus, the syntax of a standard sample fetch method is one of the following : |
| 11662 | - name |
| 11663 | - name(arg1) |
| 11664 | - name(arg1,arg2) |
| 11665 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 11666 | |
| 11667 | 7.3.1. Converters |
| 11668 | ----------------- |
| 11669 | |
Willy Tarreau | e6b11e4 | 2013-11-26 19:02:32 +0100 | [diff] [blame] | 11670 | Sample fetch methods may be combined with transformations to be applied on top |
| 11671 | of the fetched sample (also called "converters"). These combinations form what |
| 11672 | is called "sample expressions" and the result is a "sample". Initially this |
| 11673 | was only supported by "stick on" and "stick store-request" directives but this |
| 11674 | has now be extended to all places where samples may be used (acls, log-format, |
| 11675 | unique-id-format, add-header, ...). |
| 11676 | |
| 11677 | These transformations are enumerated as a series of specific keywords after the |
| 11678 | sample fetch method. These keywords may equally be appended immediately after |
| 11679 | the fetch keyword's argument, delimited by a comma. These keywords can also |
| 11680 | support some arguments (eg: a netmask) which must be passed in parenthesis. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 11681 | |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 11682 | A certain category of converters are bitwise and arithmetic operators which |
| 11683 | support performing basic operations on integers. Some bitwise operations are |
| 11684 | supported (and, or, xor, cpl) and some arithmetic operations are supported |
| 11685 | (add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not, |
| 11686 | bool) which make it possible to report a match without having to write an ACL. |
| 11687 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11688 | The currently available list of transformation keywords include : |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 11689 | |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 11690 | add(<value>) |
Thierry FOURNIER | 07ee64e | 2015-07-06 23:43:03 +0200 | [diff] [blame] | 11691 | Adds <value> to the input value of type signed integer, and returns the |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 11692 | result as a signed integer. <value> can be a numeric value or a variable |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 11693 | name. The name of the variable starts with an indication about its scope. The |
| 11694 | scopes allowed are: |
| 11695 | "sess" : the variable is shared with the whole session |
| 11696 | "txn" : the variable is shared with the transaction (request and response) |
| 11697 | "req" : the variable is shared only during request processing |
| 11698 | "res" : the variable is shared only during response processing |
| 11699 | This prefix is followed by a name. The separator is a '.'. The name may only |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 11700 | contain characters 'a-z', 'A-Z', '0-9' and '_'. |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 11701 | |
| 11702 | and(<value>) |
Thierry FOURNIER | 07ee64e | 2015-07-06 23:43:03 +0200 | [diff] [blame] | 11703 | Performs a bitwise "AND" between <value> and the input value of type signed |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 11704 | integer, and returns the result as an signed integer. <value> can be a |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 11705 | numeric value or a variable name. The name of the variable starts with an |
| 11706 | indication about its scope. The scopes allowed are: |
| 11707 | "sess" : the variable is shared with the whole session |
| 11708 | "txn" : the variable is shared with the transaction (request and response) |
| 11709 | "req" : the variable is shared only during request processing |
| 11710 | "res" : the variable is shared only during response processing |
| 11711 | This prefix is followed by a name. The separator is a '.'. The name may only |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 11712 | contain characters 'a-z', 'A-Z', '0-9' and '_'. |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 11713 | |
Emeric Brun | 53d1a98 | 2014-04-30 18:21:37 +0200 | [diff] [blame] | 11714 | base64 |
| 11715 | Converts a binary input sample to a base64 string. It is used to log or |
| 11716 | transfer binary content in a way that can be reliably transferred (eg: |
| 11717 | an SSL ID can be copied in a header). |
| 11718 | |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 11719 | bool |
Thierry FOURNIER | 07ee64e | 2015-07-06 23:43:03 +0200 | [diff] [blame] | 11720 | Returns a boolean TRUE if the input value of type signed integer is |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 11721 | non-null, otherwise returns FALSE. Used in conjunction with and(), it can be |
| 11722 | used to report true/false for bit testing on input values (eg: verify the |
| 11723 | presence of a flag). |
| 11724 | |
Emeric Brun | 54c4ac8 | 2014-11-03 15:32:43 +0100 | [diff] [blame] | 11725 | bytes(<offset>[,<length>]) |
| 11726 | Extracts some bytes from an input binary sample. The result is a binary |
| 11727 | sample starting at an offset (in bytes) of the original sample and |
| 11728 | optionnaly truncated at the given length. |
| 11729 | |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 11730 | cpl |
Thierry FOURNIER | 07ee64e | 2015-07-06 23:43:03 +0200 | [diff] [blame] | 11731 | Takes the input value of type signed integer, applies a ones-complement |
| 11732 | (flips all bits) and returns the result as an signed integer. |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 11733 | |
Willy Tarreau | 8059977 | 2015-01-20 19:35:24 +0100 | [diff] [blame] | 11734 | crc32([<avalanche>]) |
| 11735 | Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32 |
| 11736 | hash function. Optionally, it is possible to apply a full avalanche hash |
| 11737 | function to the output if the optional <avalanche> argument equals 1. This |
| 11738 | converter uses the same functions as used by the various hash-based load |
| 11739 | balancing algorithms, so it will provide exactly the same results. It is |
| 11740 | provided for compatibility with other software which want a CRC32 to be |
| 11741 | computed on some input keys, so it follows the most common implementation as |
| 11742 | found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms |
| 11743 | but may provide a better or at least less predictable distribution. It must |
| 11744 | not be used for security purposes as a 32-bit hash is trivial to break. See |
| 11745 | also "djb2", "sdbm", "wt6" and the "hash-type" directive. |
| 11746 | |
David Carlier | 29b3ca3 | 2015-09-25 14:09:21 +0100 | [diff] [blame] | 11747 | da-csv-conv(<prop>[,<prop>*]) |
David Carlier | 4542b10 | 2015-06-01 13:54:29 +0200 | [diff] [blame] | 11748 | Asks the DeviceAtlas converter to identify the User Agent string passed on |
| 11749 | input, and to emit a string made of the concatenation of the properties |
| 11750 | enumerated in argument, delimited by the separator defined by the global |
| 11751 | keyword "deviceatlas-property-separator", or by default the pipe character |
David Carlier | 840b024 | 2016-03-16 10:09:55 +0000 | [diff] [blame] | 11752 | ('|'). There's a limit of 12 different properties imposed by the haproxy |
David Carlier | 4542b10 | 2015-06-01 13:54:29 +0200 | [diff] [blame] | 11753 | configuration language. |
| 11754 | |
| 11755 | Example: |
| 11756 | frontend www |
Cyril Bonté | 307ee1e | 2015-09-28 23:16:06 +0200 | [diff] [blame] | 11757 | bind *:8881 |
| 11758 | default_backend servers |
David Carlier | 840b024 | 2016-03-16 10:09:55 +0000 | [diff] [blame] | 11759 | http-request set-header X-DeviceAtlas-Data %[req.fhdr(User-Agent),da-csv(primaryHardwareType,osName,osVersion,browserName,browserVersion,browserRenderingEngine)] |
David Carlier | 4542b10 | 2015-06-01 13:54:29 +0200 | [diff] [blame] | 11760 | |
Thierry FOURNIER | 9687c77 | 2015-05-07 15:46:29 +0200 | [diff] [blame] | 11761 | debug |
| 11762 | This converter is used as debug tool. It dumps on screen the content and the |
| 11763 | type of the input sample. The sample is returned as is on its output. This |
| 11764 | converter only exists when haproxy was built with debugging enabled. |
| 11765 | |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 11766 | div(<value>) |
Thierry FOURNIER | 07ee64e | 2015-07-06 23:43:03 +0200 | [diff] [blame] | 11767 | Divides the input value of type signed integer by <value>, and returns the |
| 11768 | result as an signed integer. If <value> is null, the largest unsigned |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 11769 | integer is returned (typically 2^63-1). <value> can be a numeric value or a |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 11770 | variable name. The name of the variable starts with an indication about its |
| 11771 | scope. The scopes allowed are: |
| 11772 | "sess" : the variable is shared with the whole session |
| 11773 | "txn" : the variable is shared with the transaction (request and response) |
| 11774 | "req" : the variable is shared only during request processing |
| 11775 | "res" : the variable is shared only during response processing |
| 11776 | This prefix is followed by a name. The separator is a '.'. The name may only |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 11777 | contain characters 'a-z', 'A-Z', '0-9' and '_'. |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 11778 | |
Willy Tarreau | 23ec4ca | 2014-07-15 20:15:37 +0200 | [diff] [blame] | 11779 | djb2([<avalanche>]) |
| 11780 | Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2 |
| 11781 | hash function. Optionally, it is possible to apply a full avalanche hash |
| 11782 | function to the output if the optional <avalanche> argument equals 1. This |
| 11783 | converter uses the same functions as used by the various hash-based load |
| 11784 | balancing algorithms, so it will provide exactly the same results. It is |
| 11785 | mostly intended for debugging, but can be used as a stick-table entry to |
| 11786 | collect rough statistics. It must not be used for security purposes as a |
Willy Tarreau | 8059977 | 2015-01-20 19:35:24 +0100 | [diff] [blame] | 11787 | 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6" and the |
| 11788 | "hash-type" directive. |
Willy Tarreau | 23ec4ca | 2014-07-15 20:15:37 +0200 | [diff] [blame] | 11789 | |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 11790 | even |
Thierry FOURNIER | 07ee64e | 2015-07-06 23:43:03 +0200 | [diff] [blame] | 11791 | Returns a boolean TRUE if the input value of type signed integer is even |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 11792 | otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool". |
| 11793 | |
Emeric Brun | f399b0d | 2014-11-03 17:07:03 +0100 | [diff] [blame] | 11794 | field(<index>,<delimiters>) |
| 11795 | Extracts the substring at the given index considering given delimiters from |
| 11796 | an input string. Indexes start at 1 and delimiters are a string formatted |
| 11797 | list of chars. |
| 11798 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 11799 | hex |
| 11800 | Converts a binary input sample to an hex string containing two hex digits per |
| 11801 | input byte. It is used to log or transfer hex dumps of some binary input data |
| 11802 | in a way that can be reliably transferred (eg: an SSL ID can be copied in a |
| 11803 | header). |
Thierry FOURNIER | 2f49d6d | 2014-03-12 15:01:52 +0100 | [diff] [blame] | 11804 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 11805 | http_date([<offset>]) |
| 11806 | Converts an integer supposed to contain a date since epoch to a string |
| 11807 | representing this date in a format suitable for use in HTTP header fields. If |
| 11808 | an offset value is specified, then it is a number of seconds that is added to |
| 11809 | the date before the conversion is operated. This is particularly useful to |
| 11810 | emit Date header fields, Expires values in responses when combined with a |
| 11811 | positive offset, or Last-Modified values when the offset is negative. |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 11812 | |
Willy Tarreau | d9f316a | 2014-07-10 14:03:38 +0200 | [diff] [blame] | 11813 | in_table(<table>) |
| 11814 | Uses the string representation of the input sample to perform a look up in |
| 11815 | the specified table. If the key is not found in the table, a boolean false |
| 11816 | is returned. Otherwise a boolean true is returned. This can be used to verify |
| 11817 | the presence of a certain key in a table tracking some elements (eg: whether |
| 11818 | or not a source IP address or an Authorization header was already seen). |
| 11819 | |
Willy Tarreau | ffcb2e4 | 2014-07-10 16:29:08 +0200 | [diff] [blame] | 11820 | ipmask(<mask>) |
| 11821 | Apply a mask to an IPv4 address, and use the result for lookups and storage. |
| 11822 | This can be used to make all hosts within a certain mask to share the same |
| 11823 | table entries and as such use the same server. The mask can be passed in |
| 11824 | dotted form (eg: 255.255.255.0) or in CIDR form (eg: 24). |
| 11825 | |
Thierry FOURNIER | 317e1c4 | 2014-08-12 10:20:47 +0200 | [diff] [blame] | 11826 | json([<input-code>]) |
| 11827 | Escapes the input string and produces an ASCII ouput string ready to use as a |
| 11828 | JSON string. The converter tries to decode the input string according to the |
| 11829 | <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8"" or |
| 11830 | "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types |
| 11831 | of errors: |
| 11832 | - bad UTF-8 sequence (lone continuation byte, bad number of continuation |
| 11833 | bytes, ...) |
| 11834 | - invalid range (the decoded value is within a UTF-8 prohibited range), |
| 11835 | - code overlong (the value is encoded with more bytes than necessary). |
| 11836 | |
| 11837 | The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8 |
| 11838 | character is greater than 0xffff because the JSON string escape specification |
| 11839 | only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists |
| 11840 | in 4 variants designated by a combination of two suffix letters : "p" for |
| 11841 | "permissive" and "s" for "silently ignore". The behaviors of the decoders |
| 11842 | are : |
| 11843 | - "ascii" : never fails ; |
| 11844 | - "utf8" : fails on any detected errors ; |
| 11845 | - "utf8s" : never fails, but removes characters corresponding to errors ; |
| 11846 | - "utf8p" : accepts and fixes the overlong errors, but fails on any other |
| 11847 | error ; |
| 11848 | - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes |
| 11849 | characters corresponding to the other errors. |
| 11850 | |
| 11851 | This converter is particularly useful for building properly escaped JSON for |
| 11852 | logging to servers which consume JSON-formated traffic logs. |
| 11853 | |
| 11854 | Example: |
| 11855 | capture request header user-agent len 150 |
| 11856 | capture request header Host len 15 |
Mac Browning | e83345d | 2016-03-14 14:46:01 -0400 | [diff] [blame] | 11857 | log-format {"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json("utf8s")]"} |
Thierry FOURNIER | 317e1c4 | 2014-08-12 10:20:47 +0200 | [diff] [blame] | 11858 | |
| 11859 | Input request from client 127.0.0.1: |
| 11860 | GET / HTTP/1.0 |
| 11861 | User-Agent: Very "Ugly" UA 1/2 |
| 11862 | |
| 11863 | Output log: |
| 11864 | {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"} |
| 11865 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 11866 | language(<value>[,<default>]) |
| 11867 | Returns the value with the highest q-factor from a list as extracted from the |
| 11868 | "accept-language" header using "req.fhdr". Values with no q-factor have a |
| 11869 | q-factor of 1. Values with a q-factor of 0 are dropped. Only values which |
| 11870 | belong to the list of semi-colon delimited <values> will be considered. The |
| 11871 | argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the |
| 11872 | given list and a default value is provided, it is returned. Note that language |
| 11873 | names may have a variant after a dash ('-'). If this variant is present in the |
| 11874 | list, it will be matched, but if it is not, only the base language is checked. |
| 11875 | The match is case-sensitive, and the output string is always one of those |
| 11876 | provided in arguments. The ordering of arguments is meaningless, only the |
| 11877 | ordering of the values in the request counts, as the first value among |
| 11878 | multiple sharing the same q-factor is used. |
Thierry FOURNIER | ad90351 | 2014-04-11 17:51:01 +0200 | [diff] [blame] | 11879 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 11880 | Example : |
Thierry FOURNIER | ad90351 | 2014-04-11 17:51:01 +0200 | [diff] [blame] | 11881 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 11882 | # this configuration switches to the backend matching a |
| 11883 | # given language based on the request : |
Thierry FOURNIER | ad90351 | 2014-04-11 17:51:01 +0200 | [diff] [blame] | 11884 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 11885 | acl es req.fhdr(accept-language),language(es;fr;en) -m str es |
| 11886 | acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr |
| 11887 | acl en req.fhdr(accept-language),language(es;fr;en) -m str en |
| 11888 | use_backend spanish if es |
| 11889 | use_backend french if fr |
| 11890 | use_backend english if en |
| 11891 | default_backend choose_your_language |
Thierry FOURNIER | ad90351 | 2014-04-11 17:51:01 +0200 | [diff] [blame] | 11892 | |
Willy Tarreau | ffcb2e4 | 2014-07-10 16:29:08 +0200 | [diff] [blame] | 11893 | lower |
| 11894 | Convert a string sample to lower case. This can only be placed after a string |
| 11895 | sample fetch function or after a transformation keyword returning a string |
| 11896 | type. The result is of type string. |
| 11897 | |
Willy Tarreau | 0dbfdba | 2014-07-10 16:37:47 +0200 | [diff] [blame] | 11898 | ltime(<format>[,<offset>]) |
| 11899 | Converts an integer supposed to contain a date since epoch to a string |
| 11900 | representing this date in local time using a format defined by the <format> |
| 11901 | string using strftime(3). The purpose is to allow any date format to be used |
| 11902 | in logs. An optional <offset> in seconds may be applied to the input date |
| 11903 | (positive or negative). See the strftime() man page for the format supported |
| 11904 | by your operating system. See also the utime converter. |
| 11905 | |
| 11906 | Example : |
| 11907 | |
| 11908 | # Emit two colons, one with the local time and another with ip:port |
| 11909 | # Eg: 20140710162350 127.0.0.1:57325 |
| 11910 | log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp |
| 11911 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 11912 | map(<map_file>[,<default_value>]) |
| 11913 | map_<match_type>(<map_file>[,<default_value>]) |
| 11914 | map_<match_type>_<output_type>(<map_file>[,<default_value>]) |
| 11915 | Search the input value from <map_file> using the <match_type> matching method, |
| 11916 | and return the associated value converted to the type <output_type>. If the |
| 11917 | input value cannot be found in the <map_file>, the converter returns the |
| 11918 | <default_value>. If the <default_value> is not set, the converter fails and |
| 11919 | acts as if no input value could be fetched. If the <match_type> is not set, it |
| 11920 | defaults to "str". Likewise, if the <output_type> is not set, it defaults to |
| 11921 | "str". For convenience, the "map" keyword is an alias for "map_str" and maps a |
| 11922 | string to another string. |
Thierry FOURNIER | d5f624d | 2013-11-26 11:52:33 +0100 | [diff] [blame] | 11923 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 11924 | It is important to avoid overlapping between the keys : IP addresses and |
| 11925 | strings are stored in trees, so the first of the finest match will be used. |
| 11926 | Other keys are stored in lists, so the first matching occurrence will be used. |
Thierry FOURNIER | d5f624d | 2013-11-26 11:52:33 +0100 | [diff] [blame] | 11927 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 11928 | The following array contains the list of all map functions avalaible sorted by |
| 11929 | input type, match type and output type. |
Thierry FOURNIER | d5f624d | 2013-11-26 11:52:33 +0100 | [diff] [blame] | 11930 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 11931 | input type | match method | output type str | output type int | output type ip |
| 11932 | -----------+--------------+-----------------+-----------------+--------------- |
| 11933 | str | str | map_str | map_str_int | map_str_ip |
| 11934 | -----------+--------------+-----------------+-----------------+--------------- |
Willy Tarreau | 787a4c0 | 2014-05-10 07:55:30 +0200 | [diff] [blame] | 11935 | str | beg | map_beg | map_beg_int | map_end_ip |
| 11936 | -----------+--------------+-----------------+-----------------+--------------- |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 11937 | str | sub | map_sub | map_sub_int | map_sub_ip |
| 11938 | -----------+--------------+-----------------+-----------------+--------------- |
| 11939 | str | dir | map_dir | map_dir_int | map_dir_ip |
| 11940 | -----------+--------------+-----------------+-----------------+--------------- |
| 11941 | str | dom | map_dom | map_dom_int | map_dom_ip |
| 11942 | -----------+--------------+-----------------+-----------------+--------------- |
| 11943 | str | end | map_end | map_end_int | map_end_ip |
| 11944 | -----------+--------------+-----------------+-----------------+--------------- |
Thierry Fournier | 8feaa66 | 2016-02-10 22:55:20 +0100 | [diff] [blame] | 11945 | | | map_reg | | |
| 11946 | str | reg +-----------------+ map_reg_int | map_reg_ip |
| 11947 | | | map_regm | | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 11948 | -----------+--------------+-----------------+-----------------+--------------- |
| 11949 | int | int | map_int | map_int_int | map_int_ip |
| 11950 | -----------+--------------+-----------------+-----------------+--------------- |
| 11951 | ip | ip | map_ip | map_ip_int | map_ip_ip |
| 11952 | -----------+--------------+-----------------+-----------------+--------------- |
Thierry FOURNIER | d5f624d | 2013-11-26 11:52:33 +0100 | [diff] [blame] | 11953 | |
Thierry Fournier | 8feaa66 | 2016-02-10 22:55:20 +0100 | [diff] [blame] | 11954 | The special map called "map_regm" expect matching zone in the regular |
| 11955 | expression and modify the output replacing back reference (like "\1") by |
| 11956 | the corresponding match text. |
| 11957 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 11958 | The file contains one key + value per line. Lines which start with '#' are |
| 11959 | ignored, just like empty lines. Leading tabs and spaces are stripped. The key |
| 11960 | is then the first "word" (series of non-space/tabs characters), and the value |
| 11961 | is what follows this series of space/tab till the end of the line excluding |
| 11962 | trailing spaces/tabs. |
Thierry FOURNIER | d5f624d | 2013-11-26 11:52:33 +0100 | [diff] [blame] | 11963 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 11964 | Example : |
| 11965 | |
| 11966 | # this is a comment and is ignored |
| 11967 | 2.22.246.0/23 United Kingdom \n |
| 11968 | <-><-----------><--><------------><----> |
| 11969 | | | | | `- trailing spaces ignored |
| 11970 | | | | `---------- value |
| 11971 | | | `-------------------- middle spaces ignored |
| 11972 | | `---------------------------- key |
| 11973 | `------------------------------------ leading spaces ignored |
| 11974 | |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 11975 | mod(<value>) |
Thierry FOURNIER | 07ee64e | 2015-07-06 23:43:03 +0200 | [diff] [blame] | 11976 | Divides the input value of type signed integer by <value>, and returns the |
| 11977 | remainder as an signed integer. If <value> is null, then zero is returned. |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 11978 | <value> can be a numeric value or a variable name. The name of the variable |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 11979 | starts with an indication about its scope. The scopes allowed are: |
| 11980 | "sess" : the variable is shared with the whole session |
| 11981 | "txn" : the variable is shared with the transaction (request and response) |
| 11982 | "req" : the variable is shared only during request processing |
| 11983 | "res" : the variable is shared only during response processing |
| 11984 | This prefix is followed by a name. The separator is a '.'. The name may only |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 11985 | contain characters 'a-z', 'A-Z', '0-9' and '_'. |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 11986 | |
| 11987 | mul(<value>) |
Thierry FOURNIER | 07ee64e | 2015-07-06 23:43:03 +0200 | [diff] [blame] | 11988 | Multiplies the input value of type signed integer by <value>, and returns |
Thierry FOURNIER | 00c005c | 2015-07-08 01:10:21 +0200 | [diff] [blame] | 11989 | the product as an signed integer. In case of overflow, the largest possible |
| 11990 | value for the sign is returned so that the operation doesn't wrap around. |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 11991 | <value> can be a numeric value or a variable name. The name of the variable |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 11992 | starts with an indication about its scope. The scopes allowed are: |
| 11993 | "sess" : the variable is shared with the whole session |
| 11994 | "txn" : the variable is shared with the transaction (request and response) |
| 11995 | "req" : the variable is shared only during request processing |
| 11996 | "res" : the variable is shared only during response processing |
| 11997 | This prefix is followed by a name. The separator is a '.'. The name may only |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 11998 | contain characters 'a-z', 'A-Z', '0-9' and '_'. |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 11999 | |
| 12000 | neg |
Thierry FOURNIER | 07ee64e | 2015-07-06 23:43:03 +0200 | [diff] [blame] | 12001 | Takes the input value of type signed integer, computes the opposite value, |
| 12002 | and returns the remainder as an signed integer. 0 is identity. This operator |
| 12003 | is provided for reversed subtracts : in order to subtract the input from a |
| 12004 | constant, simply perform a "neg,add(value)". |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 12005 | |
| 12006 | not |
Thierry FOURNIER | 07ee64e | 2015-07-06 23:43:03 +0200 | [diff] [blame] | 12007 | Returns a boolean FALSE if the input value of type signed integer is |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 12008 | non-null, otherwise returns TRUE. Used in conjunction with and(), it can be |
| 12009 | used to report true/false for bit testing on input values (eg: verify the |
| 12010 | absence of a flag). |
| 12011 | |
| 12012 | odd |
Thierry FOURNIER | 07ee64e | 2015-07-06 23:43:03 +0200 | [diff] [blame] | 12013 | Returns a boolean TRUE if the input value of type signed integer is odd |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 12014 | otherwise returns FALSE. It is functionally equivalent to "and(1),bool". |
| 12015 | |
| 12016 | or(<value>) |
Thierry FOURNIER | 07ee64e | 2015-07-06 23:43:03 +0200 | [diff] [blame] | 12017 | Performs a bitwise "OR" between <value> and the input value of type signed |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 12018 | integer, and returns the result as an signed integer. <value> can be a |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 12019 | numeric value or a variable name. The name of the variable starts with an |
| 12020 | indication about its scope. The scopes allowed are: |
| 12021 | "sess" : the variable is shared with the whole session |
| 12022 | "txn" : the variable is shared with the transaction (request and response) |
| 12023 | "req" : the variable is shared only during request processing |
| 12024 | "res" : the variable is shared only during response processing |
| 12025 | This prefix is followed by a name. The separator is a '.'. The name may only |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 12026 | contain characters 'a-z', 'A-Z', '0-9' and '_'. |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 12027 | |
Willy Tarreau | c4dc350 | 2015-01-23 20:39:28 +0100 | [diff] [blame] | 12028 | regsub(<regex>,<subst>[,<flags>]) |
Willy Tarreau | 7eda849 | 2015-01-20 19:47:06 +0100 | [diff] [blame] | 12029 | Applies a regex-based substitution to the input string. It does the same |
| 12030 | operation as the well-known "sed" utility with "s/<regex>/<subst>/". By |
| 12031 | default it will replace in the input string the first occurrence of the |
| 12032 | largest part matching the regular expression <regex> with the substitution |
| 12033 | string <subst>. It is possible to replace all occurrences instead by adding |
| 12034 | the flag "g" in the third argument <flags>. It is also possible to make the |
| 12035 | regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a |
| 12036 | string, it is made up from the concatenation of all desired flags. Thus if |
| 12037 | both "i" and "g" are desired, using "gi" or "ig" will have the same effect. |
| 12038 | It is important to note that due to the current limitations of the |
Baptiste Assmann | 66025d8 | 2016-03-06 23:36:48 +0100 | [diff] [blame] | 12039 | configuration parser, some characters such as closing parenthesis, closing |
| 12040 | square brackets or comma are not possible to use in the arguments. The first |
| 12041 | use of this converter is to replace certain characters or sequence of |
| 12042 | characters with other ones. |
Willy Tarreau | 7eda849 | 2015-01-20 19:47:06 +0100 | [diff] [blame] | 12043 | |
| 12044 | Example : |
| 12045 | |
| 12046 | # de-duplicate "/" in header "x-path". |
| 12047 | # input: x-path: /////a///b/c/xzxyz/ |
| 12048 | # output: x-path: /a/b/c/xzxyz/ |
| 12049 | http-request set-header x-path %[hdr(x-path),regsub(/+,/,g)] |
| 12050 | |
Thierry FOURNIER | 35ab275 | 2015-05-28 13:22:03 +0200 | [diff] [blame] | 12051 | capture-req(<id>) |
| 12052 | Capture the string entry in the request slot <id> and returns the entry as |
| 12053 | is. If the slot doesn't exist, the capture fails silently. |
| 12054 | |
| 12055 | See also: "declare capture", "http-request capture", |
Baptiste Assmann | 5ac425c | 2015-10-21 23:13:46 +0200 | [diff] [blame] | 12056 | "http-response capture", "capture.req.hdr" and |
| 12057 | "capture.res.hdr" (sample fetches). |
Thierry FOURNIER | 35ab275 | 2015-05-28 13:22:03 +0200 | [diff] [blame] | 12058 | |
| 12059 | capture-res(<id>) |
| 12060 | Capture the string entry in the response slot <id> and returns the entry as |
| 12061 | is. If the slot doesn't exist, the capture fails silently. |
| 12062 | |
| 12063 | See also: "declare capture", "http-request capture", |
Baptiste Assmann | 5ac425c | 2015-10-21 23:13:46 +0200 | [diff] [blame] | 12064 | "http-response capture", "capture.req.hdr" and |
| 12065 | "capture.res.hdr" (sample fetches). |
Thierry FOURNIER | 35ab275 | 2015-05-28 13:22:03 +0200 | [diff] [blame] | 12066 | |
Willy Tarreau | 23ec4ca | 2014-07-15 20:15:37 +0200 | [diff] [blame] | 12067 | sdbm([<avalanche>]) |
| 12068 | Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM |
| 12069 | hash function. Optionally, it is possible to apply a full avalanche hash |
| 12070 | function to the output if the optional <avalanche> argument equals 1. This |
| 12071 | converter uses the same functions as used by the various hash-based load |
| 12072 | balancing algorithms, so it will provide exactly the same results. It is |
| 12073 | mostly intended for debugging, but can be used as a stick-table entry to |
| 12074 | collect rough statistics. It must not be used for security purposes as a |
Willy Tarreau | 8059977 | 2015-01-20 19:35:24 +0100 | [diff] [blame] | 12075 | 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6" and the |
| 12076 | "hash-type" directive. |
Willy Tarreau | 23ec4ca | 2014-07-15 20:15:37 +0200 | [diff] [blame] | 12077 | |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 12078 | set-var(<var name>) |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 12079 | Sets a variable with the input content and returns the content on the output as |
| 12080 | is. The variable keeps the value and the associated input type. The name of the |
| 12081 | variable starts with an indication about its scope. The scopes allowed are: |
| 12082 | "sess" : the variable is shared with the whole session |
| 12083 | "txn" : the variable is shared with the transaction (request and |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 12084 | response), |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 12085 | "req" : the variable is shared only during request processing, |
| 12086 | "res" : the variable is shared only during response processing. |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 12087 | This prefix is followed by a name. The separator is a '.'. The name may only |
| 12088 | contain characters 'a-z', 'A-Z', '0-9' and '_'. |
| 12089 | |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 12090 | sub(<value>) |
Thierry FOURNIER | 07ee64e | 2015-07-06 23:43:03 +0200 | [diff] [blame] | 12091 | Subtracts <value> from the input value of type signed integer, and returns |
| 12092 | the result as an signed integer. Note: in order to subtract the input from |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 12093 | a constant, simply perform a "neg,add(value)". <value> can be a numeric value |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 12094 | or a variable name. The name of the variable starts with an indication about |
| 12095 | its scope. The scopes allowed are: |
| 12096 | "sess" : the variable is shared with the whole session |
| 12097 | "txn" : the variable is shared with the transaction (request and |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 12098 | response), |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 12099 | "req" : the variable is shared only during request processing, |
| 12100 | "res" : the variable is shared only during response processing. |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 12101 | This prefix is followed by a name. The separator is a '.'. The name may only |
| 12102 | contain characters 'a-z', 'A-Z', '0-9' and '_'. |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 12103 | |
Willy Tarreau | d9f316a | 2014-07-10 14:03:38 +0200 | [diff] [blame] | 12104 | table_bytes_in_rate(<table>) |
| 12105 | Uses the string representation of the input sample to perform a look up in |
| 12106 | the specified table. If the key is not found in the table, integer value zero |
| 12107 | is returned. Otherwise the converter returns the average client-to-server |
| 12108 | bytes rate associated with the input sample in the designated table, measured |
| 12109 | in amount of bytes over the period configured in the table. See also the |
| 12110 | sc_bytes_in_rate sample fetch keyword. |
| 12111 | |
| 12112 | |
| 12113 | table_bytes_out_rate(<table>) |
| 12114 | Uses the string representation of the input sample to perform a look up in |
| 12115 | the specified table. If the key is not found in the table, integer value zero |
| 12116 | is returned. Otherwise the converter returns the average server-to-client |
| 12117 | bytes rate associated with the input sample in the designated table, measured |
| 12118 | in amount of bytes over the period configured in the table. See also the |
| 12119 | sc_bytes_out_rate sample fetch keyword. |
| 12120 | |
| 12121 | table_conn_cnt(<table>) |
| 12122 | Uses the string representation of the input sample to perform a look up in |
| 12123 | the specified table. If the key is not found in the table, integer value zero |
| 12124 | is returned. Otherwise the converter returns the cumulated amount of incoming |
| 12125 | connections associated with the input sample in the designated table. See |
| 12126 | also the sc_conn_cnt sample fetch keyword. |
| 12127 | |
| 12128 | table_conn_cur(<table>) |
| 12129 | Uses the string representation of the input sample to perform a look up in |
| 12130 | the specified table. If the key is not found in the table, integer value zero |
| 12131 | is returned. Otherwise the converter returns the current amount of concurrent |
| 12132 | tracked connections associated with the input sample in the designated table. |
| 12133 | See also the sc_conn_cur sample fetch keyword. |
| 12134 | |
| 12135 | table_conn_rate(<table>) |
| 12136 | Uses the string representation of the input sample to perform a look up in |
| 12137 | the specified table. If the key is not found in the table, integer value zero |
| 12138 | is returned. Otherwise the converter returns the average incoming connection |
| 12139 | rate associated with the input sample in the designated table. See also the |
| 12140 | sc_conn_rate sample fetch keyword. |
| 12141 | |
Thierry FOURNIER | 236657b | 2015-08-19 08:25:14 +0200 | [diff] [blame] | 12142 | table_gpt0(<table>) |
| 12143 | Uses the string representation of the input sample to perform a look up in |
| 12144 | the specified table. If the key is not found in the table, boolean value zero |
| 12145 | is returned. Otherwise the converter returns the current value of the first |
| 12146 | general purpose tag associated with the input sample in the designated table. |
| 12147 | See also the sc_get_gpt0 sample fetch keyword. |
| 12148 | |
Willy Tarreau | d9f316a | 2014-07-10 14:03:38 +0200 | [diff] [blame] | 12149 | table_gpc0(<table>) |
| 12150 | Uses the string representation of the input sample to perform a look up in |
| 12151 | the specified table. If the key is not found in the table, integer value zero |
| 12152 | is returned. Otherwise the converter returns the current value of the first |
| 12153 | general purpose counter associated with the input sample in the designated |
| 12154 | table. See also the sc_get_gpc0 sample fetch keyword. |
| 12155 | |
| 12156 | table_gpc0_rate(<table>) |
| 12157 | Uses the string representation of the input sample to perform a look up in |
| 12158 | the specified table. If the key is not found in the table, integer value zero |
| 12159 | is returned. Otherwise the converter returns the frequency which the gpc0 |
| 12160 | counter was incremented over the configured period in the table, associated |
| 12161 | with the input sample in the designated table. See also the sc_get_gpc0_rate |
| 12162 | sample fetch keyword. |
| 12163 | |
| 12164 | table_http_err_cnt(<table>) |
| 12165 | Uses the string representation of the input sample to perform a look up in |
| 12166 | the specified table. If the key is not found in the table, integer value zero |
| 12167 | is returned. Otherwise the converter returns the cumulated amount of HTTP |
| 12168 | errors associated with the input sample in the designated table. See also the |
| 12169 | sc_http_err_cnt sample fetch keyword. |
| 12170 | |
| 12171 | table_http_err_rate(<table>) |
| 12172 | Uses the string representation of the input sample to perform a look up in |
| 12173 | the specified table. If the key is not found in the table, integer value zero |
| 12174 | is returned. Otherwise the average rate of HTTP errors associated with the |
| 12175 | input sample in the designated table, measured in amount of errors over the |
| 12176 | period configured in the table. See also the sc_http_err_rate sample fetch |
| 12177 | keyword. |
| 12178 | |
| 12179 | table_http_req_cnt(<table>) |
| 12180 | Uses the string representation of the input sample to perform a look up in |
| 12181 | the specified table. If the key is not found in the table, integer value zero |
| 12182 | is returned. Otherwise the converter returns the cumulated amount of HTTP |
| 12183 | requests associated with the input sample in the designated table. See also |
| 12184 | the sc_http_req_cnt sample fetch keyword. |
| 12185 | |
| 12186 | table_http_req_rate(<table>) |
| 12187 | Uses the string representation of the input sample to perform a look up in |
| 12188 | the specified table. If the key is not found in the table, integer value zero |
| 12189 | is returned. Otherwise the average rate of HTTP requests associated with the |
| 12190 | input sample in the designated table, measured in amount of requests over the |
| 12191 | period configured in the table. See also the sc_http_req_rate sample fetch |
| 12192 | keyword. |
| 12193 | |
| 12194 | table_kbytes_in(<table>) |
| 12195 | Uses the string representation of the input sample to perform a look up in |
| 12196 | the specified table. If the key is not found in the table, integer value zero |
| 12197 | is returned. Otherwise the converter returns the cumulated amount of client- |
| 12198 | to-server data associated with the input sample in the designated table, |
| 12199 | measured in kilobytes. The test is currently performed on 32-bit integers, |
| 12200 | which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch |
| 12201 | keyword. |
| 12202 | |
| 12203 | table_kbytes_out(<table>) |
| 12204 | Uses the string representation of the input sample to perform a look up in |
| 12205 | the specified table. If the key is not found in the table, integer value zero |
| 12206 | is returned. Otherwise the converter returns the cumulated amount of server- |
| 12207 | to-client data associated with the input sample in the designated table, |
| 12208 | measured in kilobytes. The test is currently performed on 32-bit integers, |
| 12209 | which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch |
| 12210 | keyword. |
| 12211 | |
| 12212 | table_server_id(<table>) |
| 12213 | Uses the string representation of the input sample to perform a look up in |
| 12214 | the specified table. If the key is not found in the table, integer value zero |
| 12215 | is returned. Otherwise the converter returns the server ID associated with |
| 12216 | the input sample in the designated table. A server ID is associated to a |
| 12217 | sample by a "stick" rule when a connection to a server succeeds. A server ID |
| 12218 | zero means that no server is associated with this key. |
| 12219 | |
| 12220 | table_sess_cnt(<table>) |
| 12221 | Uses the string representation of the input sample to perform a look up in |
| 12222 | the specified table. If the key is not found in the table, integer value zero |
| 12223 | is returned. Otherwise the converter returns the cumulated amount of incoming |
| 12224 | sessions associated with the input sample in the designated table. Note that |
| 12225 | a session here refers to an incoming connection being accepted by the |
| 12226 | "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch |
| 12227 | keyword. |
| 12228 | |
| 12229 | table_sess_rate(<table>) |
| 12230 | Uses the string representation of the input sample to perform a look up in |
| 12231 | the specified table. If the key is not found in the table, integer value zero |
| 12232 | is returned. Otherwise the converter returns the average incoming session |
| 12233 | rate associated with the input sample in the designated table. Note that a |
| 12234 | session here refers to an incoming connection being accepted by the |
| 12235 | "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch |
| 12236 | keyword. |
| 12237 | |
| 12238 | table_trackers(<table>) |
| 12239 | Uses the string representation of the input sample to perform a look up in |
| 12240 | the specified table. If the key is not found in the table, integer value zero |
| 12241 | is returned. Otherwise the converter returns the current amount of concurrent |
| 12242 | connections tracking the same key as the input sample in the designated |
| 12243 | table. It differs from table_conn_cur in that it does not rely on any stored |
| 12244 | information but on the table's reference count (the "use" value which is |
| 12245 | returned by "show table" on the CLI). This may sometimes be more suited for |
| 12246 | layer7 tracking. It can be used to tell a server how many concurrent |
| 12247 | connections there are from a given address for example. See also the |
| 12248 | sc_trackers sample fetch keyword. |
| 12249 | |
Willy Tarreau | ffcb2e4 | 2014-07-10 16:29:08 +0200 | [diff] [blame] | 12250 | upper |
| 12251 | Convert a string sample to upper case. This can only be placed after a string |
| 12252 | sample fetch function or after a transformation keyword returning a string |
| 12253 | type. The result is of type string. |
| 12254 | |
Thierry FOURNIER | 82ff3c9 | 2015-05-07 15:46:20 +0200 | [diff] [blame] | 12255 | url_dec |
| 12256 | Takes an url-encoded string provided as input and returns the decoded |
| 12257 | version as output. The input and the output are of type string. |
| 12258 | |
Willy Tarreau | 0dbfdba | 2014-07-10 16:37:47 +0200 | [diff] [blame] | 12259 | utime(<format>[,<offset>]) |
| 12260 | Converts an integer supposed to contain a date since epoch to a string |
| 12261 | representing this date in UTC time using a format defined by the <format> |
| 12262 | string using strftime(3). The purpose is to allow any date format to be used |
| 12263 | in logs. An optional <offset> in seconds may be applied to the input date |
| 12264 | (positive or negative). See the strftime() man page for the format supported |
| 12265 | by your operating system. See also the ltime converter. |
| 12266 | |
| 12267 | Example : |
| 12268 | |
| 12269 | # Emit two colons, one with the UTC time and another with ip:port |
| 12270 | # Eg: 20140710162350 127.0.0.1:57325 |
| 12271 | log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp |
| 12272 | |
Emeric Brun | c9a0f6d | 2014-11-25 14:09:01 +0100 | [diff] [blame] | 12273 | word(<index>,<delimiters>) |
| 12274 | Extracts the nth word considering given delimiters from an input string. |
| 12275 | Indexes start at 1 and delimiters are a string formatted list of chars. |
| 12276 | |
Willy Tarreau | 23ec4ca | 2014-07-15 20:15:37 +0200 | [diff] [blame] | 12277 | wt6([<avalanche>]) |
| 12278 | Hashes a binary input sample into an unsigned 32-bit quantity using the WT6 |
| 12279 | hash function. Optionally, it is possible to apply a full avalanche hash |
| 12280 | function to the output if the optional <avalanche> argument equals 1. This |
| 12281 | converter uses the same functions as used by the various hash-based load |
| 12282 | balancing algorithms, so it will provide exactly the same results. It is |
| 12283 | mostly intended for debugging, but can be used as a stick-table entry to |
| 12284 | collect rough statistics. It must not be used for security purposes as a |
Willy Tarreau | 8059977 | 2015-01-20 19:35:24 +0100 | [diff] [blame] | 12285 | 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", and the |
| 12286 | "hash-type" directive. |
Willy Tarreau | 23ec4ca | 2014-07-15 20:15:37 +0200 | [diff] [blame] | 12287 | |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 12288 | xor(<value>) |
| 12289 | Performs a bitwise "XOR" (exclusive OR) between <value> and the input value |
Thierry FOURNIER | 07ee64e | 2015-07-06 23:43:03 +0200 | [diff] [blame] | 12290 | of type signed integer, and returns the result as an signed integer. |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 12291 | <value> can be a numeric value or a variable name. The name of the variable |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 12292 | starts with an indication about its scope. The scopes allowed are: |
| 12293 | "sess" : the variable is shared with the whole session |
| 12294 | "txn" : the variable is shared with the transaction (request and |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 12295 | response), |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 12296 | "req" : the variable is shared only during request processing, |
| 12297 | "res" : the variable is shared only during response processing. |
Thierry FOURNIER | 5d86fae | 2015-07-07 21:10:16 +0200 | [diff] [blame] | 12298 | This prefix is followed by a name. The separator is a '.'. The name may only |
| 12299 | contain characters 'a-z', 'A-Z', '0-9' and '_'. |
Willy Tarreau | 9770787 | 2015-01-27 15:12:13 +0100 | [diff] [blame] | 12300 | |
Thierry FOURNIER | d5f624d | 2013-11-26 11:52:33 +0100 | [diff] [blame] | 12301 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 12302 | 7.3.2. Fetching samples from internal states |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12303 | -------------------------------------------- |
| 12304 | |
| 12305 | A first set of sample fetch methods applies to internal information which does |
| 12306 | not even relate to any client information. These ones are sometimes used with |
| 12307 | "monitor-fail" directives to report an internal status to external watchers. |
| 12308 | The sample fetch methods described in this section are usable anywhere. |
| 12309 | |
| 12310 | always_false : boolean |
| 12311 | Always returns the boolean "false" value. It may be used with ACLs as a |
| 12312 | temporary replacement for another one when adjusting configurations. |
| 12313 | |
| 12314 | always_true : boolean |
| 12315 | Always returns the boolean "true" value. It may be used with ACLs as a |
| 12316 | temporary replacement for another one when adjusting configurations. |
| 12317 | |
| 12318 | avg_queue([<backend>]) : integer |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 12319 | Returns the total number of queued connections of the designated backend |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12320 | divided by the number of active servers. The current backend is used if no |
| 12321 | backend is specified. This is very similar to "queue" except that the size of |
| 12322 | the farm is considered, in order to give a more accurate measurement of the |
| 12323 | time it may take for a new connection to be processed. The main usage is with |
| 12324 | ACL to return a sorry page to new users when it becomes certain they will get |
| 12325 | a degraded service, or to pass to the backend servers in a header so that |
| 12326 | they decide to work in degraded mode or to disable some functions to speed up |
| 12327 | the processing a bit. Note that in the event there would not be any active |
| 12328 | server anymore, twice the number of queued connections would be considered as |
| 12329 | the measured value. This is a fair estimate, as we expect one server to get |
| 12330 | back soon anyway, but we still prefer to send new traffic to another backend |
| 12331 | if in better shape. See also the "queue", "be_conn", and "be_sess_rate" |
| 12332 | sample fetches. |
Krzysztof Piotr Oledzki | 346f76d | 2010-01-12 21:59:30 +0100 | [diff] [blame] | 12333 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12334 | be_conn([<backend>]) : integer |
Willy Tarreau | a36af91 | 2009-10-10 12:02:45 +0200 | [diff] [blame] | 12335 | Applies to the number of currently established connections on the backend, |
| 12336 | possibly including the connection being evaluated. If no backend name is |
| 12337 | specified, the current one is used. But it is also possible to check another |
| 12338 | backend. It can be used to use a specific farm when the nominal one is full. |
| 12339 | See also the "fe_conn", "queue" and "be_sess_rate" criteria. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 12340 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12341 | be_sess_rate([<backend>]) : integer |
| 12342 | Returns an integer value corresponding to the sessions creation rate on the |
| 12343 | backend, in number of new sessions per second. This is used with ACLs to |
| 12344 | switch to an alternate backend when an expensive or fragile one reaches too |
| 12345 | high a session rate, or to limit abuse of service (eg. prevent sucking of an |
| 12346 | online dictionary). It can also be useful to add this element to logs using a |
| 12347 | log-format directive. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 12348 | |
| 12349 | Example : |
| 12350 | # Redirect to an error page if the dictionary is requested too often |
| 12351 | backend dynamic |
| 12352 | mode http |
| 12353 | acl being_scanned be_sess_rate gt 100 |
| 12354 | redirect location /denied.html if being_scanned |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 12355 | |
Thierry FOURNIER | cc10329 | 2015-06-06 19:30:17 +0200 | [diff] [blame] | 12356 | bin(<hexa>) : bin |
| 12357 | Returns a binary chain. The input is the hexadecimal representation |
| 12358 | of the string. |
| 12359 | |
| 12360 | bool(<bool>) : bool |
| 12361 | Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'. |
| 12362 | 'false' and '0' are the same. 'true' and '1' are the same. |
| 12363 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12364 | connslots([<backend>]) : integer |
| 12365 | Returns an integer value corresponding to the number of connection slots |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 12366 | still available in the backend, by totaling the maximum amount of |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12367 | connections on all servers and the maximum queue size. This is probably only |
| 12368 | used with ACLs. |
Tait Clarridge | 7896d52 | 2012-12-05 21:39:31 -0500 | [diff] [blame] | 12369 | |
Jeffrey 'jf' Lim | 5051d7b | 2008-09-04 01:03:03 +0800 | [diff] [blame] | 12370 | The basic idea here is to be able to measure the number of connection "slots" |
Willy Tarreau | 55165fe | 2009-05-10 12:02:55 +0200 | [diff] [blame] | 12371 | still available (connection + queue), so that anything beyond that (intended |
Jeffrey 'jf' Lim | 5051d7b | 2008-09-04 01:03:03 +0800 | [diff] [blame] | 12372 | usage; see "use_backend" keyword) can be redirected to a different backend. |
| 12373 | |
Willy Tarreau | 55165fe | 2009-05-10 12:02:55 +0200 | [diff] [blame] | 12374 | 'connslots' = number of available server connection slots, + number of |
| 12375 | available server queue slots. |
Jeffrey 'jf' Lim | 5051d7b | 2008-09-04 01:03:03 +0800 | [diff] [blame] | 12376 | |
Willy Tarreau | a36af91 | 2009-10-10 12:02:45 +0200 | [diff] [blame] | 12377 | Note that while "fe_conn" may be used, "connslots" comes in especially |
Willy Tarreau | 55165fe | 2009-05-10 12:02:55 +0200 | [diff] [blame] | 12378 | useful when you have a case of traffic going to one single ip, splitting into |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12379 | multiple backends (perhaps using ACLs to do name-based load balancing) and |
Willy Tarreau | 55165fe | 2009-05-10 12:02:55 +0200 | [diff] [blame] | 12380 | you want to be able to differentiate between different backends, and their |
| 12381 | available "connslots". Also, whereas "nbsrv" only measures servers that are |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12382 | actually *down*, this fetch is more fine-grained and looks into the number of |
Willy Tarreau | a36af91 | 2009-10-10 12:02:45 +0200 | [diff] [blame] | 12383 | available connection slots as well. See also "queue" and "avg_queue". |
Jeffrey 'jf' Lim | 5051d7b | 2008-09-04 01:03:03 +0800 | [diff] [blame] | 12384 | |
Willy Tarreau | 55165fe | 2009-05-10 12:02:55 +0200 | [diff] [blame] | 12385 | OTHER CAVEATS AND NOTES: at this point in time, the code does not take care |
| 12386 | of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0, |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12387 | then this fetch clearly does not make sense, in which case the value returned |
Willy Tarreau | 55165fe | 2009-05-10 12:02:55 +0200 | [diff] [blame] | 12388 | will be -1. |
Jeffrey 'jf' Lim | 5051d7b | 2008-09-04 01:03:03 +0800 | [diff] [blame] | 12389 | |
Willy Tarreau | 6236d3a | 2013-07-25 14:28:25 +0200 | [diff] [blame] | 12390 | date([<offset>]) : integer |
| 12391 | Returns the current date as the epoch (number of seconds since 01/01/1970). |
| 12392 | If an offset value is specified, then it is a number of seconds that is added |
| 12393 | to the current date before returning the value. This is particularly useful |
| 12394 | to compute relative dates, as both positive and negative offsets are allowed. |
Willy Tarreau | 276fae9 | 2013-07-25 14:36:01 +0200 | [diff] [blame] | 12395 | It is useful combined with the http_date converter. |
| 12396 | |
| 12397 | Example : |
| 12398 | |
| 12399 | # set an expires header to now+1 hour in every response |
| 12400 | http-response set-header Expires %[date(3600),http_date] |
Willy Tarreau | 6236d3a | 2013-07-25 14:28:25 +0200 | [diff] [blame] | 12401 | |
Willy Tarreau | 595ec54 | 2013-06-12 21:34:28 +0200 | [diff] [blame] | 12402 | env(<name>) : string |
| 12403 | Returns a string containing the value of environment variable <name>. As a |
| 12404 | reminder, environment variables are per-process and are sampled when the |
| 12405 | process starts. This can be useful to pass some information to a next hop |
| 12406 | server, or with ACLs to take specific action when the process is started a |
| 12407 | certain way. |
| 12408 | |
| 12409 | Examples : |
| 12410 | # Pass the Via header to next hop with the local hostname in it |
| 12411 | http-request add-header Via 1.1\ %[env(HOSTNAME)] |
| 12412 | |
| 12413 | # reject cookie-less requests when the STOP environment variable is set |
| 12414 | http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found } |
| 12415 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12416 | fe_conn([<frontend>]) : integer |
| 12417 | Returns the number of currently established connections on the frontend, |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 12418 | possibly including the connection being evaluated. If no frontend name is |
| 12419 | specified, the current one is used. But it is also possible to check another |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12420 | frontend. It can be used to return a sorry page before hard-blocking, or to |
| 12421 | use a specific backend to drain new requests when the farm is considered |
| 12422 | full. This is mostly used with ACLs but can also be used to pass some |
| 12423 | statistics to servers in HTTP headers. See also the "dst_conn", "be_conn", |
| 12424 | "fe_sess_rate" fetches. |
Willy Tarreau | a36af91 | 2009-10-10 12:02:45 +0200 | [diff] [blame] | 12425 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12426 | fe_sess_rate([<frontend>]) : integer |
| 12427 | Returns an integer value corresponding to the sessions creation rate on the |
| 12428 | frontend, in number of new sessions per second. This is used with ACLs to |
| 12429 | limit the incoming session rate to an acceptable range in order to prevent |
| 12430 | abuse of service at the earliest moment, for example when combined with other |
| 12431 | layer 4 ACLs in order to force the clients to wait a bit for the rate to go |
| 12432 | down below the limit. It can also be useful to add this element to logs using |
| 12433 | a log-format directive. See also the "rate-limit sessions" directive for use |
| 12434 | in frontends. |
Willy Tarreau | 079ff0a | 2009-03-05 21:34:28 +0100 | [diff] [blame] | 12435 | |
| 12436 | Example : |
| 12437 | # This frontend limits incoming mails to 10/s with a max of 100 |
| 12438 | # concurrent connections. We accept any connection below 10/s, and |
| 12439 | # force excess clients to wait for 100 ms. Since clients are limited to |
| 12440 | # 100 max, there cannot be more than 10 incoming mails per second. |
| 12441 | frontend mail |
| 12442 | bind :25 |
| 12443 | mode tcp |
| 12444 | maxconn 100 |
| 12445 | acl too_fast fe_sess_rate ge 10 |
| 12446 | tcp-request inspect-delay 100ms |
| 12447 | tcp-request content accept if ! too_fast |
| 12448 | tcp-request content accept if WAIT_END |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 12449 | |
Thierry FOURNIER | 07ee64e | 2015-07-06 23:43:03 +0200 | [diff] [blame] | 12450 | int(<integer>) : signed integer |
| 12451 | Returns a signed integer. |
| 12452 | |
Thierry FOURNIER | cc10329 | 2015-06-06 19:30:17 +0200 | [diff] [blame] | 12453 | ipv4(<ipv4>) : ipv4 |
| 12454 | Returns an ipv4. |
| 12455 | |
| 12456 | ipv6(<ipv6>) : ipv6 |
| 12457 | Returns an ipv6. |
| 12458 | |
| 12459 | meth(<method>) : method |
| 12460 | Returns a method. |
| 12461 | |
Willy Tarreau | 0f30d26 | 2014-11-24 16:02:05 +0100 | [diff] [blame] | 12462 | nbproc : integer |
| 12463 | Returns an integer value corresponding to the number of processes that were |
| 12464 | started (it equals the global "nbproc" setting). This is useful for logging |
| 12465 | and debugging purposes. |
| 12466 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12467 | nbsrv([<backend>]) : integer |
| 12468 | Returns an integer value corresponding to the number of usable servers of |
| 12469 | either the current backend or the named backend. This is mostly used with |
| 12470 | ACLs but can also be useful when added to logs. This is normally used to |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 12471 | switch to an alternate backend when the number of servers is too low to |
| 12472 | to handle some load. It is useful to report a failure when combined with |
| 12473 | "monitor fail". |
Willy Tarreau | 079ff0a | 2009-03-05 21:34:28 +0100 | [diff] [blame] | 12474 | |
Willy Tarreau | 0f30d26 | 2014-11-24 16:02:05 +0100 | [diff] [blame] | 12475 | proc : integer |
| 12476 | Returns an integer value corresponding to the position of the process calling |
| 12477 | the function, between 1 and global.nbproc. This is useful for logging and |
| 12478 | debugging purposes. |
| 12479 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12480 | queue([<backend>]) : integer |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 12481 | Returns the total number of queued connections of the designated backend, |
| 12482 | including all the connections in server queues. If no backend name is |
| 12483 | specified, the current one is used, but it is also possible to check another |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12484 | one. This is useful with ACLs or to pass statistics to backend servers. This |
| 12485 | can be used to take actions when queuing goes above a known level, generally |
| 12486 | indicating a surge of traffic or a massive slowdown on the servers. One |
| 12487 | possible action could be to reject new users but still accept old ones. See |
| 12488 | also the "avg_queue", "be_conn", and "be_sess_rate" fetches. |
| 12489 | |
Willy Tarreau | 84310e2 | 2014-02-14 11:59:04 +0100 | [diff] [blame] | 12490 | rand([<range>]) : integer |
| 12491 | Returns a random integer value within a range of <range> possible values, |
| 12492 | starting at zero. If the range is not specified, it defaults to 2^32, which |
| 12493 | gives numbers between 0 and 4294967295. It can be useful to pass some values |
| 12494 | needed to take some routing decisions for example, or just for debugging |
| 12495 | purposes. This random must not be used for security purposes. |
| 12496 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12497 | srv_conn([<backend>/]<server>) : integer |
| 12498 | Returns an integer value corresponding to the number of currently established |
| 12499 | connections on the designated server, possibly including the connection being |
| 12500 | evaluated. If <backend> is omitted, then the server is looked up in the |
| 12501 | current backend. It can be used to use a specific farm when one server is |
| 12502 | full, or to inform the server about our view of the number of active |
| 12503 | connections with it. See also the "fe_conn", "be_conn" and "queue" fetch |
| 12504 | methods. |
| 12505 | |
| 12506 | srv_is_up([<backend>/]<server>) : boolean |
| 12507 | Returns true when the designated server is UP, and false when it is either |
| 12508 | DOWN or in maintenance mode. If <backend> is omitted, then the server is |
| 12509 | looked up in the current backend. It is mainly used to take action based on |
| 12510 | an external status reported via a health check (eg: a geographical site's |
| 12511 | availability). Another possible use which is more of a hack consists in |
| 12512 | using dummy servers as boolean variables that can be enabled or disabled from |
| 12513 | the CLI, so that rules depending on those ACLs can be tweaked in realtime. |
| 12514 | |
| 12515 | srv_sess_rate([<backend>/]<server>) : integer |
| 12516 | Returns an integer corresponding to the sessions creation rate on the |
| 12517 | designated server, in number of new sessions per second. If <backend> is |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 12518 | omitted, then the server is looked up in the current backend. This is mostly |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12519 | used with ACLs but can make sense with logs too. This is used to switch to an |
| 12520 | alternate backend when an expensive or fragile one reaches too high a session |
| 12521 | rate, or to limit abuse of service (eg. prevent latent requests from |
| 12522 | overloading servers). |
| 12523 | |
| 12524 | Example : |
| 12525 | # Redirect to a separate back |
| 12526 | acl srv1_full srv_sess_rate(be1/srv1) gt 50 |
| 12527 | acl srv2_full srv_sess_rate(be1/srv2) gt 50 |
| 12528 | use_backend be2 if srv1_full or srv2_full |
| 12529 | |
Willy Tarreau | 0f30d26 | 2014-11-24 16:02:05 +0100 | [diff] [blame] | 12530 | stopping : boolean |
| 12531 | Returns TRUE if the process calling the function is currently stopping. This |
| 12532 | can be useful for logging, or for relaxing certain checks or helping close |
| 12533 | certain connections upon graceful shutdown. |
| 12534 | |
Thierry FOURNIER | cc10329 | 2015-06-06 19:30:17 +0200 | [diff] [blame] | 12535 | str(<string>) : string |
| 12536 | Returns a string. |
| 12537 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12538 | table_avl([<table>]) : integer |
| 12539 | Returns the total number of available entries in the current proxy's |
| 12540 | stick-table or in the designated stick-table. See also table_cnt. |
| 12541 | |
| 12542 | table_cnt([<table>]) : integer |
| 12543 | Returns the total number of entries currently in use in the current proxy's |
| 12544 | stick-table or in the designated stick-table. See also src_conn_cnt and |
| 12545 | table_avl for other entry counting methods. |
| 12546 | |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 12547 | var(<var-name>) : undefined |
| 12548 | Returns a variable with the stored type. If the variable is not set, the |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 12549 | sample fetch fails. The name of the variable starts with an indication |
| 12550 | about its scope. The scopes allowed are: |
| 12551 | "sess" : the variable is shared with the whole session |
| 12552 | "txn" : the variable is shared with the transaction (request and |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 12553 | response), |
Daniel Schneller | 0b54705 | 2016-03-21 20:46:57 +0100 | [diff] [blame] | 12554 | "req" : the variable is shared only during request processing, |
| 12555 | "res" : the variable is shared only during response processing. |
Thierry FOURNIER | 4834bc7 | 2015-06-06 19:29:07 +0200 | [diff] [blame] | 12556 | This prefix is followed by a name. The separator is a '.'. The name may only |
| 12557 | contain characters 'a-z', 'A-Z', '0-9' and '_'. |
| 12558 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 12559 | 7.3.3. Fetching samples at Layer 4 |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12560 | ---------------------------------- |
| 12561 | |
| 12562 | The layer 4 usually describes just the transport layer which in haproxy is |
| 12563 | closest to the connection, where no content is yet made available. The fetch |
| 12564 | methods described here are usable as low as the "tcp-request connection" rule |
| 12565 | sets unless they require some future information. Those generally include |
| 12566 | TCP/IP addresses and ports, as well as elements from stick-tables related to |
Willy Tarreau | 4d4149c | 2013-07-23 19:33:46 +0200 | [diff] [blame] | 12567 | the incoming connection. For retrieving a value from a sticky counters, the |
| 12568 | counter number can be explicitly set as 0, 1, or 2 using the pre-defined |
| 12569 | "sc0_", "sc1_", or "sc2_" prefix, or it can be specified as the first integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12570 | argument when using the "sc_" prefix. An optional table may be specified with |
| 12571 | the "sc*" form, in which case the currently tracked key will be looked up into |
| 12572 | this alternate table instead of the table currently being tracked. |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12573 | |
| 12574 | be_id : integer |
| 12575 | Returns an integer containing the current backend's id. It can be used in |
| 12576 | frontends with responses to check which backend processed the request. |
| 12577 | |
| 12578 | dst : ip |
| 12579 | This is the destination IPv4 address of the connection on the client side, |
| 12580 | which is the address the client connected to. It can be useful when running |
| 12581 | in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables. |
| 12582 | On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to |
| 12583 | RFC 4291. |
| 12584 | |
| 12585 | dst_conn : integer |
| 12586 | Returns an integer value corresponding to the number of currently established |
| 12587 | connections on the same socket including the one being evaluated. It is |
| 12588 | normally used with ACLs but can as well be used to pass the information to |
| 12589 | servers in an HTTP header or in logs. It can be used to either return a sorry |
| 12590 | page before hard-blocking, or to use a specific backend to drain new requests |
| 12591 | when the socket is considered saturated. This offers the ability to assign |
| 12592 | different limits to different listening ports or addresses. See also the |
| 12593 | "fe_conn" and "be_conn" fetches. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 12594 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12595 | dst_port : integer |
| 12596 | Returns an integer value corresponding to the destination TCP port of the |
| 12597 | connection on the client side, which is the port the client connected to. |
| 12598 | This might be used when running in transparent mode, when assigning dynamic |
| 12599 | ports to some clients for a whole application session, to stick all users to |
| 12600 | a same server, or to pass the destination port information to a server using |
| 12601 | an HTTP header. |
| 12602 | |
| 12603 | fe_id : integer |
| 12604 | Returns an integer containing the current frontend's id. It can be used in |
| 12605 | backends to check from which backend it was called, or to stick all users |
| 12606 | coming via a same frontend to the same server. |
| 12607 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12608 | sc_bytes_in_rate(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12609 | sc0_bytes_in_rate([<table>]) : integer |
| 12610 | sc1_bytes_in_rate([<table>]) : integer |
| 12611 | sc2_bytes_in_rate([<table>]) : integer |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12612 | Returns the average client-to-server bytes rate from the currently tracked |
| 12613 | counters, measured in amount of bytes over the period configured in the |
| 12614 | table. See also src_bytes_in_rate. |
| 12615 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12616 | sc_bytes_out_rate(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12617 | sc0_bytes_out_rate([<table>]) : integer |
| 12618 | sc1_bytes_out_rate([<table>]) : integer |
| 12619 | sc2_bytes_out_rate([<table>]) : integer |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12620 | Returns the average server-to-client bytes rate from the currently tracked |
| 12621 | counters, measured in amount of bytes over the period configured in the |
| 12622 | table. See also src_bytes_out_rate. |
| 12623 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12624 | sc_clr_gpc0(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12625 | sc0_clr_gpc0([<table>]) : integer |
| 12626 | sc1_clr_gpc0([<table>]) : integer |
| 12627 | sc2_clr_gpc0([<table>]) : integer |
Willy Tarreau | f73cd11 | 2011-08-13 01:45:16 +0200 | [diff] [blame] | 12628 | Clears the first General Purpose Counter associated to the currently tracked |
| 12629 | counters, and returns its previous value. Before the first invocation, the |
Willy Tarreau | 869948b | 2013-01-04 14:14:57 +0100 | [diff] [blame] | 12630 | stored value is zero, so first invocation will always return zero. This is |
| 12631 | typically used as a second ACL in an expression in order to mark a connection |
| 12632 | when a first ACL was verified : |
Willy Tarreau | f73cd11 | 2011-08-13 01:45:16 +0200 | [diff] [blame] | 12633 | |
| 12634 | # block if 5 consecutive requests continue to come faster than 10 sess |
| 12635 | # per second, and reset the counter as soon as the traffic slows down. |
Willy Tarreau | be4a3ef | 2013-06-17 15:04:07 +0200 | [diff] [blame] | 12636 | acl abuse sc0_http_req_rate gt 10 |
| 12637 | acl kill sc0_inc_gpc0 gt 5 |
| 12638 | acl save sc0_clr_gpc0 ge 0 |
Willy Tarreau | f73cd11 | 2011-08-13 01:45:16 +0200 | [diff] [blame] | 12639 | tcp-request connection accept if !abuse save |
| 12640 | tcp-request connection reject if abuse kill |
| 12641 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12642 | sc_conn_cnt(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12643 | sc0_conn_cnt([<table>]) : integer |
| 12644 | sc1_conn_cnt([<table>]) : integer |
| 12645 | sc2_conn_cnt([<table>]) : integer |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12646 | Returns the cumulated number of incoming connections from currently tracked |
| 12647 | counters. See also src_conn_cnt. |
| 12648 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12649 | sc_conn_cur(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12650 | sc0_conn_cur([<table>]) : integer |
| 12651 | sc1_conn_cur([<table>]) : integer |
| 12652 | sc2_conn_cur([<table>]) : integer |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12653 | Returns the current amount of concurrent connections tracking the same |
| 12654 | tracked counters. This number is automatically incremented when tracking |
| 12655 | begins and decremented when tracking stops. See also src_conn_cur. |
| 12656 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12657 | sc_conn_rate(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12658 | sc0_conn_rate([<table>]) : integer |
| 12659 | sc1_conn_rate([<table>]) : integer |
| 12660 | sc2_conn_rate([<table>]) : integer |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12661 | Returns the average connection rate from the currently tracked counters, |
| 12662 | measured in amount of connections over the period configured in the table. |
| 12663 | See also src_conn_rate. |
| 12664 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12665 | sc_get_gpc0(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12666 | sc0_get_gpc0([<table>]) : integer |
| 12667 | sc1_get_gpc0([<table>]) : integer |
| 12668 | sc2_get_gpc0([<table>]) : integer |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12669 | Returns the value of the first General Purpose Counter associated to the |
Willy Tarreau | 4d4149c | 2013-07-23 19:33:46 +0200 | [diff] [blame] | 12670 | currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0. |
Willy Tarreau | ba2ffd1 | 2013-05-29 15:54:14 +0200 | [diff] [blame] | 12671 | |
Thierry FOURNIER | 236657b | 2015-08-19 08:25:14 +0200 | [diff] [blame] | 12672 | sc_get_gpt0(<ctr>[,<table>]) : integer |
| 12673 | sc0_get_gpt0([<table>]) : integer |
| 12674 | sc1_get_gpt0([<table>]) : integer |
| 12675 | sc2_get_gpt0([<table>]) : integer |
| 12676 | Returns the value of the first General Purpose Tag associated to the |
| 12677 | currently tracked counters. See also src_get_gpt0. |
| 12678 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12679 | sc_gpc0_rate(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12680 | sc0_gpc0_rate([<table>]) : integer |
| 12681 | sc1_gpc0_rate([<table>]) : integer |
| 12682 | sc2_gpc0_rate([<table>]) : integer |
Willy Tarreau | ba2ffd1 | 2013-05-29 15:54:14 +0200 | [diff] [blame] | 12683 | Returns the average increment rate of the first General Purpose Counter |
| 12684 | associated to the currently tracked counters. It reports the frequency |
| 12685 | which the gpc0 counter was incremented over the configured period. See also |
Willy Tarreau | 4d4149c | 2013-07-23 19:33:46 +0200 | [diff] [blame] | 12686 | src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note |
| 12687 | that the "gpc0_rate" counter must be stored in the stick-table for a value to |
| 12688 | be returned, as "gpc0" only holds the event count. |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12689 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12690 | sc_http_err_cnt(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12691 | sc0_http_err_cnt([<table>]) : integer |
| 12692 | sc1_http_err_cnt([<table>]) : integer |
| 12693 | sc2_http_err_cnt([<table>]) : integer |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12694 | Returns the cumulated number of HTTP errors from the currently tracked |
| 12695 | counters. This includes the both request errors and 4xx error responses. |
| 12696 | See also src_http_err_cnt. |
| 12697 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12698 | sc_http_err_rate(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12699 | sc0_http_err_rate([<table>]) : integer |
| 12700 | sc1_http_err_rate([<table>]) : integer |
| 12701 | sc2_http_err_rate([<table>]) : integer |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12702 | Returns the average rate of HTTP errors from the currently tracked counters, |
| 12703 | measured in amount of errors over the period configured in the table. This |
| 12704 | includes the both request errors and 4xx error responses. See also |
| 12705 | src_http_err_rate. |
| 12706 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12707 | sc_http_req_cnt(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12708 | sc0_http_req_cnt([<table>]) : integer |
| 12709 | sc1_http_req_cnt([<table>]) : integer |
| 12710 | sc2_http_req_cnt([<table>]) : integer |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12711 | Returns the cumulated number of HTTP requests from the currently tracked |
| 12712 | counters. This includes every started request, valid or not. See also |
| 12713 | src_http_req_cnt. |
| 12714 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12715 | sc_http_req_rate(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12716 | sc0_http_req_rate([<table>]) : integer |
| 12717 | sc1_http_req_rate([<table>]) : integer |
| 12718 | sc2_http_req_rate([<table>]) : integer |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12719 | Returns the average rate of HTTP requests from the currently tracked |
| 12720 | counters, measured in amount of requests over the period configured in |
| 12721 | the table. This includes every started request, valid or not. See also |
| 12722 | src_http_req_rate. |
| 12723 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12724 | sc_inc_gpc0(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12725 | sc0_inc_gpc0([<table>]) : integer |
| 12726 | sc1_inc_gpc0([<table>]) : integer |
| 12727 | sc2_inc_gpc0([<table>]) : integer |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12728 | Increments the first General Purpose Counter associated to the currently |
Willy Tarreau | 869948b | 2013-01-04 14:14:57 +0100 | [diff] [blame] | 12729 | tracked counters, and returns its new value. Before the first invocation, |
| 12730 | the stored value is zero, so first invocation will increase it to 1 and will |
| 12731 | return 1. This is typically used as a second ACL in an expression in order |
| 12732 | to mark a connection when a first ACL was verified : |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12733 | |
Willy Tarreau | be4a3ef | 2013-06-17 15:04:07 +0200 | [diff] [blame] | 12734 | acl abuse sc0_http_req_rate gt 10 |
| 12735 | acl kill sc0_inc_gpc0 gt 0 |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12736 | tcp-request connection reject if abuse kill |
| 12737 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12738 | sc_kbytes_in(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12739 | sc0_kbytes_in([<table>]) : integer |
| 12740 | sc1_kbytes_in([<table>]) : integer |
| 12741 | sc2_kbytes_in([<table>]) : integer |
Willy Tarreau | a01b974 | 2014-07-10 15:29:24 +0200 | [diff] [blame] | 12742 | Returns the total amount of client-to-server data from the currently tracked |
| 12743 | counters, measured in kilobytes. The test is currently performed on 32-bit |
| 12744 | integers, which limits values to 4 terabytes. See also src_kbytes_in. |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12745 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12746 | sc_kbytes_out(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12747 | sc0_kbytes_out([<table>]) : integer |
| 12748 | sc1_kbytes_out([<table>]) : integer |
| 12749 | sc2_kbytes_out([<table>]) : integer |
Willy Tarreau | a01b974 | 2014-07-10 15:29:24 +0200 | [diff] [blame] | 12750 | Returns the total amount of server-to-client data from the currently tracked |
| 12751 | counters, measured in kilobytes. The test is currently performed on 32-bit |
| 12752 | integers, which limits values to 4 terabytes. See also src_kbytes_out. |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12753 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12754 | sc_sess_cnt(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12755 | sc0_sess_cnt([<table>]) : integer |
| 12756 | sc1_sess_cnt([<table>]) : integer |
| 12757 | sc2_sess_cnt([<table>]) : integer |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12758 | Returns the cumulated number of incoming connections that were transformed |
| 12759 | into sessions, which means that they were accepted by a "tcp-request |
| 12760 | connection" rule, from the currently tracked counters. A backend may count |
| 12761 | more sessions than connections because each connection could result in many |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 12762 | backend sessions if some HTTP keep-alive is performed over the connection |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12763 | with the client. See also src_sess_cnt. |
| 12764 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12765 | sc_sess_rate(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12766 | sc0_sess_rate([<table>]) : integer |
| 12767 | sc1_sess_rate([<table>]) : integer |
| 12768 | sc2_sess_rate([<table>]) : integer |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12769 | Returns the average session rate from the currently tracked counters, |
| 12770 | measured in amount of sessions over the period configured in the table. A |
| 12771 | session is a connection that got past the early "tcp-request connection" |
| 12772 | rules. A backend may count more sessions than connections because each |
| 12773 | connection could result in many backend sessions if some HTTP keep-alive is |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 12774 | performed over the connection with the client. See also src_sess_rate. |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12775 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12776 | sc_tracked(<ctr>[,<table>]) : boolean |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12777 | sc0_tracked([<table>]) : boolean |
| 12778 | sc1_tracked([<table>]) : boolean |
| 12779 | sc2_tracked([<table>]) : boolean |
Willy Tarreau | 6f1615f | 2013-06-03 15:15:22 +0200 | [diff] [blame] | 12780 | Returns true if the designated session counter is currently being tracked by |
| 12781 | the current session. This can be useful when deciding whether or not we want |
| 12782 | to set some values in a header passed to the server. |
| 12783 | |
Cyril Bonté | 62ba870 | 2014-04-22 23:52:25 +0200 | [diff] [blame] | 12784 | sc_trackers(<ctr>[,<table>]) : integer |
Willy Tarreau | 0f791d4 | 2013-07-23 19:56:43 +0200 | [diff] [blame] | 12785 | sc0_trackers([<table>]) : integer |
| 12786 | sc1_trackers([<table>]) : integer |
| 12787 | sc2_trackers([<table>]) : integer |
Willy Tarreau | 2406db4 | 2012-12-09 12:16:43 +0100 | [diff] [blame] | 12788 | Returns the current amount of concurrent connections tracking the same |
| 12789 | tracked counters. This number is automatically incremented when tracking |
Willy Tarreau | be4a3ef | 2013-06-17 15:04:07 +0200 | [diff] [blame] | 12790 | begins and decremented when tracking stops. It differs from sc0_conn_cur in |
Willy Tarreau | 2406db4 | 2012-12-09 12:16:43 +0100 | [diff] [blame] | 12791 | that it does not rely on any stored information but on the table's reference |
| 12792 | count (the "use" value which is returned by "show table" on the CLI). This |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12793 | may sometimes be more suited for layer7 tracking. It can be used to tell a |
| 12794 | server how many concurrent connections there are from a given address for |
| 12795 | example. |
Willy Tarreau | 2406db4 | 2012-12-09 12:16:43 +0100 | [diff] [blame] | 12796 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12797 | so_id : integer |
| 12798 | Returns an integer containing the current listening socket's id. It is useful |
| 12799 | in frontends involving many "bind" lines, or to stick all users coming via a |
| 12800 | same socket to the same server. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 12801 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12802 | src : ip |
| 12803 | This is the source IPv4 address of the client of the session. It is of type |
| 12804 | IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are |
| 12805 | mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the |
| 12806 | TCP-level source address which is used, and not the address of a client |
| 12807 | behind a proxy. However if the "accept-proxy" bind directive is used, it can |
| 12808 | be the address of a client behind another PROXY-protocol compatible component |
| 12809 | for all rule sets except "tcp-request connection" which sees the real address. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 12810 | |
Thierry FOURNIER | d5f624d | 2013-11-26 11:52:33 +0100 | [diff] [blame] | 12811 | Example: |
| 12812 | # add an HTTP header in requests with the originating address' country |
| 12813 | http-request set-header X-Country %[src,map_ip(geoip.lst)] |
| 12814 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12815 | src_bytes_in_rate([<table>]) : integer |
| 12816 | Returns the average bytes rate from the incoming connection's source address |
| 12817 | in the current proxy's stick-table or in the designated stick-table, measured |
| 12818 | in amount of bytes over the period configured in the table. If the address is |
Willy Tarreau | 4d4149c | 2013-07-23 19:33:46 +0200 | [diff] [blame] | 12819 | not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate. |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12820 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12821 | src_bytes_out_rate([<table>]) : integer |
| 12822 | Returns the average bytes rate to the incoming connection's source address in |
| 12823 | the current proxy's stick-table or in the designated stick-table, measured in |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12824 | amount of bytes over the period configured in the table. If the address is |
Willy Tarreau | 4d4149c | 2013-07-23 19:33:46 +0200 | [diff] [blame] | 12825 | not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate. |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12826 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12827 | src_clr_gpc0([<table>]) : integer |
| 12828 | Clears the first General Purpose Counter associated to the incoming |
| 12829 | connection's source address in the current proxy's stick-table or in the |
| 12830 | designated stick-table, and returns its previous value. If the address is not |
| 12831 | found, an entry is created and 0 is returned. This is typically used as a |
| 12832 | second ACL in an expression in order to mark a connection when a first ACL |
| 12833 | was verified : |
Willy Tarreau | f73cd11 | 2011-08-13 01:45:16 +0200 | [diff] [blame] | 12834 | |
| 12835 | # block if 5 consecutive requests continue to come faster than 10 sess |
| 12836 | # per second, and reset the counter as soon as the traffic slows down. |
| 12837 | acl abuse src_http_req_rate gt 10 |
| 12838 | acl kill src_inc_gpc0 gt 5 |
Willy Tarreau | 869948b | 2013-01-04 14:14:57 +0100 | [diff] [blame] | 12839 | acl save src_clr_gpc0 ge 0 |
Willy Tarreau | f73cd11 | 2011-08-13 01:45:16 +0200 | [diff] [blame] | 12840 | tcp-request connection accept if !abuse save |
| 12841 | tcp-request connection reject if abuse kill |
| 12842 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12843 | src_conn_cnt([<table>]) : integer |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12844 | Returns the cumulated number of connections initiated from the current |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12845 | incoming connection's source address in the current proxy's stick-table or in |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12846 | the designated stick-table. If the address is not found, zero is returned. |
Willy Tarreau | 4d4149c | 2013-07-23 19:33:46 +0200 | [diff] [blame] | 12847 | See also sc/sc0/sc1/sc2_conn_cnt. |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12848 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12849 | src_conn_cur([<table>]) : integer |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12850 | Returns the current amount of concurrent connections initiated from the |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12851 | current incoming connection's source address in the current proxy's |
| 12852 | stick-table or in the designated stick-table. If the address is not found, |
Willy Tarreau | 4d4149c | 2013-07-23 19:33:46 +0200 | [diff] [blame] | 12853 | zero is returned. See also sc/sc0/sc1/sc2_conn_cur. |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12854 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12855 | src_conn_rate([<table>]) : integer |
| 12856 | Returns the average connection rate from the incoming connection's source |
| 12857 | address in the current proxy's stick-table or in the designated stick-table, |
| 12858 | measured in amount of connections over the period configured in the table. If |
Willy Tarreau | 4d4149c | 2013-07-23 19:33:46 +0200 | [diff] [blame] | 12859 | the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate. |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12860 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12861 | src_get_gpc0([<table>]) : integer |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12862 | Returns the value of the first General Purpose Counter associated to the |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12863 | incoming connection's source address in the current proxy's stick-table or in |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12864 | the designated stick-table. If the address is not found, zero is returned. |
Willy Tarreau | 4d4149c | 2013-07-23 19:33:46 +0200 | [diff] [blame] | 12865 | See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0. |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12866 | |
Thierry FOURNIER | 236657b | 2015-08-19 08:25:14 +0200 | [diff] [blame] | 12867 | src_get_gpt0([<table>]) : integer |
| 12868 | Returns the value of the first General Purpose Tag associated to the |
| 12869 | incoming connection's source address in the current proxy's stick-table or in |
| 12870 | the designated stick-table. If the address is not found, zero is returned. |
| 12871 | See also sc/sc0/sc1/sc2_get_gpt0. |
| 12872 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12873 | src_gpc0_rate([<table>]) : integer |
Willy Tarreau | ba2ffd1 | 2013-05-29 15:54:14 +0200 | [diff] [blame] | 12874 | Returns the average increment rate of the first General Purpose Counter |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12875 | associated to the incoming connection's source address in the current proxy's |
Willy Tarreau | ba2ffd1 | 2013-05-29 15:54:14 +0200 | [diff] [blame] | 12876 | stick-table or in the designated stick-table. It reports the frequency |
| 12877 | which the gpc0 counter was incremented over the configured period. See also |
Willy Tarreau | 4d4149c | 2013-07-23 19:33:46 +0200 | [diff] [blame] | 12878 | sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note |
| 12879 | that the "gpc0_rate" counter must be stored in the stick-table for a value to |
| 12880 | be returned, as "gpc0" only holds the event count. |
Willy Tarreau | ba2ffd1 | 2013-05-29 15:54:14 +0200 | [diff] [blame] | 12881 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12882 | src_http_err_cnt([<table>]) : integer |
| 12883 | Returns the cumulated number of HTTP errors from the incoming connection's |
| 12884 | source address in the current proxy's stick-table or in the designated |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12885 | stick-table. This includes the both request errors and 4xx error responses. |
Willy Tarreau | 4d4149c | 2013-07-23 19:33:46 +0200 | [diff] [blame] | 12886 | See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12887 | returned. |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12888 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12889 | src_http_err_rate([<table>]) : integer |
| 12890 | Returns the average rate of HTTP errors from the incoming connection's source |
| 12891 | address in the current proxy's stick-table or in the designated stick-table, |
| 12892 | measured in amount of errors over the period configured in the table. This |
| 12893 | includes the both request errors and 4xx error responses. If the address is |
Willy Tarreau | 4d4149c | 2013-07-23 19:33:46 +0200 | [diff] [blame] | 12894 | not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate. |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12895 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12896 | src_http_req_cnt([<table>]) : integer |
| 12897 | Returns the cumulated number of HTTP requests from the incoming connection's |
| 12898 | source address in the current proxy's stick-table or in the designated stick- |
| 12899 | table. This includes every started request, valid or not. If the address is |
Willy Tarreau | 4d4149c | 2013-07-23 19:33:46 +0200 | [diff] [blame] | 12900 | not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt. |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12901 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12902 | src_http_req_rate([<table>]) : integer |
| 12903 | Returns the average rate of HTTP requests from the incoming connection's |
| 12904 | source address in the current proxy's stick-table or in the designated stick- |
| 12905 | table, measured in amount of requests over the period configured in the |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12906 | table. This includes every started request, valid or not. If the address is |
Willy Tarreau | 4d4149c | 2013-07-23 19:33:46 +0200 | [diff] [blame] | 12907 | not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate. |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12908 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12909 | src_inc_gpc0([<table>]) : integer |
| 12910 | Increments the first General Purpose Counter associated to the incoming |
| 12911 | connection's source address in the current proxy's stick-table or in the |
| 12912 | designated stick-table, and returns its new value. If the address is not |
Willy Tarreau | be4a3ef | 2013-06-17 15:04:07 +0200 | [diff] [blame] | 12913 | found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0. |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12914 | This is typically used as a second ACL in an expression in order to mark a |
| 12915 | connection when a first ACL was verified : |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12916 | |
| 12917 | acl abuse src_http_req_rate gt 10 |
Willy Tarreau | 869948b | 2013-01-04 14:14:57 +0100 | [diff] [blame] | 12918 | acl kill src_inc_gpc0 gt 0 |
Willy Tarreau | e965652 | 2010-08-17 15:40:09 +0200 | [diff] [blame] | 12919 | tcp-request connection reject if abuse kill |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12920 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12921 | src_kbytes_in([<table>]) : integer |
Willy Tarreau | a01b974 | 2014-07-10 15:29:24 +0200 | [diff] [blame] | 12922 | Returns the total amount of data received from the incoming connection's |
| 12923 | source address in the current proxy's stick-table or in the designated |
| 12924 | stick-table, measured in kilobytes. If the address is not found, zero is |
| 12925 | returned. The test is currently performed on 32-bit integers, which limits |
| 12926 | values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in. |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12927 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12928 | src_kbytes_out([<table>]) : integer |
Willy Tarreau | a01b974 | 2014-07-10 15:29:24 +0200 | [diff] [blame] | 12929 | Returns the total amount of data sent to the incoming connection's source |
| 12930 | address in the current proxy's stick-table or in the designated stick-table, |
| 12931 | measured in kilobytes. If the address is not found, zero is returned. The |
| 12932 | test is currently performed on 32-bit integers, which limits values to 4 |
| 12933 | terabytes. See also sc/sc0/sc1/sc2_kbytes_out. |
Willy Tarreau | a975b8f | 2010-06-05 19:13:27 +0200 | [diff] [blame] | 12934 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12935 | src_port : integer |
| 12936 | Returns an integer value corresponding to the TCP source port of the |
| 12937 | connection on the client side, which is the port the client connected from. |
| 12938 | Usage of this function is very limited as modern protocols do not care much |
| 12939 | about source ports nowadays. |
Willy Tarreau | 079ff0a | 2009-03-05 21:34:28 +0100 | [diff] [blame] | 12940 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12941 | src_sess_cnt([<table>]) : integer |
| 12942 | Returns the cumulated number of connections initiated from the incoming |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12943 | connection's source IPv4 address in the current proxy's stick-table or in the |
| 12944 | designated stick-table, that were transformed into sessions, which means that |
| 12945 | they were accepted by "tcp-request" rules. If the address is not found, zero |
Willy Tarreau | 4d4149c | 2013-07-23 19:33:46 +0200 | [diff] [blame] | 12946 | is returned. See also sc/sc0/sc1/sc2_sess_cnt. |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12947 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12948 | src_sess_rate([<table>]) : integer |
| 12949 | Returns the average session rate from the incoming connection's source |
| 12950 | address in the current proxy's stick-table or in the designated stick-table, |
| 12951 | measured in amount of sessions over the period configured in the table. A |
| 12952 | session is a connection that went past the early "tcp-request" rules. If the |
Willy Tarreau | 4d4149c | 2013-07-23 19:33:46 +0200 | [diff] [blame] | 12953 | address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate. |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12954 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12955 | src_updt_conn_cnt([<table>]) : integer |
| 12956 | Creates or updates the entry associated to the incoming connection's source |
| 12957 | address in the current proxy's stick-table or in the designated stick-table. |
| 12958 | This table must be configured to store the "conn_cnt" data type, otherwise |
| 12959 | the match will be ignored. The current count is incremented by one, and the |
| 12960 | expiration timer refreshed. The updated count is returned, so this match |
| 12961 | can't return zero. This was used to reject service abusers based on their |
| 12962 | source address. Note: it is recommended to use the more complete "track-sc*" |
| 12963 | actions in "tcp-request" rules instead. |
Willy Tarreau | a975b8f | 2010-06-05 19:13:27 +0200 | [diff] [blame] | 12964 | |
| 12965 | Example : |
| 12966 | # This frontend limits incoming SSH connections to 3 per 10 second for |
| 12967 | # each source address, and rejects excess connections until a 10 second |
| 12968 | # silence is observed. At most 20 addresses are tracked. |
| 12969 | listen ssh |
| 12970 | bind :22 |
| 12971 | mode tcp |
| 12972 | maxconn 100 |
Willy Tarreau | c9705a1 | 2010-07-27 20:05:50 +0200 | [diff] [blame] | 12973 | stick-table type ip size 20 expire 10s store conn_cnt |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12974 | tcp-request content reject if { src_updt_conn_cnt gt 3 } |
Willy Tarreau | a975b8f | 2010-06-05 19:13:27 +0200 | [diff] [blame] | 12975 | server local 127.0.0.1:22 |
| 12976 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12977 | srv_id : integer |
| 12978 | Returns an integer containing the server's id when processing the response. |
| 12979 | While it's almost only used with ACLs, it may be used for logging or |
| 12980 | debugging. |
Hervé COMMOWICK | daa824e | 2011-08-05 12:09:44 +0200 | [diff] [blame] | 12981 | |
Hervé COMMOWICK | 35ed801 | 2010-12-15 14:04:51 +0100 | [diff] [blame] | 12982 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 12983 | 7.3.4. Fetching samples at Layer 5 |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12984 | ---------------------------------- |
Willy Tarreau | 0b1cd94 | 2010-05-16 22:18:27 +0200 | [diff] [blame] | 12985 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 12986 | The layer 5 usually describes just the session layer which in haproxy is |
| 12987 | closest to the session once all the connection handshakes are finished, but |
| 12988 | when no content is yet made available. The fetch methods described here are |
| 12989 | usable as low as the "tcp-request content" rule sets unless they require some |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 12990 | future information. Those generally include the results of SSL negotiations. |
Willy Tarreau | c735a07 | 2011-03-29 00:57:02 +0200 | [diff] [blame] | 12991 | |
Emeric Brun | 645ae79 | 2014-04-30 14:21:06 +0200 | [diff] [blame] | 12992 | ssl_bc : boolean |
| 12993 | Returns true when the back connection was made via an SSL/TLS transport |
| 12994 | layer and is locally deciphered. This means the outgoing connection was made |
| 12995 | other a server with the "ssl" option. |
| 12996 | |
| 12997 | ssl_bc_alg_keysize : integer |
| 12998 | Returns the symmetric cipher key size supported in bits when the outgoing |
| 12999 | connection was made over an SSL/TLS transport layer. |
| 13000 | |
| 13001 | ssl_bc_cipher : string |
| 13002 | Returns the name of the used cipher when the outgoing connection was made |
| 13003 | over an SSL/TLS transport layer. |
| 13004 | |
| 13005 | ssl_bc_protocol : string |
| 13006 | Returns the name of the used protocol when the outgoing connection was made |
| 13007 | over an SSL/TLS transport layer. |
| 13008 | |
Emeric Brun | b73a9b0 | 2014-04-30 18:49:19 +0200 | [diff] [blame] | 13009 | ssl_bc_unique_id : binary |
Emeric Brun | 645ae79 | 2014-04-30 14:21:06 +0200 | [diff] [blame] | 13010 | When the outgoing connection was made over an SSL/TLS transport layer, |
Emeric Brun | b73a9b0 | 2014-04-30 18:49:19 +0200 | [diff] [blame] | 13011 | returns the TLS unique ID as defined in RFC5929 section 3. The unique id |
| 13012 | can be encoded to base64 using the converter: "ssl_bc_unique_id,base64". |
Emeric Brun | 645ae79 | 2014-04-30 14:21:06 +0200 | [diff] [blame] | 13013 | |
| 13014 | ssl_bc_session_id : binary |
| 13015 | Returns the SSL ID of the back connection when the outgoing connection was |
| 13016 | made over an SSL/TLS transport layer. It is useful to log if we want to know |
| 13017 | if session was reused or not. |
| 13018 | |
| 13019 | ssl_bc_use_keysize : integer |
| 13020 | Returns the symmetric cipher key size used in bits when the outgoing |
| 13021 | connection was made over an SSL/TLS transport layer. |
| 13022 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13023 | ssl_c_ca_err : integer |
| 13024 | When the incoming connection was made over an SSL/TLS transport layer, |
| 13025 | returns the ID of the first error detected during verification of the client |
| 13026 | certificate at depth > 0, or 0 if no error was encountered during this |
| 13027 | verification process. Please refer to your SSL library's documentation to |
| 13028 | find the exhaustive list of error codes. |
Willy Tarreau | c735a07 | 2011-03-29 00:57:02 +0200 | [diff] [blame] | 13029 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13030 | ssl_c_ca_err_depth : integer |
| 13031 | When the incoming connection was made over an SSL/TLS transport layer, |
| 13032 | returns the depth in the CA chain of the first error detected during the |
| 13033 | verification of the client certificate. If no error is encountered, 0 is |
| 13034 | returned. |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 13035 | |
Emeric Brun | 43e7958 | 2014-10-29 19:03:26 +0100 | [diff] [blame] | 13036 | ssl_c_der : binary |
| 13037 | Returns the DER formatted certificate presented by the client when the |
| 13038 | incoming connection was made over an SSL/TLS transport layer. When used for |
| 13039 | an ACL, the value(s) to match against can be passed in hexadecimal form. |
| 13040 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13041 | ssl_c_err : integer |
| 13042 | When the incoming connection was made over an SSL/TLS transport layer, |
| 13043 | returns the ID of the first error detected during verification at depth 0, or |
| 13044 | 0 if no error was encountered during this verification process. Please refer |
| 13045 | to your SSL library's documentation to find the exhaustive list of error |
| 13046 | codes. |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 13047 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13048 | ssl_c_i_dn([<entry>[,<occ>]]) : string |
| 13049 | When the incoming connection was made over an SSL/TLS transport layer, |
| 13050 | returns the full distinguished name of the issuer of the certificate |
| 13051 | presented by the client when no <entry> is specified, or the value of the |
| 13052 | first given entry found from the beginning of the DN. If a positive/negative |
| 13053 | occurrence number is specified as the optional second argument, it returns |
| 13054 | the value of the nth given entry value from the beginning/end of the DN. |
| 13055 | For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and |
| 13056 | "ssl_c_i_dn(CN)" retrieves the common name. |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 13057 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13058 | ssl_c_key_alg : string |
| 13059 | Returns the name of the algorithm used to generate the key of the certificate |
| 13060 | presented by the client when the incoming connection was made over an SSL/TLS |
| 13061 | transport layer. |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 13062 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13063 | ssl_c_notafter : string |
| 13064 | Returns the end date presented by the client as a formatted string |
| 13065 | YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS |
| 13066 | transport layer. |
Emeric Brun | bede3d0 | 2009-06-30 17:54:00 +0200 | [diff] [blame] | 13067 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13068 | ssl_c_notbefore : string |
| 13069 | Returns the start date presented by the client as a formatted string |
| 13070 | YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS |
| 13071 | transport layer. |
Willy Tarreau | b6672b5 | 2011-12-12 17:23:41 +0100 | [diff] [blame] | 13072 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13073 | ssl_c_s_dn([<entry>[,<occ>]]) : string |
| 13074 | When the incoming connection was made over an SSL/TLS transport layer, |
| 13075 | returns the full distinguished name of the subject of the certificate |
| 13076 | presented by the client when no <entry> is specified, or the value of the |
| 13077 | first given entry found from the beginning of the DN. If a positive/negative |
| 13078 | occurrence number is specified as the optional second argument, it returns |
| 13079 | the value of the nth given entry value from the beginning/end of the DN. |
| 13080 | For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and |
| 13081 | "ssl_c_s_dn(CN)" retrieves the common name. |
Willy Tarreau | b6672b5 | 2011-12-12 17:23:41 +0100 | [diff] [blame] | 13082 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13083 | ssl_c_serial : binary |
| 13084 | Returns the serial of the certificate presented by the client when the |
| 13085 | incoming connection was made over an SSL/TLS transport layer. When used for |
| 13086 | an ACL, the value(s) to match against can be passed in hexadecimal form. |
Emeric Brun | 2525b6b | 2012-10-18 15:59:43 +0200 | [diff] [blame] | 13087 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13088 | ssl_c_sha1 : binary |
| 13089 | Returns the SHA-1 fingerprint of the certificate presented by the client when |
| 13090 | the incoming connection was made over an SSL/TLS transport layer. This can be |
| 13091 | used to stick a client to a server, or to pass this information to a server. |
Willy Tarreau | 2d0caa3 | 2014-07-02 19:01:22 +0200 | [diff] [blame] | 13092 | Note that the output is binary, so if you want to pass that signature to the |
| 13093 | server, you need to encode it in hex or base64, such as in the example below: |
| 13094 | |
| 13095 | http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex] |
Emeric Brun | 2525b6b | 2012-10-18 15:59:43 +0200 | [diff] [blame] | 13096 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13097 | ssl_c_sig_alg : string |
| 13098 | Returns the name of the algorithm used to sign the certificate presented by |
| 13099 | the client when the incoming connection was made over an SSL/TLS transport |
| 13100 | layer. |
Emeric Brun | 8785589 | 2012-10-17 17:39:35 +0200 | [diff] [blame] | 13101 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13102 | ssl_c_used : boolean |
| 13103 | Returns true if current SSL session uses a client certificate even if current |
| 13104 | connection uses SSL session resumption. See also "ssl_fc_has_crt". |
Emeric Brun | 7f56e74 | 2012-10-19 18:15:40 +0200 | [diff] [blame] | 13105 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13106 | ssl_c_verify : integer |
| 13107 | Returns the verify result error ID when the incoming connection was made over |
| 13108 | an SSL/TLS transport layer, otherwise zero if no error is encountered. Please |
| 13109 | refer to your SSL library's documentation for an exhaustive list of error |
| 13110 | codes. |
Emeric Brun | ce5ad80 | 2012-10-22 14:11:22 +0200 | [diff] [blame] | 13111 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13112 | ssl_c_version : integer |
| 13113 | Returns the version of the certificate presented by the client when the |
| 13114 | incoming connection was made over an SSL/TLS transport layer. |
Emeric Brun | ce5ad80 | 2012-10-22 14:11:22 +0200 | [diff] [blame] | 13115 | |
Emeric Brun | 43e7958 | 2014-10-29 19:03:26 +0100 | [diff] [blame] | 13116 | ssl_f_der : binary |
| 13117 | Returns the DER formatted certificate presented by the frontend when the |
| 13118 | incoming connection was made over an SSL/TLS transport layer. When used for |
| 13119 | an ACL, the value(s) to match against can be passed in hexadecimal form. |
| 13120 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13121 | ssl_f_i_dn([<entry>[,<occ>]]) : string |
| 13122 | When the incoming connection was made over an SSL/TLS transport layer, |
| 13123 | returns the full distinguished name of the issuer of the certificate |
| 13124 | presented by the frontend when no <entry> is specified, or the value of the |
| 13125 | first given entry found from the beginning of the DN. If a positive/negative |
Emeric Brun | 8785589 | 2012-10-17 17:39:35 +0200 | [diff] [blame] | 13126 | occurrence number is specified as the optional second argument, it returns |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13127 | the value of the nth given entry value from the beginning/end of the DN. |
| 13128 | For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and |
| 13129 | "ssl_f_i_dn(CN)" retrieves the common name. |
Emeric Brun | 8785589 | 2012-10-17 17:39:35 +0200 | [diff] [blame] | 13130 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13131 | ssl_f_key_alg : string |
| 13132 | Returns the name of the algorithm used to generate the key of the certificate |
| 13133 | presented by the frontend when the incoming connection was made over an |
| 13134 | SSL/TLS transport layer. |
Emeric Brun | 7f56e74 | 2012-10-19 18:15:40 +0200 | [diff] [blame] | 13135 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13136 | ssl_f_notafter : string |
| 13137 | Returns the end date presented by the frontend as a formatted string |
| 13138 | YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS |
| 13139 | transport layer. |
Emeric Brun | 2525b6b | 2012-10-18 15:59:43 +0200 | [diff] [blame] | 13140 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13141 | ssl_f_notbefore : string |
| 13142 | Returns the start date presented by the frontend as a formatted string |
| 13143 | YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS |
| 13144 | transport layer. |
Emeric Brun | 8785589 | 2012-10-17 17:39:35 +0200 | [diff] [blame] | 13145 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13146 | ssl_f_s_dn([<entry>[,<occ>]]) : string |
| 13147 | When the incoming connection was made over an SSL/TLS transport layer, |
| 13148 | returns the full distinguished name of the subject of the certificate |
| 13149 | presented by the frontend when no <entry> is specified, or the value of the |
| 13150 | first given entry found from the beginning of the DN. If a positive/negative |
| 13151 | occurrence number is specified as the optional second argument, it returns |
| 13152 | the value of the nth given entry value from the beginning/end of the DN. |
| 13153 | For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and |
| 13154 | "ssl_f_s_dn(CN)" retrieves the common name. |
Emeric Brun | ce5ad80 | 2012-10-22 14:11:22 +0200 | [diff] [blame] | 13155 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13156 | ssl_f_serial : binary |
| 13157 | Returns the serial of the certificate presented by the frontend when the |
| 13158 | incoming connection was made over an SSL/TLS transport layer. When used for |
| 13159 | an ACL, the value(s) to match against can be passed in hexadecimal form. |
Emeric Brun | 8785589 | 2012-10-17 17:39:35 +0200 | [diff] [blame] | 13160 | |
Emeric Brun | 55f4fa8 | 2014-04-30 17:11:25 +0200 | [diff] [blame] | 13161 | ssl_f_sha1 : binary |
| 13162 | Returns the SHA-1 fingerprint of the certificate presented by the frontend |
| 13163 | when the incoming connection was made over an SSL/TLS transport layer. This |
| 13164 | can be used to know which certificate was chosen using SNI. |
| 13165 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13166 | ssl_f_sig_alg : string |
| 13167 | Returns the name of the algorithm used to sign the certificate presented by |
| 13168 | the frontend when the incoming connection was made over an SSL/TLS transport |
| 13169 | layer. |
Emeric Brun | 7f56e74 | 2012-10-19 18:15:40 +0200 | [diff] [blame] | 13170 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13171 | ssl_f_version : integer |
| 13172 | Returns the version of the certificate presented by the frontend when the |
| 13173 | incoming connection was made over an SSL/TLS transport layer. |
| 13174 | |
| 13175 | ssl_fc : boolean |
Emeric Brun | 2525b6b | 2012-10-18 15:59:43 +0200 | [diff] [blame] | 13176 | Returns true when the front connection was made via an SSL/TLS transport |
| 13177 | layer and is locally deciphered. This means it has matched a socket declared |
| 13178 | with a "bind" line having the "ssl" option. |
| 13179 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13180 | Example : |
| 13181 | # This passes "X-Proto: https" to servers when client connects over SSL |
| 13182 | listen http-https |
| 13183 | bind :80 |
| 13184 | bind :443 ssl crt /etc/haproxy.pem |
| 13185 | http-request add-header X-Proto https if { ssl_fc } |
| 13186 | |
| 13187 | ssl_fc_alg_keysize : integer |
| 13188 | Returns the symmetric cipher key size supported in bits when the incoming |
| 13189 | connection was made over an SSL/TLS transport layer. |
| 13190 | |
| 13191 | ssl_fc_alpn : string |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 13192 | This extracts the Application Layer Protocol Negotiation field from an |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13193 | incoming connection made via a TLS transport layer and locally deciphered by |
| 13194 | haproxy. The result is a string containing the protocol name advertised by |
| 13195 | the client. The SSL library must have been built with support for TLS |
| 13196 | extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is |
| 13197 | not advertised unless the "alpn" keyword on the "bind" line specifies a |
| 13198 | protocol list. Also, nothing forces the client to pick a protocol from this |
| 13199 | list, any other one may be requested. The TLS ALPN extension is meant to |
| 13200 | replace the TLS NPN extension. See also "ssl_fc_npn". |
| 13201 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13202 | ssl_fc_cipher : string |
| 13203 | Returns the name of the used cipher when the incoming connection was made |
| 13204 | over an SSL/TLS transport layer. |
Willy Tarreau | ab861d3 | 2013-04-02 02:30:41 +0200 | [diff] [blame] | 13205 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13206 | ssl_fc_has_crt : boolean |
Emeric Brun | 2525b6b | 2012-10-18 15:59:43 +0200 | [diff] [blame] | 13207 | Returns true if a client certificate is present in an incoming connection over |
| 13208 | SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'. |
Emeric Brun | 9143d37 | 2012-12-20 15:44:16 +0100 | [diff] [blame] | 13209 | Note: on SSL session resumption with Session ID or TLS ticket, client |
| 13210 | certificate is not present in the current connection but may be retrieved |
| 13211 | from the cache or the ticket. So prefer "ssl_c_used" if you want to check if |
| 13212 | current SSL session uses a client certificate. |
Emeric Brun | 2525b6b | 2012-10-18 15:59:43 +0200 | [diff] [blame] | 13213 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13214 | ssl_fc_has_sni : boolean |
| 13215 | This checks for the presence of a Server Name Indication TLS extension (SNI) |
Willy Tarreau | f7bc57c | 2012-10-03 00:19:48 +0200 | [diff] [blame] | 13216 | in an incoming connection was made over an SSL/TLS transport layer. Returns |
| 13217 | true when the incoming connection presents a TLS SNI field. This requires |
| 13218 | that the SSL library is build with support for TLS extensions enabled (check |
| 13219 | haproxy -vv). |
Willy Tarreau | 7875d09 | 2012-09-10 08:20:03 +0200 | [diff] [blame] | 13220 | |
Nenad Merdanovic | 26ea822 | 2015-05-18 02:28:57 +0200 | [diff] [blame] | 13221 | ssl_fc_is_resumed: boolean |
| 13222 | Returns true if the SSL/TLS session has been resumed through the use of |
| 13223 | SSL session cache or TLS tickets. |
| 13224 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13225 | ssl_fc_npn : string |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 13226 | This extracts the Next Protocol Negotiation field from an incoming connection |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13227 | made via a TLS transport layer and locally deciphered by haproxy. The result |
| 13228 | is a string containing the protocol name advertised by the client. The SSL |
| 13229 | library must have been built with support for TLS extensions enabled (check |
| 13230 | haproxy -vv). Note that the TLS NPN extension is not advertised unless the |
| 13231 | "npn" keyword on the "bind" line specifies a protocol list. Also, nothing |
| 13232 | forces the client to pick a protocol from this list, any other one may be |
| 13233 | requested. Please note that the TLS NPN extension was replaced with ALPN. |
Willy Tarreau | a33c654 | 2012-10-15 13:19:06 +0200 | [diff] [blame] | 13234 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13235 | ssl_fc_protocol : string |
| 13236 | Returns the name of the used protocol when the incoming connection was made |
| 13237 | over an SSL/TLS transport layer. |
Willy Tarreau | 7875d09 | 2012-09-10 08:20:03 +0200 | [diff] [blame] | 13238 | |
Emeric Brun | b73a9b0 | 2014-04-30 18:49:19 +0200 | [diff] [blame] | 13239 | ssl_fc_unique_id : binary |
David S | c1ad52e | 2014-04-08 18:48:47 -0400 | [diff] [blame] | 13240 | When the incoming connection was made over an SSL/TLS transport layer, |
Emeric Brun | b73a9b0 | 2014-04-30 18:49:19 +0200 | [diff] [blame] | 13241 | returns the TLS unique ID as defined in RFC5929 section 3. The unique id |
| 13242 | can be encoded to base64 using the converter: "ssl_bc_unique_id,base64". |
David S | c1ad52e | 2014-04-08 18:48:47 -0400 | [diff] [blame] | 13243 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13244 | ssl_fc_session_id : binary |
| 13245 | Returns the SSL ID of the front connection when the incoming connection was |
| 13246 | made over an SSL/TLS transport layer. It is useful to stick a given client to |
| 13247 | a server. It is important to note that some browsers refresh their session ID |
| 13248 | every few minutes. |
Willy Tarreau | 7875d09 | 2012-09-10 08:20:03 +0200 | [diff] [blame] | 13249 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13250 | ssl_fc_sni : string |
| 13251 | This extracts the Server Name Indication TLS extension (SNI) field from an |
| 13252 | incoming connection made via an SSL/TLS transport layer and locally |
| 13253 | deciphered by haproxy. The result (when present) typically is a string |
| 13254 | matching the HTTPS host name (253 chars or less). The SSL library must have |
| 13255 | been built with support for TLS extensions enabled (check haproxy -vv). |
| 13256 | |
| 13257 | This fetch is different from "req_ssl_sni" above in that it applies to the |
| 13258 | connection being deciphered by haproxy and not to SSL contents being blindly |
| 13259 | forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This |
Cyril Bonté | 9c1eb1e | 2012-10-09 22:45:34 +0200 | [diff] [blame] | 13260 | requires that the SSL library is build with support for TLS extensions |
| 13261 | enabled (check haproxy -vv). |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 13262 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13263 | ACL derivatives : |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13264 | ssl_fc_sni_end : suffix match |
| 13265 | ssl_fc_sni_reg : regex match |
Emeric Brun | 589fcad | 2012-10-16 14:13:26 +0200 | [diff] [blame] | 13266 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13267 | ssl_fc_use_keysize : integer |
| 13268 | Returns the symmetric cipher key size used in bits when the incoming |
| 13269 | connection was made over an SSL/TLS transport layer. |
Willy Tarreau | b6fb420 | 2008-07-20 11:18:28 +0200 | [diff] [blame] | 13270 | |
Willy Tarreau | b6fb420 | 2008-07-20 11:18:28 +0200 | [diff] [blame] | 13271 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 13272 | 7.3.5. Fetching samples from buffer contents (Layer 6) |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13273 | ------------------------------------------------------ |
Willy Tarreau | b6fb420 | 2008-07-20 11:18:28 +0200 | [diff] [blame] | 13274 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13275 | Fetching samples from buffer contents is a bit different from the previous |
| 13276 | sample fetches above because the sampled data are ephemeral. These data can |
| 13277 | only be used when they're available and will be lost when they're forwarded. |
| 13278 | For this reason, samples fetched from buffer contents during a request cannot |
| 13279 | be used in a response for example. Even while the data are being fetched, they |
| 13280 | can change. Sometimes it is necessary to set some delays or combine multiple |
| 13281 | sample fetch methods to ensure that the expected data are complete and usable, |
| 13282 | for example through TCP request content inspection. Please see the "tcp-request |
| 13283 | content" keyword for more detailed information on the subject. |
Willy Tarreau | 6264477 | 2008-07-16 18:36:06 +0200 | [diff] [blame] | 13284 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13285 | payload(<offset>,<length>) : binary (deprecated) |
| 13286 | This is an alias for "req.payload" when used in the context of a request (eg: |
| 13287 | "stick on", "stick match"), and for "res.payload" when used in the context of |
| 13288 | a response such as in "stick store response". |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 13289 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13290 | payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated) |
| 13291 | This is an alias for "req.payload_lv" when used in the context of a request |
| 13292 | (eg: "stick on", "stick match"), and for "res.payload_lv" when used in the |
| 13293 | context of a response such as in "stick store response". |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 13294 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13295 | req.len : integer |
| 13296 | req_len : integer (deprecated) |
| 13297 | Returns an integer value corresponding to the number of bytes present in the |
| 13298 | request buffer. This is mostly used in ACL. It is important to understand |
| 13299 | that this test does not return false as long as the buffer is changing. This |
| 13300 | means that a check with equality to zero will almost always immediately match |
| 13301 | at the beginning of the session, while a test for more data will wait for |
| 13302 | that data to come in and return false only when haproxy is certain that no |
| 13303 | more data will come in. This test was designed to be used with TCP request |
| 13304 | content inspection. |
Willy Tarreau | a7ad50c | 2012-04-29 15:39:40 +0200 | [diff] [blame] | 13305 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13306 | req.payload(<offset>,<length>) : binary |
| 13307 | This extracts a binary block of <length> bytes and starting at byte <offset> |
Willy Tarreau | 00f0084 | 2013-08-02 11:07:32 +0200 | [diff] [blame] | 13308 | in the request buffer. As a special case, if the <length> argument is zero, |
| 13309 | the the whole buffer from <offset> to the end is extracted. This can be used |
| 13310 | with ACLs in order to check for the presence of some content in a buffer at |
| 13311 | any location. |
Willy Tarreau | a7ad50c | 2012-04-29 15:39:40 +0200 | [diff] [blame] | 13312 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13313 | ACL alternatives : |
| 13314 | payload(<offset>,<length>) : hex binary match |
Willy Tarreau | a7ad50c | 2012-04-29 15:39:40 +0200 | [diff] [blame] | 13315 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13316 | req.payload_lv(<offset1>,<length>[,<offset2>]) : binary |
| 13317 | This extracts a binary block whose size is specified at <offset1> for <length> |
| 13318 | bytes, and which starts at <offset2> if specified or just after the length in |
| 13319 | the request buffer. The <offset2> parameter also supports relative offsets if |
| 13320 | prepended with a '+' or '-' sign. |
Willy Tarreau | a7ad50c | 2012-04-29 15:39:40 +0200 | [diff] [blame] | 13321 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13322 | ACL alternatives : |
| 13323 | payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match |
Willy Tarreau | a7ad50c | 2012-04-29 15:39:40 +0200 | [diff] [blame] | 13324 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13325 | Example : please consult the example from the "stick store-response" keyword. |
Willy Tarreau | a7ad50c | 2012-04-29 15:39:40 +0200 | [diff] [blame] | 13326 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13327 | req.proto_http : boolean |
| 13328 | req_proto_http : boolean (deprecated) |
| 13329 | Returns true when data in the request buffer look like HTTP and correctly |
| 13330 | parses as such. It is the same parser as the common HTTP request parser which |
| 13331 | is used so there should be no surprises. The test does not match until the |
| 13332 | request is complete, failed or timed out. This test may be used to report the |
| 13333 | protocol in TCP logs, but the biggest use is to block TCP request analysis |
| 13334 | until a complete HTTP request is present in the buffer, for example to track |
| 13335 | a header. |
Willy Tarreau | a7ad50c | 2012-04-29 15:39:40 +0200 | [diff] [blame] | 13336 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13337 | Example: |
| 13338 | # track request counts per "base" (concatenation of Host+URL) |
| 13339 | tcp-request inspect-delay 10s |
| 13340 | tcp-request content reject if !HTTP |
Willy Tarreau | be4a3ef | 2013-06-17 15:04:07 +0200 | [diff] [blame] | 13341 | tcp-request content track-sc0 base table req-rate |
Willy Tarreau | a7ad50c | 2012-04-29 15:39:40 +0200 | [diff] [blame] | 13342 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13343 | req.rdp_cookie([<name>]) : string |
| 13344 | rdp_cookie([<name>]) : string (deprecated) |
| 13345 | When the request buffer looks like the RDP protocol, extracts the RDP cookie |
| 13346 | <name>, or any cookie if unspecified. The parser only checks for the first |
| 13347 | cookie, as illustrated in the RDP protocol specification. The cookie name is |
| 13348 | case insensitive. Generally the "MSTS" cookie name will be used, as it can |
| 13349 | contain the user name of the client connecting to the server if properly |
| 13350 | configured on the client. The "MSTSHASH" cookie is often used as well for |
| 13351 | session stickiness to servers. |
Willy Tarreau | 04aa6a9 | 2012-04-06 18:57:55 +0200 | [diff] [blame] | 13352 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13353 | This differs from "balance rdp-cookie" in that any balancing algorithm may be |
| 13354 | used and thus the distribution of clients to backend servers is not linked to |
| 13355 | a hash of the RDP cookie. It is envisaged that using a balancing algorithm |
| 13356 | such as "balance roundrobin" or "balance leastconn" will lead to a more even |
| 13357 | distribution of clients to backend servers than the hash used by "balance |
| 13358 | rdp-cookie". |
Willy Tarreau | 04aa6a9 | 2012-04-06 18:57:55 +0200 | [diff] [blame] | 13359 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13360 | ACL derivatives : |
| 13361 | req_rdp_cookie([<name>]) : exact string match |
Willy Tarreau | 04aa6a9 | 2012-04-06 18:57:55 +0200 | [diff] [blame] | 13362 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13363 | Example : |
| 13364 | listen tse-farm |
| 13365 | bind 0.0.0.0:3389 |
| 13366 | # wait up to 5s for an RDP cookie in the request |
| 13367 | tcp-request inspect-delay 5s |
| 13368 | tcp-request content accept if RDP_COOKIE |
| 13369 | # apply RDP cookie persistence |
| 13370 | persist rdp-cookie |
| 13371 | # Persist based on the mstshash cookie |
| 13372 | # This is only useful makes sense if |
| 13373 | # balance rdp-cookie is not used |
| 13374 | stick-table type string size 204800 |
| 13375 | stick on req.rdp_cookie(mstshash) |
| 13376 | server srv1 1.1.1.1:3389 |
| 13377 | server srv1 1.1.1.2:3389 |
Willy Tarreau | 04aa6a9 | 2012-04-06 18:57:55 +0200 | [diff] [blame] | 13378 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13379 | See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the |
| 13380 | "req_rdp_cookie" ACL. |
Willy Tarreau | 04aa6a9 | 2012-04-06 18:57:55 +0200 | [diff] [blame] | 13381 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13382 | req.rdp_cookie_cnt([name]) : integer |
| 13383 | rdp_cookie_cnt([name]) : integer (deprecated) |
| 13384 | Tries to parse the request buffer as RDP protocol, then returns an integer |
| 13385 | corresponding to the number of RDP cookies found. If an optional cookie name |
| 13386 | is passed, only cookies matching this name are considered. This is mostly |
| 13387 | used in ACL. |
Willy Tarreau | 04aa6a9 | 2012-04-06 18:57:55 +0200 | [diff] [blame] | 13388 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13389 | ACL derivatives : |
| 13390 | req_rdp_cookie_cnt([<name>]) : integer match |
Willy Tarreau | 04aa6a9 | 2012-04-06 18:57:55 +0200 | [diff] [blame] | 13391 | |
Nenad Merdanovic | 5fc7d7e | 2015-07-07 22:00:17 +0200 | [diff] [blame] | 13392 | req.ssl_ec_ext : boolean |
| 13393 | Returns a boolean identifying if client sent the Supported Elliptic Curves |
| 13394 | Extension as defined in RFC4492, section 5.1. within the SSL ClientHello |
Cyril Bonté | 307ee1e | 2015-09-28 23:16:06 +0200 | [diff] [blame] | 13395 | message. This can be used to present ECC compatible clients with EC |
| 13396 | certificate and to use RSA for all others, on the same IP address. Note that |
| 13397 | this only applies to raw contents found in the request buffer and not to |
| 13398 | contents deciphered via an SSL data layer, so this will not work with "bind" |
| 13399 | lines having the "ssl" option. |
Nenad Merdanovic | 5fc7d7e | 2015-07-07 22:00:17 +0200 | [diff] [blame] | 13400 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13401 | req.ssl_hello_type : integer |
| 13402 | req_ssl_hello_type : integer (deprecated) |
| 13403 | Returns an integer value containing the type of the SSL hello message found |
| 13404 | in the request buffer if the buffer contains data that parse as a complete |
| 13405 | SSL (v3 or superior) client hello message. Note that this only applies to raw |
| 13406 | contents found in the request buffer and not to contents deciphered via an |
| 13407 | SSL data layer, so this will not work with "bind" lines having the "ssl" |
| 13408 | option. This is mostly used in ACL to detect presence of an SSL hello message |
| 13409 | that is supposed to contain an SSL session ID usable for stickiness. |
Willy Tarreau | 04aa6a9 | 2012-04-06 18:57:55 +0200 | [diff] [blame] | 13410 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13411 | req.ssl_sni : string |
| 13412 | req_ssl_sni : string (deprecated) |
| 13413 | Returns a string containing the value of the Server Name TLS extension sent |
| 13414 | by a client in a TLS stream passing through the request buffer if the buffer |
| 13415 | contains data that parse as a complete SSL (v3 or superior) client hello |
| 13416 | message. Note that this only applies to raw contents found in the request |
| 13417 | buffer and not to contents deciphered via an SSL data layer, so this will not |
| 13418 | work with "bind" lines having the "ssl" option. SNI normally contains the |
| 13419 | name of the host the client tries to connect to (for recent browsers). SNI is |
| 13420 | useful for allowing or denying access to certain hosts when SSL/TLS is used |
| 13421 | by the client. This test was designed to be used with TCP request content |
| 13422 | inspection. If content switching is needed, it is recommended to first wait |
| 13423 | for a complete client hello (type 1), like in the example below. See also |
| 13424 | "ssl_fc_sni". |
Willy Tarreau | 04aa6a9 | 2012-04-06 18:57:55 +0200 | [diff] [blame] | 13425 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13426 | ACL derivatives : |
| 13427 | req_ssl_sni : exact string match |
Willy Tarreau | 04aa6a9 | 2012-04-06 18:57:55 +0200 | [diff] [blame] | 13428 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13429 | Examples : |
| 13430 | # Wait for a client hello for at most 5 seconds |
| 13431 | tcp-request inspect-delay 5s |
| 13432 | tcp-request content accept if { req_ssl_hello_type 1 } |
| 13433 | use_backend bk_allow if { req_ssl_sni -f allowed_sites } |
| 13434 | default_backend bk_sorry_page |
Willy Tarreau | 04aa6a9 | 2012-04-06 18:57:55 +0200 | [diff] [blame] | 13435 | |
Pradeep Jindal | bb2acf5 | 2015-09-29 10:12:57 +0530 | [diff] [blame] | 13436 | req.ssl_st_ext : integer |
| 13437 | Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077) |
| 13438 | Returns 1 if the client sent SessionTicket TLS Extension |
| 13439 | Returns 2 if the client also sent non-zero length TLS SessionTicket |
| 13440 | Note that this only applies to raw contents found in the request buffer and |
| 13441 | not to contents deciphered via an SSL data layer, so this will not work with |
| 13442 | "bind" lines having the "ssl" option. This can for example be used to detect |
| 13443 | whether the client sent a SessionTicket or not and stick it accordingly, if |
| 13444 | no SessionTicket then stick on SessionID or don't stick as there's no server |
| 13445 | side state is there when SessionTickets are in use. |
| 13446 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13447 | req.ssl_ver : integer |
| 13448 | req_ssl_ver : integer (deprecated) |
| 13449 | Returns an integer value containing the version of the SSL/TLS protocol of a |
| 13450 | stream present in the request buffer. Both SSLv2 hello messages and SSLv3 |
| 13451 | messages are supported. TLSv1 is announced as SSL version 3.1. The value is |
| 13452 | composed of the major version multiplied by 65536, added to the minor |
| 13453 | version. Note that this only applies to raw contents found in the request |
| 13454 | buffer and not to contents deciphered via an SSL data layer, so this will not |
| 13455 | work with "bind" lines having the "ssl" option. The ACL version of the test |
| 13456 | matches against a decimal notation in the form MAJOR.MINOR (eg: 3.1). This |
| 13457 | fetch is mostly used in ACL. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13458 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13459 | ACL derivatives : |
| 13460 | req_ssl_ver : decimal match |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13461 | |
Willy Tarreau | 47e8eba | 2013-09-11 23:28:46 +0200 | [diff] [blame] | 13462 | res.len : integer |
| 13463 | Returns an integer value corresponding to the number of bytes present in the |
| 13464 | response buffer. This is mostly used in ACL. It is important to understand |
| 13465 | that this test does not return false as long as the buffer is changing. This |
| 13466 | means that a check with equality to zero will almost always immediately match |
| 13467 | at the beginning of the session, while a test for more data will wait for |
| 13468 | that data to come in and return false only when haproxy is certain that no |
| 13469 | more data will come in. This test was designed to be used with TCP response |
| 13470 | content inspection. |
| 13471 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13472 | res.payload(<offset>,<length>) : binary |
| 13473 | This extracts a binary block of <length> bytes and starting at byte <offset> |
Willy Tarreau | 00f0084 | 2013-08-02 11:07:32 +0200 | [diff] [blame] | 13474 | in the response buffer. As a special case, if the <length> argument is zero, |
| 13475 | the the whole buffer from <offset> to the end is extracted. This can be used |
| 13476 | with ACLs in order to check for the presence of some content in a buffer at |
| 13477 | any location. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13478 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13479 | res.payload_lv(<offset1>,<length>[,<offset2>]) : binary |
| 13480 | This extracts a binary block whose size is specified at <offset1> for <length> |
| 13481 | bytes, and which starts at <offset2> if specified or just after the length in |
| 13482 | the response buffer. The <offset2> parameter also supports relative offsets |
| 13483 | if prepended with a '+' or '-' sign. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13484 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13485 | Example : please consult the example from the "stick store-response" keyword. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13486 | |
Willy Tarreau | 971f7b6 | 2015-09-29 14:06:59 +0200 | [diff] [blame] | 13487 | res.ssl_hello_type : integer |
| 13488 | rep_ssl_hello_type : integer (deprecated) |
| 13489 | Returns an integer value containing the type of the SSL hello message found |
| 13490 | in the response buffer if the buffer contains data that parses as a complete |
| 13491 | SSL (v3 or superior) hello message. Note that this only applies to raw |
| 13492 | contents found in the response buffer and not to contents deciphered via an |
| 13493 | SSL data layer, so this will not work with "server" lines having the "ssl" |
| 13494 | option. This is mostly used in ACL to detect presence of an SSL hello message |
| 13495 | that is supposed to contain an SSL session ID usable for stickiness. |
| 13496 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13497 | wait_end : boolean |
| 13498 | This fetch either returns true when the inspection period is over, or does |
| 13499 | not fetch. It is only used in ACLs, in conjunction with content analysis to |
| 13500 | avoid returning a wrong verdict early. It may also be used to delay some |
| 13501 | actions, such as a delayed reject for some special addresses. Since it either |
| 13502 | stops the rules evaluation or immediately returns true, it is recommended to |
| 13503 | use this acl as the last one in a rule. Please note that the default ACL |
| 13504 | "WAIT_END" is always usable without prior declaration. This test was designed |
| 13505 | to be used with TCP request content inspection. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13506 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13507 | Examples : |
| 13508 | # delay every incoming request by 2 seconds |
| 13509 | tcp-request inspect-delay 2s |
| 13510 | tcp-request content accept if WAIT_END |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13511 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13512 | # don't immediately tell bad guys they are rejected |
| 13513 | tcp-request inspect-delay 10s |
| 13514 | acl goodguys src 10.0.0.0/24 |
| 13515 | acl badguys src 10.0.1.0/24 |
| 13516 | tcp-request content accept if goodguys |
| 13517 | tcp-request content reject if badguys WAIT_END |
| 13518 | tcp-request content reject |
| 13519 | |
| 13520 | |
Thierry FOURNIER | 060762e | 2014-04-23 13:29:15 +0200 | [diff] [blame] | 13521 | 7.3.6. Fetching HTTP samples (Layer 7) |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13522 | -------------------------------------- |
| 13523 | |
| 13524 | It is possible to fetch samples from HTTP contents, requests and responses. |
| 13525 | This application layer is also called layer 7. It is only possible to fetch the |
| 13526 | data in this section when a full HTTP request or response has been parsed from |
| 13527 | its respective request or response buffer. This is always the case with all |
| 13528 | HTTP specific rules and for sections running with "mode http". When using TCP |
| 13529 | content inspection, it may be necessary to support an inspection delay in order |
| 13530 | to let the request or response come in first. These fetches may require a bit |
| 13531 | more CPU resources than the layer 4 ones, but not much since the request and |
| 13532 | response are indexed. |
| 13533 | |
| 13534 | base : string |
| 13535 | This returns the concatenation of the first Host header and the path part of |
| 13536 | the request, which starts at the first slash and ends before the question |
| 13537 | mark. It can be useful in virtual hosted environments to detect URL abuses as |
| 13538 | well as to improve shared caches efficiency. Using this with a limited size |
| 13539 | stick table also allows one to collect statistics about most commonly |
| 13540 | requested objects by host/path. With ACLs it can allow simple content |
| 13541 | switching rules involving the host and the path at the same time, such as |
| 13542 | "www.example.com/favicon.ico". See also "path" and "uri". |
| 13543 | |
| 13544 | ACL derivatives : |
| 13545 | base : exact string match |
| 13546 | base_beg : prefix match |
| 13547 | base_dir : subdir match |
| 13548 | base_dom : domain match |
| 13549 | base_end : suffix match |
| 13550 | base_len : length match |
| 13551 | base_reg : regex match |
| 13552 | base_sub : substring match |
| 13553 | |
| 13554 | base32 : integer |
| 13555 | This returns a 32-bit hash of the value returned by the "base" fetch method |
| 13556 | above. This is useful to track per-URL activity on high traffic sites without |
| 13557 | having to store all URLs. Instead a shorter hash is stored, saving a lot of |
Willy Tarreau | 23ec4ca | 2014-07-15 20:15:37 +0200 | [diff] [blame] | 13558 | memory. The output type is an unsigned integer. The hash function used is |
| 13559 | SDBM with full avalanche on the output. Technically, base32 is exactly equal |
| 13560 | to "base,sdbm(1)". |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13561 | |
| 13562 | base32+src : binary |
| 13563 | This returns the concatenation of the base32 fetch above and the src fetch |
| 13564 | below. The resulting type is of type binary, with a size of 8 or 20 bytes |
| 13565 | depending on the source address family. This can be used to track per-IP, |
| 13566 | per-URL counters. |
| 13567 | |
William Lallemand | 65ad6e1 | 2014-01-31 15:08:02 +0100 | [diff] [blame] | 13568 | capture.req.hdr(<idx>) : string |
| 13569 | This extracts the content of the header captured by the "capture request |
| 13570 | header", idx is the position of the capture keyword in the configuration. |
| 13571 | The first entry is an index of 0. See also: "capture request header". |
| 13572 | |
| 13573 | capture.req.method : string |
| 13574 | This extracts the METHOD of an HTTP request. It can be used in both request |
| 13575 | and response. Unlike "method", it can be used in both request and response |
| 13576 | because it's allocated. |
| 13577 | |
| 13578 | capture.req.uri : string |
| 13579 | This extracts the request's URI, which starts at the first slash and ends |
| 13580 | before the first space in the request (without the host part). Unlike "path" |
| 13581 | and "url", it can be used in both request and response because it's |
| 13582 | allocated. |
| 13583 | |
Willy Tarreau | 3c1b5ec | 2014-04-24 23:41:57 +0200 | [diff] [blame] | 13584 | capture.req.ver : string |
| 13585 | This extracts the request's HTTP version and returns either "HTTP/1.0" or |
| 13586 | "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and |
| 13587 | logs because it relies on a persistent flag. |
| 13588 | |
William Lallemand | 65ad6e1 | 2014-01-31 15:08:02 +0100 | [diff] [blame] | 13589 | capture.res.hdr(<idx>) : string |
| 13590 | This extracts the content of the header captured by the "capture response |
| 13591 | header", idx is the position of the capture keyword in the configuration. |
| 13592 | The first entry is an index of 0. |
| 13593 | See also: "capture response header" |
| 13594 | |
Willy Tarreau | 3c1b5ec | 2014-04-24 23:41:57 +0200 | [diff] [blame] | 13595 | capture.res.ver : string |
| 13596 | This extracts the response's HTTP version and returns either "HTTP/1.0" or |
| 13597 | "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a |
| 13598 | persistent flag. |
| 13599 | |
Willy Tarreau | a5910cc | 2015-05-02 00:46:08 +0200 | [diff] [blame] | 13600 | req.body : binary |
| 13601 | This returns the HTTP request's available body as a block of data. It |
| 13602 | requires that the request body has been buffered made available using |
| 13603 | "option http-buffer-request". In case of chunked-encoded body, currently only |
| 13604 | the first chunk is analyzed. |
| 13605 | |
Thierry FOURNIER | 9826c77 | 2015-05-20 15:50:54 +0200 | [diff] [blame] | 13606 | req.body_param([<name>) : string |
| 13607 | This fetch assumes that the body of the POST request is url-encoded. The user |
| 13608 | can check if the "content-type" contains the value |
| 13609 | "application/x-www-form-urlencoded". This extracts the first occurrence of the |
| 13610 | parameter <name> in the body, which ends before '&'. The parameter name is |
| 13611 | case-sensitive. If no name is given, any parameter will match, and the first |
| 13612 | one will be returned. The result is a string corresponding to the value of the |
| 13613 | parameter <name> as presented in the request body (no URL decoding is |
| 13614 | performed). Note that the ACL version of this fetch iterates over multiple |
| 13615 | parameters and will iteratively report all parameters values if no name is |
| 13616 | given. |
| 13617 | |
Willy Tarreau | a5910cc | 2015-05-02 00:46:08 +0200 | [diff] [blame] | 13618 | req.body_len : integer |
| 13619 | This returns the length of the HTTP request's available body in bytes. It may |
| 13620 | be lower than the advertised length if the body is larger than the buffer. It |
| 13621 | requires that the request body has been buffered made available using |
| 13622 | "option http-buffer-request". |
| 13623 | |
| 13624 | req.body_size : integer |
| 13625 | This returns the advertised length of the HTTP request's body in bytes. It |
| 13626 | will represent the advertised Content-Length header, or the size of the first |
| 13627 | chunk in case of chunked encoding. In order to parse the chunks, it requires |
| 13628 | that the request body has been buffered made available using |
| 13629 | "option http-buffer-request". |
| 13630 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13631 | req.cook([<name>]) : string |
| 13632 | cook([<name>]) : string (deprecated) |
| 13633 | This extracts the last occurrence of the cookie name <name> on a "Cookie" |
| 13634 | header line from the request, and returns its value as string. If no name is |
| 13635 | specified, the first cookie value is returned. When used with ACLs, all |
| 13636 | matching cookies are evaluated. Spaces around the name and the value are |
| 13637 | ignored as requested by the Cookie header specification (RFC6265). The cookie |
| 13638 | name is case-sensitive. Empty cookies are valid, so an empty cookie may very |
| 13639 | well return an empty value if it is present. Use the "found" match to detect |
| 13640 | presence. Use the res.cook() variant for response cookies sent by the server. |
| 13641 | |
| 13642 | ACL derivatives : |
| 13643 | cook([<name>]) : exact string match |
| 13644 | cook_beg([<name>]) : prefix match |
| 13645 | cook_dir([<name>]) : subdir match |
| 13646 | cook_dom([<name>]) : domain match |
| 13647 | cook_end([<name>]) : suffix match |
| 13648 | cook_len([<name>]) : length match |
| 13649 | cook_reg([<name>]) : regex match |
| 13650 | cook_sub([<name>]) : substring match |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13651 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13652 | req.cook_cnt([<name>]) : integer |
| 13653 | cook_cnt([<name>]) : integer (deprecated) |
| 13654 | Returns an integer value representing the number of occurrences of the cookie |
| 13655 | <name> in the request, or all cookies if <name> is not specified. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13656 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13657 | req.cook_val([<name>]) : integer |
| 13658 | cook_val([<name>]) : integer (deprecated) |
| 13659 | This extracts the last occurrence of the cookie name <name> on a "Cookie" |
| 13660 | header line from the request, and converts its value to an integer which is |
| 13661 | returned. If no name is specified, the first cookie value is returned. When |
| 13662 | used in ACLs, all matching names are iterated over until a value matches. |
Willy Tarreau | 0e69854 | 2011-09-16 08:32:32 +0200 | [diff] [blame] | 13663 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13664 | cookie([<name>]) : string (deprecated) |
| 13665 | This extracts the last occurrence of the cookie name <name> on a "Cookie" |
| 13666 | header line from the request, or a "Set-Cookie" header from the response, and |
| 13667 | returns its value as a string. A typical use is to get multiple clients |
| 13668 | sharing a same profile use the same server. This can be similar to what |
Willy Tarreau | 294d0f0 | 2015-08-10 19:40:12 +0200 | [diff] [blame] | 13669 | "appsession" did with the "request-learn" statement, but with support for |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13670 | multi-peer synchronization and state keeping across restarts. If no name is |
| 13671 | specified, the first cookie value is returned. This fetch should not be used |
| 13672 | anymore and should be replaced by req.cook() or res.cook() instead as it |
| 13673 | ambiguously uses the direction based on the context where it is used. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13674 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13675 | hdr([<name>[,<occ>]]) : string |
| 13676 | This is equivalent to req.hdr() when used on requests, and to res.hdr() when |
| 13677 | used on responses. Please refer to these respective fetches for more details. |
| 13678 | In case of doubt about the fetch direction, please use the explicit ones. |
| 13679 | Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 13680 | unambiguously apply to the request headers. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13681 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13682 | req.fhdr(<name>[,<occ>]) : string |
| 13683 | This extracts the last occurrence of header <name> in an HTTP request. When |
| 13684 | used from an ACL, all occurrences are iterated over until a match is found. |
| 13685 | Optionally, a specific occurrence might be specified as a position number. |
| 13686 | Positive values indicate a position from the first occurrence, with 1 being |
| 13687 | the first one. Negative values indicate positions relative to the last one, |
| 13688 | with -1 being the last one. It differs from req.hdr() in that any commas |
| 13689 | present in the value are returned and are not used as delimiters. This is |
| 13690 | sometimes useful with headers such as User-Agent. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13691 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13692 | req.fhdr_cnt([<name>]) : integer |
| 13693 | Returns an integer value representing the number of occurrences of request |
| 13694 | header field name <name>, or the total number of header fields if <name> is |
| 13695 | not specified. Contrary to its req.hdr_cnt() cousin, this function returns |
| 13696 | the number of full line headers and does not stop on commas. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13697 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13698 | req.hdr([<name>[,<occ>]]) : string |
| 13699 | This extracts the last occurrence of header <name> in an HTTP request. When |
| 13700 | used from an ACL, all occurrences are iterated over until a match is found. |
| 13701 | Optionally, a specific occurrence might be specified as a position number. |
| 13702 | Positive values indicate a position from the first occurrence, with 1 being |
| 13703 | the first one. Negative values indicate positions relative to the last one, |
| 13704 | with -1 being the last one. A typical use is with the X-Forwarded-For header |
| 13705 | once converted to IP, associated with an IP stick-table. The function |
| 13706 | considers any comma as a delimiter for distinct values. If full-line headers |
| 13707 | are desired instead, use req.fhdr(). Please carefully check RFC2616 to know |
| 13708 | how certain headers are supposed to be parsed. Also, some of them are case |
| 13709 | insensitive (eg: Connection). |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13710 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13711 | ACL derivatives : |
| 13712 | hdr([<name>[,<occ>]]) : exact string match |
| 13713 | hdr_beg([<name>[,<occ>]]) : prefix match |
| 13714 | hdr_dir([<name>[,<occ>]]) : subdir match |
| 13715 | hdr_dom([<name>[,<occ>]]) : domain match |
| 13716 | hdr_end([<name>[,<occ>]]) : suffix match |
| 13717 | hdr_len([<name>[,<occ>]]) : length match |
| 13718 | hdr_reg([<name>[,<occ>]]) : regex match |
| 13719 | hdr_sub([<name>[,<occ>]]) : substring match |
| 13720 | |
| 13721 | req.hdr_cnt([<name>]) : integer |
| 13722 | hdr_cnt([<header>]) : integer (deprecated) |
| 13723 | Returns an integer value representing the number of occurrences of request |
| 13724 | header field name <name>, or the total number of header field values if |
| 13725 | <name> is not specified. It is important to remember that one header line may |
| 13726 | count as several headers if it has several values. The function considers any |
| 13727 | comma as a delimiter for distinct values. If full-line headers are desired |
| 13728 | instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to |
| 13729 | detect presence, absence or abuse of a specific header, as well as to block |
| 13730 | request smuggling attacks by rejecting requests which contain more than one |
| 13731 | of certain headers. See "req.hdr" for more information on header matching. |
| 13732 | |
| 13733 | req.hdr_ip([<name>[,<occ>]]) : ip |
| 13734 | hdr_ip([<name>[,<occ>]]) : ip (deprecated) |
| 13735 | This extracts the last occurrence of header <name> in an HTTP request, |
| 13736 | converts it to an IPv4 or IPv6 address and returns this address. When used |
| 13737 | with ACLs, all occurrences are checked, and if <name> is omitted, every value |
| 13738 | of every header is checked. Optionally, a specific occurrence might be |
| 13739 | specified as a position number. Positive values indicate a position from the |
| 13740 | first occurrence, with 1 being the first one. Negative values indicate |
| 13741 | positions relative to the last one, with -1 being the last one. A typical use |
| 13742 | is with the X-Forwarded-For and X-Client-IP headers. |
| 13743 | |
| 13744 | req.hdr_val([<name>[,<occ>]]) : integer |
| 13745 | hdr_val([<name>[,<occ>]]) : integer (deprecated) |
| 13746 | This extracts the last occurrence of header <name> in an HTTP request, and |
| 13747 | converts it to an integer value. When used with ACLs, all occurrences are |
| 13748 | checked, and if <name> is omitted, every value of every header is checked. |
| 13749 | Optionally, a specific occurrence might be specified as a position number. |
| 13750 | Positive values indicate a position from the first occurrence, with 1 being |
| 13751 | the first one. Negative values indicate positions relative to the last one, |
| 13752 | with -1 being the last one. A typical use is with the X-Forwarded-For header. |
| 13753 | |
| 13754 | http_auth(<userlist>) : boolean |
| 13755 | Returns a boolean indicating whether the authentication data received from |
| 13756 | the client match a username & password stored in the specified userlist. This |
| 13757 | fetch function is not really useful outside of ACLs. Currently only http |
| 13758 | basic auth is supported. |
| 13759 | |
Thierry FOURNIER | 9eec0a6 | 2014-01-22 18:38:02 +0100 | [diff] [blame] | 13760 | http_auth_group(<userlist>) : string |
| 13761 | Returns a string corresponding to the user name found in the authentication |
| 13762 | data received from the client if both the user name and password are valid |
| 13763 | according to the specified userlist. The main purpose is to use it in ACLs |
| 13764 | where it is then checked whether the user belongs to any group within a list. |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13765 | This fetch function is not really useful outside of ACLs. Currently only http |
| 13766 | basic auth is supported. |
| 13767 | |
| 13768 | ACL derivatives : |
Thierry FOURNIER | 9eec0a6 | 2014-01-22 18:38:02 +0100 | [diff] [blame] | 13769 | http_auth_group(<userlist>) : group ... |
| 13770 | Returns true when the user extracted from the request and whose password is |
| 13771 | valid according to the specified userlist belongs to at least one of the |
| 13772 | groups. |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13773 | |
| 13774 | http_first_req : boolean |
Willy Tarreau | 7f18e52 | 2010-10-22 20:04:13 +0200 | [diff] [blame] | 13775 | Returns true when the request being processed is the first one of the |
| 13776 | connection. This can be used to add or remove headers that may be missing |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13777 | from some requests when a request is not the first one, or to help grouping |
| 13778 | requests in the logs. |
Willy Tarreau | 7f18e52 | 2010-10-22 20:04:13 +0200 | [diff] [blame] | 13779 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13780 | method : integer + string |
| 13781 | Returns an integer value corresponding to the method in the HTTP request. For |
| 13782 | example, "GET" equals 1 (check sources to establish the matching). Value 9 |
| 13783 | means "other method" and may be converted to a string extracted from the |
| 13784 | stream. This should not be used directly as a sample, this is only meant to |
| 13785 | be used from ACLs, which transparently convert methods from patterns to these |
| 13786 | integer + string values. Some predefined ACL already check for most common |
| 13787 | methods. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 13788 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13789 | ACL derivatives : |
| 13790 | method : case insensitive method match |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 13791 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13792 | Example : |
| 13793 | # only accept GET and HEAD requests |
| 13794 | acl valid_method method GET HEAD |
| 13795 | http-request deny if ! valid_method |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 13796 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13797 | path : string |
| 13798 | This extracts the request's URL path, which starts at the first slash and |
| 13799 | ends before the question mark (without the host part). A typical use is with |
| 13800 | prefetch-capable caches, and with portals which need to aggregate multiple |
| 13801 | information from databases and keep them in caches. Note that with outgoing |
| 13802 | caches, it would be wiser to use "url" instead. With ACLs, it's typically |
| 13803 | used to match exact file names (eg: "/login.php"), or directory parts using |
| 13804 | the derivative forms. See also the "url" and "base" fetch methods. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 13805 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13806 | ACL derivatives : |
| 13807 | path : exact string match |
| 13808 | path_beg : prefix match |
| 13809 | path_dir : subdir match |
| 13810 | path_dom : domain match |
| 13811 | path_end : suffix match |
| 13812 | path_len : length match |
| 13813 | path_reg : regex match |
| 13814 | path_sub : substring match |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 13815 | |
Willy Tarreau | 49ad95c | 2015-01-19 15:06:26 +0100 | [diff] [blame] | 13816 | query : string |
| 13817 | This extracts the request's query string, which starts after the first |
| 13818 | question mark. If no question mark is present, this fetch returns nothing. If |
| 13819 | a question mark is present but nothing follows, it returns an empty string. |
| 13820 | This means it's possible to easily know whether a query string is present |
| 13821 | using the "found" matching method. This fetch is the completemnt of "path" |
| 13822 | which stops before the question mark. |
| 13823 | |
Willy Tarreau | eb27ec7 | 2015-02-20 13:55:29 +0100 | [diff] [blame] | 13824 | req.hdr_names([<delim>]) : string |
| 13825 | This builds a string made from the concatenation of all header names as they |
| 13826 | appear in the request when the rule is evaluated. The default delimiter is |
| 13827 | the comma (',') but it may be overridden as an optional argument <delim>. In |
| 13828 | this case, only the first character of <delim> is considered. |
| 13829 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13830 | req.ver : string |
| 13831 | req_ver : string (deprecated) |
| 13832 | Returns the version string from the HTTP request, for example "1.1". This can |
| 13833 | be useful for logs, but is mostly there for ACL. Some predefined ACL already |
| 13834 | check for versions 1.0 and 1.1. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13835 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13836 | ACL derivatives : |
| 13837 | req_ver : exact string match |
Willy Tarreau | 0e69854 | 2011-09-16 08:32:32 +0200 | [diff] [blame] | 13838 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13839 | res.comp : boolean |
| 13840 | Returns the boolean "true" value if the response has been compressed by |
| 13841 | HAProxy, otherwise returns boolean "false". This may be used to add |
| 13842 | information in the logs. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 13843 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13844 | res.comp_algo : string |
| 13845 | Returns a string containing the name of the algorithm used if the response |
| 13846 | was compressed by HAProxy, for example : "deflate". This may be used to add |
| 13847 | some information in the logs. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13848 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13849 | res.cook([<name>]) : string |
| 13850 | scook([<name>]) : string (deprecated) |
| 13851 | This extracts the last occurrence of the cookie name <name> on a "Set-Cookie" |
| 13852 | header line from the response, and returns its value as string. If no name is |
| 13853 | specified, the first cookie value is returned. |
Willy Tarreau | 0ce3aa0 | 2012-04-25 18:46:33 +0200 | [diff] [blame] | 13854 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13855 | ACL derivatives : |
| 13856 | scook([<name>] : exact string match |
Willy Tarreau | 0ce3aa0 | 2012-04-25 18:46:33 +0200 | [diff] [blame] | 13857 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13858 | res.cook_cnt([<name>]) : integer |
| 13859 | scook_cnt([<name>]) : integer (deprecated) |
| 13860 | Returns an integer value representing the number of occurrences of the cookie |
| 13861 | <name> in the response, or all cookies if <name> is not specified. This is |
| 13862 | mostly useful when combined with ACLs to detect suspicious responses. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13863 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13864 | res.cook_val([<name>]) : integer |
| 13865 | scook_val([<name>]) : integer (deprecated) |
| 13866 | This extracts the last occurrence of the cookie name <name> on a "Set-Cookie" |
| 13867 | header line from the response, and converts its value to an integer which is |
| 13868 | returned. If no name is specified, the first cookie value is returned. |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 13869 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13870 | res.fhdr([<name>[,<occ>]]) : string |
| 13871 | This extracts the last occurrence of header <name> in an HTTP response, or of |
| 13872 | the last header if no <name> is specified. Optionally, a specific occurrence |
| 13873 | might be specified as a position number. Positive values indicate a position |
| 13874 | from the first occurrence, with 1 being the first one. Negative values |
| 13875 | indicate positions relative to the last one, with -1 being the last one. It |
| 13876 | differs from res.hdr() in that any commas present in the value are returned |
| 13877 | and are not used as delimiters. If this is not desired, the res.hdr() fetch |
| 13878 | should be used instead. This is sometimes useful with headers such as Date or |
| 13879 | Expires. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 13880 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13881 | res.fhdr_cnt([<name>]) : integer |
| 13882 | Returns an integer value representing the number of occurrences of response |
| 13883 | header field name <name>, or the total number of header fields if <name> is |
| 13884 | not specified. Contrary to its res.hdr_cnt() cousin, this function returns |
| 13885 | the number of full line headers and does not stop on commas. If this is not |
| 13886 | desired, the res.hdr_cnt() fetch should be used instead. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 13887 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13888 | res.hdr([<name>[,<occ>]]) : string |
| 13889 | shdr([<name>[,<occ>]]) : string (deprecated) |
| 13890 | This extracts the last occurrence of header <name> in an HTTP response, or of |
| 13891 | the last header if no <name> is specified. Optionally, a specific occurrence |
| 13892 | might be specified as a position number. Positive values indicate a position |
| 13893 | from the first occurrence, with 1 being the first one. Negative values |
| 13894 | indicate positions relative to the last one, with -1 being the last one. This |
| 13895 | can be useful to learn some data into a stick-table. The function considers |
| 13896 | any comma as a delimiter for distinct values. If this is not desired, the |
| 13897 | res.fhdr() fetch should be used instead. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 13898 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13899 | ACL derivatives : |
| 13900 | shdr([<name>[,<occ>]]) : exact string match |
| 13901 | shdr_beg([<name>[,<occ>]]) : prefix match |
| 13902 | shdr_dir([<name>[,<occ>]]) : subdir match |
| 13903 | shdr_dom([<name>[,<occ>]]) : domain match |
| 13904 | shdr_end([<name>[,<occ>]]) : suffix match |
| 13905 | shdr_len([<name>[,<occ>]]) : length match |
| 13906 | shdr_reg([<name>[,<occ>]]) : regex match |
| 13907 | shdr_sub([<name>[,<occ>]]) : substring match |
| 13908 | |
| 13909 | res.hdr_cnt([<name>]) : integer |
| 13910 | shdr_cnt([<name>]) : integer (deprecated) |
| 13911 | Returns an integer value representing the number of occurrences of response |
| 13912 | header field name <name>, or the total number of header fields if <name> is |
| 13913 | not specified. The function considers any comma as a delimiter for distinct |
| 13914 | values. If this is not desired, the res.fhdr_cnt() fetch should be used |
| 13915 | instead. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 13916 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13917 | res.hdr_ip([<name>[,<occ>]]) : ip |
| 13918 | shdr_ip([<name>[,<occ>]]) : ip (deprecated) |
| 13919 | This extracts the last occurrence of header <name> in an HTTP response, |
| 13920 | convert it to an IPv4 or IPv6 address and returns this address. Optionally, a |
| 13921 | specific occurrence might be specified as a position number. Positive values |
| 13922 | indicate a position from the first occurrence, with 1 being the first one. |
| 13923 | Negative values indicate positions relative to the last one, with -1 being |
| 13924 | the last one. This can be useful to learn some data into a stick table. |
Willy Tarreau | 6a06a40 | 2007-07-15 20:15:28 +0200 | [diff] [blame] | 13925 | |
Willy Tarreau | eb27ec7 | 2015-02-20 13:55:29 +0100 | [diff] [blame] | 13926 | res.hdr_names([<delim>]) : string |
| 13927 | This builds a string made from the concatenation of all header names as they |
| 13928 | appear in the response when the rule is evaluated. The default delimiter is |
| 13929 | the comma (',') but it may be overridden as an optional argument <delim>. In |
| 13930 | this case, only the first character of <delim> is considered. |
| 13931 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13932 | res.hdr_val([<name>[,<occ>]]) : integer |
| 13933 | shdr_val([<name>[,<occ>]]) : integer (deprecated) |
| 13934 | This extracts the last occurrence of header <name> in an HTTP response, and |
| 13935 | converts it to an integer value. Optionally, a specific occurrence might be |
| 13936 | specified as a position number. Positive values indicate a position from the |
| 13937 | first occurrence, with 1 being the first one. Negative values indicate |
| 13938 | positions relative to the last one, with -1 being the last one. This can be |
| 13939 | useful to learn some data into a stick table. |
Alexandre Cassen | 5eb1a90 | 2007-11-29 15:43:32 +0100 | [diff] [blame] | 13940 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13941 | res.ver : string |
| 13942 | resp_ver : string (deprecated) |
| 13943 | Returns the version string from the HTTP response, for example "1.1". This |
| 13944 | can be useful for logs, but is mostly there for ACL. |
Willy Tarreau | 0e69854 | 2011-09-16 08:32:32 +0200 | [diff] [blame] | 13945 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13946 | ACL derivatives : |
| 13947 | resp_ver : exact string match |
Alexandre Cassen | 5eb1a90 | 2007-11-29 15:43:32 +0100 | [diff] [blame] | 13948 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13949 | set-cookie([<name>]) : string (deprecated) |
| 13950 | This extracts the last occurrence of the cookie name <name> on a "Set-Cookie" |
| 13951 | header line from the response and uses the corresponding value to match. This |
Willy Tarreau | 294d0f0 | 2015-08-10 19:40:12 +0200 | [diff] [blame] | 13952 | can be comparable to what "appsession" did with default options, but with |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13953 | support for multi-peer synchronization and state keeping across restarts. |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 13954 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13955 | This fetch function is deprecated and has been superseded by the "res.cook" |
| 13956 | fetch. This keyword will disappear soon. |
Krzysztof Piotr Oledzki | 6b35ce1 | 2010-02-01 23:35:44 +0100 | [diff] [blame] | 13957 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13958 | status : integer |
| 13959 | Returns an integer containing the HTTP status code in the HTTP response, for |
| 13960 | example, 302. It is mostly used within ACLs and integer ranges, for example, |
| 13961 | to remove any Location header if the response is not a 3xx. |
Willy Tarreau | 25c1ebc | 2012-04-25 16:21:44 +0200 | [diff] [blame] | 13962 | |
Thierry Fournier | 0e00dca | 2016-04-07 15:47:40 +0200 | [diff] [blame] | 13963 | unique-id : string |
| 13964 | Returns the unique-id attached to the request. The directive |
| 13965 | "unique-id-format" must be set. If it is not set, the unique-id sample fetch |
| 13966 | fails. Note that the unique-id is usually used with HTTP requests, however this |
| 13967 | sample fetch can be used with other protocols. Obviously, if it is used with |
| 13968 | other protocols than HTTP, the unique-id-format directive must not contain |
| 13969 | HTTP parts. See: unique-id-format and unique-id-header |
| 13970 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13971 | url : string |
| 13972 | This extracts the request's URL as presented in the request. A typical use is |
| 13973 | with prefetch-capable caches, and with portals which need to aggregate |
| 13974 | multiple information from databases and keep them in caches. With ACLs, using |
| 13975 | "path" is preferred over using "url", because clients may send a full URL as |
| 13976 | is normally done with proxies. The only real use is to match "*" which does |
| 13977 | not match in "path", and for which there is already a predefined ACL. See |
| 13978 | also "path" and "base". |
Willy Tarreau | 25c1ebc | 2012-04-25 16:21:44 +0200 | [diff] [blame] | 13979 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13980 | ACL derivatives : |
| 13981 | url : exact string match |
| 13982 | url_beg : prefix match |
| 13983 | url_dir : subdir match |
| 13984 | url_dom : domain match |
| 13985 | url_end : suffix match |
| 13986 | url_len : length match |
| 13987 | url_reg : regex match |
| 13988 | url_sub : substring match |
Willy Tarreau | 25c1ebc | 2012-04-25 16:21:44 +0200 | [diff] [blame] | 13989 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13990 | url_ip : ip |
| 13991 | This extracts the IP address from the request's URL when the host part is |
| 13992 | presented as an IP address. Its use is very limited. For instance, a |
| 13993 | monitoring system might use this field as an alternative for the source IP in |
| 13994 | order to test what path a given source address would follow, or to force an |
| 13995 | entry in a table for a given source address. With ACLs it can be used to |
| 13996 | restrict access to certain systems through a proxy, for example when combined |
| 13997 | with option "http_proxy". |
Willy Tarreau | 25c1ebc | 2012-04-25 16:21:44 +0200 | [diff] [blame] | 13998 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 13999 | url_port : integer |
| 14000 | This extracts the port part from the request's URL. Note that if the port is |
| 14001 | not specified in the request, port 80 is assumed. With ACLs it can be used to |
| 14002 | restrict access to certain systems through a proxy, for example when combined |
| 14003 | with option "http_proxy". |
Willy Tarreau | 25c1ebc | 2012-04-25 16:21:44 +0200 | [diff] [blame] | 14004 | |
Willy Tarreau | 1ede1da | 2015-05-07 16:06:18 +0200 | [diff] [blame] | 14005 | urlp([<name>[,<delim>]]) : string |
| 14006 | url_param([<name>[,<delim>]]) : string |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 14007 | This extracts the first occurrence of the parameter <name> in the query |
| 14008 | string, which begins after either '?' or <delim>, and which ends before '&', |
Willy Tarreau | 1ede1da | 2015-05-07 16:06:18 +0200 | [diff] [blame] | 14009 | ';' or <delim>. The parameter name is case-sensitive. If no name is given, |
| 14010 | any parameter will match, and the first one will be returned. The result is |
| 14011 | a string corresponding to the value of the parameter <name> as presented in |
| 14012 | the request (no URL decoding is performed). This can be used for session |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 14013 | stickiness based on a client ID, to extract an application cookie passed as a |
| 14014 | URL parameter, or in ACLs to apply some checks. Note that the ACL version of |
Willy Tarreau | 1ede1da | 2015-05-07 16:06:18 +0200 | [diff] [blame] | 14015 | this fetch iterates over multiple parameters and will iteratively report all |
| 14016 | parameters values if no name is given |
Willy Tarreau | 25c1ebc | 2012-04-25 16:21:44 +0200 | [diff] [blame] | 14017 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 14018 | ACL derivatives : |
| 14019 | urlp(<name>[,<delim>]) : exact string match |
| 14020 | urlp_beg(<name>[,<delim>]) : prefix match |
| 14021 | urlp_dir(<name>[,<delim>]) : subdir match |
| 14022 | urlp_dom(<name>[,<delim>]) : domain match |
| 14023 | urlp_end(<name>[,<delim>]) : suffix match |
| 14024 | urlp_len(<name>[,<delim>]) : length match |
| 14025 | urlp_reg(<name>[,<delim>]) : regex match |
| 14026 | urlp_sub(<name>[,<delim>]) : substring match |
Willy Tarreau | 25c1ebc | 2012-04-25 16:21:44 +0200 | [diff] [blame] | 14027 | |
Willy Tarreau | 25c1ebc | 2012-04-25 16:21:44 +0200 | [diff] [blame] | 14028 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 14029 | Example : |
| 14030 | # match http://example.com/foo?PHPSESSIONID=some_id |
| 14031 | stick on urlp(PHPSESSIONID) |
| 14032 | # match http://example.com/foo;JSESSIONID=some_id |
| 14033 | stick on urlp(JSESSIONID,;) |
Willy Tarreau | 25c1ebc | 2012-04-25 16:21:44 +0200 | [diff] [blame] | 14034 | |
Willy Tarreau | 1ede1da | 2015-05-07 16:06:18 +0200 | [diff] [blame] | 14035 | urlp_val([<name>[,<delim>])] : integer |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 14036 | See "urlp" above. This one extracts the URL parameter <name> in the request |
| 14037 | and converts it to an integer value. This can be used for session stickiness |
| 14038 | based on a user ID for example, or with ACLs to match a page number or price. |
Willy Tarreau | a9fddca | 2012-07-31 07:51:48 +0200 | [diff] [blame] | 14039 | |
Willy Tarreau | 198a744 | 2008-01-17 12:05:32 +0100 | [diff] [blame] | 14040 | |
Willy Tarreau | 74ca504 | 2013-06-11 23:12:07 +0200 | [diff] [blame] | 14041 | 7.4. Pre-defined ACLs |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14042 | --------------------- |
Willy Tarreau | ced2701 | 2008-01-17 20:35:34 +0100 | [diff] [blame] | 14043 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14044 | Some predefined ACLs are hard-coded so that they do not have to be declared in |
| 14045 | every frontend which needs them. They all have their names in upper case in |
Patrick Mézard | 2382ad6 | 2010-05-09 10:43:32 +0200 | [diff] [blame] | 14046 | order to avoid confusion. Their equivalence is provided below. |
Willy Tarreau | ced2701 | 2008-01-17 20:35:34 +0100 | [diff] [blame] | 14047 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14048 | ACL name Equivalent to Usage |
| 14049 | ---------------+-----------------------------+--------------------------------- |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14050 | FALSE always_false never match |
Willy Tarreau | 2492d5b | 2009-07-11 00:06:00 +0200 | [diff] [blame] | 14051 | HTTP req_proto_http match if protocol is valid HTTP |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14052 | HTTP_1.0 req_ver 1.0 match HTTP version 1.0 |
| 14053 | HTTP_1.1 req_ver 1.1 match HTTP version 1.1 |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 14054 | HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length |
| 14055 | HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme |
| 14056 | HTTP_URL_SLASH url_beg / match URL beginning with "/" |
| 14057 | HTTP_URL_STAR url * match URL equal to "*" |
| 14058 | LOCALHOST src 127.0.0.1/8 match connection from local host |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14059 | METH_CONNECT method CONNECT match HTTP CONNECT method |
Daniel Schneller | 9ff96c7 | 2016-04-11 17:45:29 +0200 | [diff] [blame] | 14060 | METH_DELETE method DELETE match HTTP DELETE method |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14061 | METH_GET method GET HEAD match HTTP GET or HEAD method |
| 14062 | METH_HEAD method HEAD match HTTP HEAD method |
| 14063 | METH_OPTIONS method OPTIONS match HTTP OPTIONS method |
| 14064 | METH_POST method POST match HTTP POST method |
Daniel Schneller | 9ff96c7 | 2016-04-11 17:45:29 +0200 | [diff] [blame] | 14065 | METH_PUT method PUT match HTTP PUT method |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14066 | METH_TRACE method TRACE match HTTP TRACE method |
Emeric Brun | bede3d0 | 2009-06-30 17:54:00 +0200 | [diff] [blame] | 14067 | RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14068 | REQ_CONTENT req_len gt 0 match data in the request buffer |
Willy Tarreau | d63335a | 2010-02-26 12:56:52 +0100 | [diff] [blame] | 14069 | TRUE always_true always match |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14070 | WAIT_END wait_end wait for end of content analysis |
| 14071 | ---------------+-----------------------------+--------------------------------- |
Willy Tarreau | ced2701 | 2008-01-17 20:35:34 +0100 | [diff] [blame] | 14072 | |
Willy Tarreau | b937b7e | 2010-01-12 15:27:54 +0100 | [diff] [blame] | 14073 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14074 | 8. Logging |
| 14075 | ---------- |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 14076 | |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14077 | One of HAProxy's strong points certainly lies is its precise logs. It probably |
| 14078 | provides the finest level of information available for such a product, which is |
| 14079 | very important for troubleshooting complex environments. Standard information |
| 14080 | provided in logs include client ports, TCP/HTTP state timers, precise session |
| 14081 | state at termination and precise termination cause, information about decisions |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 14082 | to direct traffic to a server, and of course the ability to capture arbitrary |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14083 | headers. |
| 14084 | |
| 14085 | In order to improve administrators reactivity, it offers a great transparency |
| 14086 | about encountered problems, both internal and external, and it is possible to |
| 14087 | send logs to different sources at the same time with different level filters : |
| 14088 | |
| 14089 | - global process-level logs (system errors, start/stop, etc..) |
| 14090 | - per-instance system and internal errors (lack of resource, bugs, ...) |
| 14091 | - per-instance external troubles (servers up/down, max connections) |
| 14092 | - per-instance activity (client connections), either at the establishment or |
| 14093 | at the termination. |
Jim Freeman | 9e8714b | 2015-05-26 09:16:34 -0600 | [diff] [blame] | 14094 | - per-request control of log-level, eg: |
| 14095 | http-request set-log-level silent if sensitive_request |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14096 | |
| 14097 | The ability to distribute different levels of logs to different log servers |
| 14098 | allow several production teams to interact and to fix their problems as soon |
| 14099 | as possible. For example, the system team might monitor system-wide errors, |
| 14100 | while the application team might be monitoring the up/down for their servers in |
| 14101 | real time, and the security team might analyze the activity logs with one hour |
| 14102 | delay. |
| 14103 | |
| 14104 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14105 | 8.1. Log levels |
| 14106 | --------------- |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14107 | |
Simon Horman | df791f5 | 2011-05-29 15:01:10 +0900 | [diff] [blame] | 14108 | TCP and HTTP connections can be logged with information such as the date, time, |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14109 | source IP address, destination address, connection duration, response times, |
Simon Horman | df791f5 | 2011-05-29 15:01:10 +0900 | [diff] [blame] | 14110 | HTTP request, HTTP return code, number of bytes transmitted, conditions |
| 14111 | in which the session ended, and even exchanged cookies values. For example |
| 14112 | track a particular user's problems. All messages may be sent to up to two |
| 14113 | syslog servers. Check the "log" keyword in section 4.2 for more information |
| 14114 | about log facilities. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14115 | |
| 14116 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14117 | 8.2. Log formats |
| 14118 | ---------------- |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14119 | |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 14120 | HAProxy supports 5 log formats. Several fields are common between these formats |
Simon Horman | df791f5 | 2011-05-29 15:01:10 +0900 | [diff] [blame] | 14121 | and will be detailed in the following sections. A few of them may vary |
| 14122 | slightly with the configuration, due to indicators specific to certain |
| 14123 | options. The supported formats are as follows : |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14124 | |
| 14125 | - the default format, which is very basic and very rarely used. It only |
| 14126 | provides very basic information about the incoming connection at the moment |
| 14127 | it is accepted : source IP:port, destination IP:port, and frontend-name. |
| 14128 | This mode will eventually disappear so it will not be described to great |
| 14129 | extents. |
| 14130 | |
| 14131 | - the TCP format, which is more advanced. This format is enabled when "option |
| 14132 | tcplog" is set on the frontend. HAProxy will then usually wait for the |
| 14133 | connection to terminate before logging. This format provides much richer |
| 14134 | information, such as timers, connection counts, queue size, etc... This |
| 14135 | format is recommended for pure TCP proxies. |
| 14136 | |
| 14137 | - the HTTP format, which is the most advanced for HTTP proxying. This format |
| 14138 | is enabled when "option httplog" is set on the frontend. It provides the |
| 14139 | same information as the TCP format with some HTTP-specific fields such as |
| 14140 | the request, the status code, and captures of headers and cookies. This |
| 14141 | format is recommended for HTTP proxies. |
| 14142 | |
Emeric Brun | 3a058f3 | 2009-06-30 18:26:00 +0200 | [diff] [blame] | 14143 | - the CLF HTTP format, which is equivalent to the HTTP format, but with the |
| 14144 | fields arranged in the same order as the CLF format. In this mode, all |
| 14145 | timers, captures, flags, etc... appear one per field after the end of the |
| 14146 | common fields, in the same order they appear in the standard HTTP format. |
| 14147 | |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 14148 | - the custom log format, allows you to make your own log line. |
| 14149 | |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14150 | Next sections will go deeper into details for each of these formats. Format |
| 14151 | specification will be performed on a "field" basis. Unless stated otherwise, a |
| 14152 | field is a portion of text delimited by any number of spaces. Since syslog |
| 14153 | servers are susceptible of inserting fields at the beginning of a line, it is |
| 14154 | always assumed that the first field is the one containing the process name and |
| 14155 | identifier. |
| 14156 | |
| 14157 | Note : Since log lines may be quite long, the log examples in sections below |
| 14158 | might be broken into multiple lines. The example log lines will be |
| 14159 | prefixed with 3 closing angle brackets ('>>>') and each time a log is |
| 14160 | broken into multiple lines, each non-final line will end with a |
| 14161 | backslash ('\') and the next line will start indented by two characters. |
| 14162 | |
| 14163 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14164 | 8.2.1. Default log format |
| 14165 | ------------------------- |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14166 | |
| 14167 | This format is used when no specific option is set. The log is emitted as soon |
| 14168 | as the connection is accepted. One should note that this currently is the only |
| 14169 | format which logs the request's destination IP and ports. |
| 14170 | |
| 14171 | Example : |
| 14172 | listen www |
| 14173 | mode http |
| 14174 | log global |
| 14175 | server srv1 127.0.0.1:8000 |
| 14176 | |
| 14177 | >>> Feb 6 12:12:09 localhost \ |
| 14178 | haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \ |
| 14179 | (www/HTTP) |
| 14180 | |
| 14181 | Field Format Extract from the example above |
| 14182 | 1 process_name '[' pid ']:' haproxy[14385]: |
| 14183 | 2 'Connect from' Connect from |
| 14184 | 3 source_ip ':' source_port 10.0.1.2:33312 |
| 14185 | 4 'to' to |
| 14186 | 5 destination_ip ':' destination_port 10.0.3.31:8012 |
| 14187 | 6 '(' frontend_name '/' mode ')' (www/HTTP) |
| 14188 | |
| 14189 | Detailed fields description : |
| 14190 | - "source_ip" is the IP address of the client which initiated the connection. |
| 14191 | - "source_port" is the TCP port of the client which initiated the connection. |
| 14192 | - "destination_ip" is the IP address the client connected to. |
| 14193 | - "destination_port" is the TCP port the client connected to. |
| 14194 | - "frontend_name" is the name of the frontend (or listener) which received |
| 14195 | and processed the connection. |
| 14196 | - "mode is the mode the frontend is operating (TCP or HTTP). |
| 14197 | |
Willy Tarreau | ceb24bc | 2010-11-09 12:46:41 +0100 | [diff] [blame] | 14198 | In case of a UNIX socket, the source and destination addresses are marked as |
| 14199 | "unix:" and the ports reflect the internal ID of the socket which accepted the |
| 14200 | connection (the same ID as reported in the stats). |
| 14201 | |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14202 | It is advised not to use this deprecated format for newer installations as it |
| 14203 | will eventually disappear. |
| 14204 | |
| 14205 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14206 | 8.2.2. TCP log format |
| 14207 | --------------------- |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14208 | |
| 14209 | The TCP format is used when "option tcplog" is specified in the frontend, and |
| 14210 | is the recommended format for pure TCP proxies. It provides a lot of precious |
| 14211 | information for troubleshooting. Since this format includes timers and byte |
| 14212 | counts, the log is normally emitted at the end of the session. It can be |
| 14213 | emitted earlier if "option logasap" is specified, which makes sense in most |
| 14214 | environments with long sessions such as remote terminals. Sessions which match |
| 14215 | the "monitor" rules are never logged. It is also possible not to emit logs for |
| 14216 | sessions for which no data were exchanged between the client and the server, by |
Willy Tarreau | c9bd0cc | 2009-05-10 11:57:02 +0200 | [diff] [blame] | 14217 | specifying "option dontlognull" in the frontend. Successful connections will |
| 14218 | not be logged if "option dontlog-normal" is specified in the frontend. A few |
| 14219 | fields may slightly vary depending on some configuration options, those are |
| 14220 | marked with a star ('*') after the field name below. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14221 | |
| 14222 | Example : |
| 14223 | frontend fnt |
| 14224 | mode tcp |
| 14225 | option tcplog |
| 14226 | log global |
| 14227 | default_backend bck |
| 14228 | |
| 14229 | backend bck |
| 14230 | server srv1 127.0.0.1:8000 |
| 14231 | |
| 14232 | >>> Feb 6 12:12:56 localhost \ |
| 14233 | haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \ |
| 14234 | bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0 |
| 14235 | |
| 14236 | Field Format Extract from the example above |
| 14237 | 1 process_name '[' pid ']:' haproxy[14387]: |
| 14238 | 2 client_ip ':' client_port 10.0.1.2:33313 |
| 14239 | 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443] |
| 14240 | 4 frontend_name fnt |
| 14241 | 5 backend_name '/' server_name bck/srv1 |
| 14242 | 6 Tw '/' Tc '/' Tt* 0/0/5007 |
| 14243 | 7 bytes_read* 212 |
| 14244 | 8 termination_state -- |
| 14245 | 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3 |
| 14246 | 10 srv_queue '/' backend_queue 0/0 |
| 14247 | |
| 14248 | Detailed fields description : |
| 14249 | - "client_ip" is the IP address of the client which initiated the TCP |
Willy Tarreau | ceb24bc | 2010-11-09 12:46:41 +0100 | [diff] [blame] | 14250 | connection to haproxy. If the connection was accepted on a UNIX socket |
| 14251 | instead, the IP address would be replaced with the word "unix". Note that |
| 14252 | when the connection is accepted on a socket configured with "accept-proxy" |
| 14253 | and the PROXY protocol is correctly used, then the logs will reflect the |
| 14254 | forwarded connection's information. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14255 | |
| 14256 | - "client_port" is the TCP port of the client which initiated the connection. |
Willy Tarreau | ceb24bc | 2010-11-09 12:46:41 +0100 | [diff] [blame] | 14257 | If the connection was accepted on a UNIX socket instead, the port would be |
| 14258 | replaced with the ID of the accepting socket, which is also reported in the |
| 14259 | stats interface. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14260 | |
| 14261 | - "accept_date" is the exact date when the connection was received by haproxy |
| 14262 | (which might be very slightly different from the date observed on the |
| 14263 | network if there was some queuing in the system's backlog). This is usually |
| 14264 | the same date which may appear in any upstream firewall's log. |
| 14265 | |
| 14266 | - "frontend_name" is the name of the frontend (or listener) which received |
| 14267 | and processed the connection. |
| 14268 | |
| 14269 | - "backend_name" is the name of the backend (or listener) which was selected |
| 14270 | to manage the connection to the server. This will be the same as the |
| 14271 | frontend if no switching rule has been applied, which is common for TCP |
| 14272 | applications. |
| 14273 | |
| 14274 | - "server_name" is the name of the last server to which the connection was |
| 14275 | sent, which might differ from the first one if there were connection errors |
| 14276 | and a redispatch occurred. Note that this server belongs to the backend |
| 14277 | which processed the request. If the connection was aborted before reaching |
| 14278 | a server, "<NOSRV>" is indicated instead of a server name. |
| 14279 | |
| 14280 | - "Tw" is the total time in milliseconds spent waiting in the various queues. |
| 14281 | It can be "-1" if the connection was aborted before reaching the queue. |
| 14282 | See "Timers" below for more details. |
| 14283 | |
| 14284 | - "Tc" is the total time in milliseconds spent waiting for the connection to |
| 14285 | establish to the final server, including retries. It can be "-1" if the |
| 14286 | connection was aborted before a connection could be established. See |
| 14287 | "Timers" below for more details. |
| 14288 | |
| 14289 | - "Tt" is the total time in milliseconds elapsed between the accept and the |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 14290 | last close. It covers all possible processing. There is one exception, if |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14291 | "option logasap" was specified, then the time counting stops at the moment |
| 14292 | the log is emitted. In this case, a '+' sign is prepended before the value, |
| 14293 | indicating that the final one will be larger. See "Timers" below for more |
| 14294 | details. |
| 14295 | |
| 14296 | - "bytes_read" is the total number of bytes transmitted from the server to |
| 14297 | the client when the log is emitted. If "option logasap" is specified, the |
| 14298 | this value will be prefixed with a '+' sign indicating that the final one |
| 14299 | may be larger. Please note that this value is a 64-bit counter, so log |
| 14300 | analysis tools must be able to handle it without overflowing. |
| 14301 | |
| 14302 | - "termination_state" is the condition the session was in when the session |
| 14303 | ended. This indicates the session state, which side caused the end of |
| 14304 | session to happen, and for what reason (timeout, error, ...). The normal |
| 14305 | flags should be "--", indicating the session was closed by either end with |
| 14306 | no data remaining in buffers. See below "Session state at disconnection" |
| 14307 | for more details. |
| 14308 | |
| 14309 | - "actconn" is the total number of concurrent connections on the process when |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 14310 | the session was logged. It is useful to detect when some per-process system |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14311 | limits have been reached. For instance, if actconn is close to 512 when |
| 14312 | multiple connection errors occur, chances are high that the system limits |
| 14313 | the process to use a maximum of 1024 file descriptors and that all of them |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14314 | are used. See section 3 "Global parameters" to find how to tune the system. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14315 | |
| 14316 | - "feconn" is the total number of concurrent connections on the frontend when |
| 14317 | the session was logged. It is useful to estimate the amount of resource |
| 14318 | required to sustain high loads, and to detect when the frontend's "maxconn" |
| 14319 | has been reached. Most often when this value increases by huge jumps, it is |
| 14320 | because there is congestion on the backend servers, but sometimes it can be |
| 14321 | caused by a denial of service attack. |
| 14322 | |
| 14323 | - "beconn" is the total number of concurrent connections handled by the |
| 14324 | backend when the session was logged. It includes the total number of |
| 14325 | concurrent connections active on servers as well as the number of |
| 14326 | connections pending in queues. It is useful to estimate the amount of |
| 14327 | additional servers needed to support high loads for a given application. |
| 14328 | Most often when this value increases by huge jumps, it is because there is |
| 14329 | congestion on the backend servers, but sometimes it can be caused by a |
| 14330 | denial of service attack. |
| 14331 | |
| 14332 | - "srv_conn" is the total number of concurrent connections still active on |
| 14333 | the server when the session was logged. It can never exceed the server's |
| 14334 | configured "maxconn" parameter. If this value is very often close or equal |
| 14335 | to the server's "maxconn", it means that traffic regulation is involved a |
| 14336 | lot, meaning that either the server's maxconn value is too low, or that |
| 14337 | there aren't enough servers to process the load with an optimal response |
| 14338 | time. When only one of the server's "srv_conn" is high, it usually means |
| 14339 | that this server has some trouble causing the connections to take longer to |
| 14340 | be processed than on other servers. |
| 14341 | |
| 14342 | - "retries" is the number of connection retries experienced by this session |
| 14343 | when trying to connect to the server. It must normally be zero, unless a |
| 14344 | server is being stopped at the same moment the connection was attempted. |
| 14345 | Frequent retries generally indicate either a network problem between |
| 14346 | haproxy and the server, or a misconfigured system backlog on the server |
| 14347 | preventing new connections from being queued. This field may optionally be |
| 14348 | prefixed with a '+' sign, indicating that the session has experienced a |
| 14349 | redispatch after the maximal retry count has been reached on the initial |
| 14350 | server. In this case, the server name appearing in the log is the one the |
| 14351 | connection was redispatched to, and not the first one, though both may |
| 14352 | sometimes be the same in case of hashing for instance. So as a general rule |
| 14353 | of thumb, when a '+' is present in front of the retry count, this count |
| 14354 | should not be attributed to the logged server. |
| 14355 | |
| 14356 | - "srv_queue" is the total number of requests which were processed before |
| 14357 | this one in the server queue. It is zero when the request has not gone |
| 14358 | through the server queue. It makes it possible to estimate the approximate |
| 14359 | server's response time by dividing the time spent in queue by the number of |
| 14360 | requests in the queue. It is worth noting that if a session experiences a |
| 14361 | redispatch and passes through two server queues, their positions will be |
| 14362 | cumulated. A request should not pass through both the server queue and the |
| 14363 | backend queue unless a redispatch occurs. |
| 14364 | |
| 14365 | - "backend_queue" is the total number of requests which were processed before |
| 14366 | this one in the backend's global queue. It is zero when the request has not |
| 14367 | gone through the global queue. It makes it possible to estimate the average |
| 14368 | queue length, which easily translates into a number of missing servers when |
| 14369 | divided by a server's "maxconn" parameter. It is worth noting that if a |
| 14370 | session experiences a redispatch, it may pass twice in the backend's queue, |
| 14371 | and then both positions will be cumulated. A request should not pass |
| 14372 | through both the server queue and the backend queue unless a redispatch |
| 14373 | occurs. |
| 14374 | |
| 14375 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14376 | 8.2.3. HTTP log format |
| 14377 | ---------------------- |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14378 | |
| 14379 | The HTTP format is the most complete and the best suited for HTTP proxies. It |
| 14380 | is enabled by when "option httplog" is specified in the frontend. It provides |
| 14381 | the same level of information as the TCP format with additional features which |
| 14382 | are specific to the HTTP protocol. Just like the TCP format, the log is usually |
| 14383 | emitted at the end of the session, unless "option logasap" is specified, which |
| 14384 | generally only makes sense for download sites. A session which matches the |
| 14385 | "monitor" rules will never logged. It is also possible not to log sessions for |
| 14386 | which no data were sent by the client by specifying "option dontlognull" in the |
Willy Tarreau | c9bd0cc | 2009-05-10 11:57:02 +0200 | [diff] [blame] | 14387 | frontend. Successful connections will not be logged if "option dontlog-normal" |
| 14388 | is specified in the frontend. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14389 | |
| 14390 | Most fields are shared with the TCP log, some being different. A few fields may |
| 14391 | slightly vary depending on some configuration options. Those ones are marked |
| 14392 | with a star ('*') after the field name below. |
| 14393 | |
| 14394 | Example : |
| 14395 | frontend http-in |
| 14396 | mode http |
| 14397 | option httplog |
| 14398 | log global |
| 14399 | default_backend bck |
| 14400 | |
| 14401 | backend static |
| 14402 | server srv1 127.0.0.1:8000 |
| 14403 | |
| 14404 | >>> Feb 6 12:14:14 localhost \ |
| 14405 | haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \ |
| 14406 | static/srv1 10/0/30/69/109 200 2750 - - ---- 1/1/1/1/0 0/0 {1wt.eu} \ |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 14407 | {} "GET /index.html HTTP/1.1" |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14408 | |
| 14409 | Field Format Extract from the example above |
| 14410 | 1 process_name '[' pid ']:' haproxy[14389]: |
| 14411 | 2 client_ip ':' client_port 10.0.1.2:33317 |
| 14412 | 3 '[' accept_date ']' [06/Feb/2009:12:14:14.655] |
| 14413 | 4 frontend_name http-in |
| 14414 | 5 backend_name '/' server_name static/srv1 |
| 14415 | 6 Tq '/' Tw '/' Tc '/' Tr '/' Tt* 10/0/30/69/109 |
| 14416 | 7 status_code 200 |
| 14417 | 8 bytes_read* 2750 |
| 14418 | 9 captured_request_cookie - |
| 14419 | 10 captured_response_cookie - |
| 14420 | 11 termination_state ---- |
| 14421 | 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0 |
| 14422 | 13 srv_queue '/' backend_queue 0/0 |
| 14423 | 14 '{' captured_request_headers* '}' {haproxy.1wt.eu} |
| 14424 | 15 '{' captured_response_headers* '}' {} |
| 14425 | 16 '"' http_request '"' "GET /index.html HTTP/1.1" |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 14426 | |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14427 | |
| 14428 | Detailed fields description : |
| 14429 | - "client_ip" is the IP address of the client which initiated the TCP |
Willy Tarreau | ceb24bc | 2010-11-09 12:46:41 +0100 | [diff] [blame] | 14430 | connection to haproxy. If the connection was accepted on a UNIX socket |
| 14431 | instead, the IP address would be replaced with the word "unix". Note that |
| 14432 | when the connection is accepted on a socket configured with "accept-proxy" |
| 14433 | and the PROXY protocol is correctly used, then the logs will reflect the |
| 14434 | forwarded connection's information. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14435 | |
| 14436 | - "client_port" is the TCP port of the client which initiated the connection. |
Willy Tarreau | ceb24bc | 2010-11-09 12:46:41 +0100 | [diff] [blame] | 14437 | If the connection was accepted on a UNIX socket instead, the port would be |
| 14438 | replaced with the ID of the accepting socket, which is also reported in the |
| 14439 | stats interface. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14440 | |
| 14441 | - "accept_date" is the exact date when the TCP connection was received by |
| 14442 | haproxy (which might be very slightly different from the date observed on |
| 14443 | the network if there was some queuing in the system's backlog). This is |
| 14444 | usually the same date which may appear in any upstream firewall's log. This |
| 14445 | does not depend on the fact that the client has sent the request or not. |
| 14446 | |
| 14447 | - "frontend_name" is the name of the frontend (or listener) which received |
| 14448 | and processed the connection. |
| 14449 | |
| 14450 | - "backend_name" is the name of the backend (or listener) which was selected |
| 14451 | to manage the connection to the server. This will be the same as the |
| 14452 | frontend if no switching rule has been applied. |
| 14453 | |
| 14454 | - "server_name" is the name of the last server to which the connection was |
| 14455 | sent, which might differ from the first one if there were connection errors |
| 14456 | and a redispatch occurred. Note that this server belongs to the backend |
| 14457 | which processed the request. If the request was aborted before reaching a |
| 14458 | server, "<NOSRV>" is indicated instead of a server name. If the request was |
| 14459 | intercepted by the stats subsystem, "<STATS>" is indicated instead. |
| 14460 | |
| 14461 | - "Tq" is the total time in milliseconds spent waiting for the client to send |
| 14462 | a full HTTP request, not counting data. It can be "-1" if the connection |
| 14463 | was aborted before a complete request could be received. It should always |
| 14464 | be very small because a request generally fits in one single packet. Large |
| 14465 | times here generally indicate network trouble between the client and |
| 14466 | haproxy. See "Timers" below for more details. |
| 14467 | |
| 14468 | - "Tw" is the total time in milliseconds spent waiting in the various queues. |
| 14469 | It can be "-1" if the connection was aborted before reaching the queue. |
| 14470 | See "Timers" below for more details. |
| 14471 | |
| 14472 | - "Tc" is the total time in milliseconds spent waiting for the connection to |
| 14473 | establish to the final server, including retries. It can be "-1" if the |
| 14474 | request was aborted before a connection could be established. See "Timers" |
| 14475 | below for more details. |
| 14476 | |
| 14477 | - "Tr" is the total time in milliseconds spent waiting for the server to send |
| 14478 | a full HTTP response, not counting data. It can be "-1" if the request was |
| 14479 | aborted before a complete response could be received. It generally matches |
| 14480 | the server's processing time for the request, though it may be altered by |
| 14481 | the amount of data sent by the client to the server. Large times here on |
| 14482 | "GET" requests generally indicate an overloaded server. See "Timers" below |
| 14483 | for more details. |
| 14484 | |
| 14485 | - "Tt" is the total time in milliseconds elapsed between the accept and the |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 14486 | last close. It covers all possible processing. There is one exception, if |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14487 | "option logasap" was specified, then the time counting stops at the moment |
| 14488 | the log is emitted. In this case, a '+' sign is prepended before the value, |
| 14489 | indicating that the final one will be larger. See "Timers" below for more |
| 14490 | details. |
| 14491 | |
| 14492 | - "status_code" is the HTTP status code returned to the client. This status |
| 14493 | is generally set by the server, but it might also be set by haproxy when |
| 14494 | the server cannot be reached or when its response is blocked by haproxy. |
| 14495 | |
| 14496 | - "bytes_read" is the total number of bytes transmitted to the client when |
| 14497 | the log is emitted. This does include HTTP headers. If "option logasap" is |
| 14498 | specified, the this value will be prefixed with a '+' sign indicating that |
| 14499 | the final one may be larger. Please note that this value is a 64-bit |
| 14500 | counter, so log analysis tools must be able to handle it without |
| 14501 | overflowing. |
| 14502 | |
| 14503 | - "captured_request_cookie" is an optional "name=value" entry indicating that |
| 14504 | the client had this cookie in the request. The cookie name and its maximum |
| 14505 | length are defined by the "capture cookie" statement in the frontend |
| 14506 | configuration. The field is a single dash ('-') when the option is not |
| 14507 | set. Only one cookie may be captured, it is generally used to track session |
| 14508 | ID exchanges between a client and a server to detect session crossing |
| 14509 | between clients due to application bugs. For more details, please consult |
| 14510 | the section "Capturing HTTP headers and cookies" below. |
| 14511 | |
| 14512 | - "captured_response_cookie" is an optional "name=value" entry indicating |
| 14513 | that the server has returned a cookie with its response. The cookie name |
| 14514 | and its maximum length are defined by the "capture cookie" statement in the |
| 14515 | frontend configuration. The field is a single dash ('-') when the option is |
| 14516 | not set. Only one cookie may be captured, it is generally used to track |
| 14517 | session ID exchanges between a client and a server to detect session |
| 14518 | crossing between clients due to application bugs. For more details, please |
| 14519 | consult the section "Capturing HTTP headers and cookies" below. |
| 14520 | |
| 14521 | - "termination_state" is the condition the session was in when the session |
| 14522 | ended. This indicates the session state, which side caused the end of |
| 14523 | session to happen, for what reason (timeout, error, ...), just like in TCP |
| 14524 | logs, and information about persistence operations on cookies in the last |
| 14525 | two characters. The normal flags should begin with "--", indicating the |
| 14526 | session was closed by either end with no data remaining in buffers. See |
| 14527 | below "Session state at disconnection" for more details. |
| 14528 | |
| 14529 | - "actconn" is the total number of concurrent connections on the process when |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 14530 | the session was logged. It is useful to detect when some per-process system |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14531 | limits have been reached. For instance, if actconn is close to 512 or 1024 |
| 14532 | when multiple connection errors occur, chances are high that the system |
| 14533 | limits the process to use a maximum of 1024 file descriptors and that all |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14534 | of them are used. See section 3 "Global parameters" to find how to tune the |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14535 | system. |
| 14536 | |
| 14537 | - "feconn" is the total number of concurrent connections on the frontend when |
| 14538 | the session was logged. It is useful to estimate the amount of resource |
| 14539 | required to sustain high loads, and to detect when the frontend's "maxconn" |
| 14540 | has been reached. Most often when this value increases by huge jumps, it is |
| 14541 | because there is congestion on the backend servers, but sometimes it can be |
| 14542 | caused by a denial of service attack. |
| 14543 | |
| 14544 | - "beconn" is the total number of concurrent connections handled by the |
| 14545 | backend when the session was logged. It includes the total number of |
| 14546 | concurrent connections active on servers as well as the number of |
| 14547 | connections pending in queues. It is useful to estimate the amount of |
| 14548 | additional servers needed to support high loads for a given application. |
| 14549 | Most often when this value increases by huge jumps, it is because there is |
| 14550 | congestion on the backend servers, but sometimes it can be caused by a |
| 14551 | denial of service attack. |
| 14552 | |
| 14553 | - "srv_conn" is the total number of concurrent connections still active on |
| 14554 | the server when the session was logged. It can never exceed the server's |
| 14555 | configured "maxconn" parameter. If this value is very often close or equal |
| 14556 | to the server's "maxconn", it means that traffic regulation is involved a |
| 14557 | lot, meaning that either the server's maxconn value is too low, or that |
| 14558 | there aren't enough servers to process the load with an optimal response |
| 14559 | time. When only one of the server's "srv_conn" is high, it usually means |
| 14560 | that this server has some trouble causing the requests to take longer to be |
| 14561 | processed than on other servers. |
| 14562 | |
| 14563 | - "retries" is the number of connection retries experienced by this session |
| 14564 | when trying to connect to the server. It must normally be zero, unless a |
| 14565 | server is being stopped at the same moment the connection was attempted. |
| 14566 | Frequent retries generally indicate either a network problem between |
| 14567 | haproxy and the server, or a misconfigured system backlog on the server |
| 14568 | preventing new connections from being queued. This field may optionally be |
| 14569 | prefixed with a '+' sign, indicating that the session has experienced a |
| 14570 | redispatch after the maximal retry count has been reached on the initial |
| 14571 | server. In this case, the server name appearing in the log is the one the |
| 14572 | connection was redispatched to, and not the first one, though both may |
| 14573 | sometimes be the same in case of hashing for instance. So as a general rule |
| 14574 | of thumb, when a '+' is present in front of the retry count, this count |
| 14575 | should not be attributed to the logged server. |
| 14576 | |
| 14577 | - "srv_queue" is the total number of requests which were processed before |
| 14578 | this one in the server queue. It is zero when the request has not gone |
| 14579 | through the server queue. It makes it possible to estimate the approximate |
| 14580 | server's response time by dividing the time spent in queue by the number of |
| 14581 | requests in the queue. It is worth noting that if a session experiences a |
| 14582 | redispatch and passes through two server queues, their positions will be |
| 14583 | cumulated. A request should not pass through both the server queue and the |
| 14584 | backend queue unless a redispatch occurs. |
| 14585 | |
| 14586 | - "backend_queue" is the total number of requests which were processed before |
| 14587 | this one in the backend's global queue. It is zero when the request has not |
| 14588 | gone through the global queue. It makes it possible to estimate the average |
| 14589 | queue length, which easily translates into a number of missing servers when |
| 14590 | divided by a server's "maxconn" parameter. It is worth noting that if a |
| 14591 | session experiences a redispatch, it may pass twice in the backend's queue, |
| 14592 | and then both positions will be cumulated. A request should not pass |
| 14593 | through both the server queue and the backend queue unless a redispatch |
| 14594 | occurs. |
| 14595 | |
| 14596 | - "captured_request_headers" is a list of headers captured in the request due |
| 14597 | to the presence of the "capture request header" statement in the frontend. |
| 14598 | Multiple headers can be captured, they will be delimited by a vertical bar |
| 14599 | ('|'). When no capture is enabled, the braces do not appear, causing a |
| 14600 | shift of remaining fields. It is important to note that this field may |
| 14601 | contain spaces, and that using it requires a smarter log parser than when |
| 14602 | it's not used. Please consult the section "Capturing HTTP headers and |
| 14603 | cookies" below for more details. |
| 14604 | |
| 14605 | - "captured_response_headers" is a list of headers captured in the response |
| 14606 | due to the presence of the "capture response header" statement in the |
| 14607 | frontend. Multiple headers can be captured, they will be delimited by a |
| 14608 | vertical bar ('|'). When no capture is enabled, the braces do not appear, |
| 14609 | causing a shift of remaining fields. It is important to note that this |
| 14610 | field may contain spaces, and that using it requires a smarter log parser |
| 14611 | than when it's not used. Please consult the section "Capturing HTTP headers |
| 14612 | and cookies" below for more details. |
| 14613 | |
| 14614 | - "http_request" is the complete HTTP request line, including the method, |
| 14615 | request and HTTP version string. Non-printable characters are encoded (see |
| 14616 | below the section "Non-printable characters"). This is always the last |
| 14617 | field, and it is always delimited by quotes and is the only one which can |
| 14618 | contain quotes. If new fields are added to the log format, they will be |
| 14619 | added before this field. This field might be truncated if the request is |
| 14620 | huge and does not fit in the standard syslog buffer (1024 characters). This |
| 14621 | is the reason why this field must always remain the last one. |
| 14622 | |
| 14623 | |
Cyril Bonté | dc4d903 | 2012-04-08 21:57:39 +0200 | [diff] [blame] | 14624 | 8.2.4. Custom log format |
| 14625 | ------------------------ |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 14626 | |
Willy Tarreau | 2beef58 | 2012-12-20 17:22:52 +0100 | [diff] [blame] | 14627 | The directive log-format allows you to customize the logs in http mode and tcp |
William Lallemand | bddd4fd | 2012-02-27 11:23:10 +0100 | [diff] [blame] | 14628 | mode. It takes a string as argument. |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 14629 | |
| 14630 | HAproxy understands some log format variables. % precedes log format variables. |
| 14631 | Variables can take arguments using braces ('{}'), and multiple arguments are |
| 14632 | separated by commas within the braces. Flags may be added or removed by |
| 14633 | prefixing them with a '+' or '-' sign. |
| 14634 | |
| 14635 | Special variable "%o" may be used to propagate its flags to all other |
| 14636 | variables on the same format string. This is particularly handy with quoted |
Dragan Dosen | 835b921 | 2016-02-12 13:23:03 +0100 | [diff] [blame] | 14637 | ("Q") and escaped ("E") string formats. |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 14638 | |
Willy Tarreau | c836845 | 2012-12-21 00:09:23 +0100 | [diff] [blame] | 14639 | If a variable is named between square brackets ('[' .. ']') then it is used |
Willy Tarreau | be722a2 | 2014-06-13 16:31:59 +0200 | [diff] [blame] | 14640 | as a sample expression rule (see section 7.3). This it useful to add some |
Willy Tarreau | c836845 | 2012-12-21 00:09:23 +0100 | [diff] [blame] | 14641 | less common information such as the client's SSL certificate's DN, or to log |
| 14642 | the key that would be used to store an entry into a stick table. |
| 14643 | |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 14644 | Note: spaces must be escaped. A space character is considered as a separator. |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 14645 | In order to emit a verbatim '%', it must be preceded by another '%' resulting |
Willy Tarreau | 06d97f9 | 2013-12-02 17:45:48 +0100 | [diff] [blame] | 14646 | in '%%'. HAProxy will automatically merge consecutive separators. |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 14647 | |
Dragan Dosen | 835b921 | 2016-02-12 13:23:03 +0100 | [diff] [blame] | 14648 | Note: when using the RFC5424 syslog message format, the characters '"', |
| 14649 | '\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see |
| 14650 | https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In |
| 14651 | such cases, the use of the flag "E" should be considered. |
| 14652 | |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 14653 | Flags are : |
| 14654 | * Q: quote a string |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 14655 | * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid) |
Dragan Dosen | 835b921 | 2016-02-12 13:23:03 +0100 | [diff] [blame] | 14656 | * E: escape characters '"', '\' and ']' in a string with '\' as prefix |
| 14657 | (intended purpose is for the RFC5424 structured-data log formats) |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 14658 | |
| 14659 | Example: |
| 14660 | |
| 14661 | log-format %T\ %t\ Some\ Text |
| 14662 | log-format %{+Q}o\ %t\ %s\ %{-Q}r |
| 14663 | |
Dragan Dosen | 835b921 | 2016-02-12 13:23:03 +0100 | [diff] [blame] | 14664 | log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]] |
| 14665 | |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 14666 | At the moment, the default HTTP format is defined this way : |
| 14667 | |
Willy Tarreau | 2beef58 | 2012-12-20 17:22:52 +0100 | [diff] [blame] | 14668 | log-format %ci:%cp\ [%t]\ %ft\ %b/%s\ %Tq/%Tw/%Tc/%Tr/%Tt\ %ST\ %B\ %CC\ \ |
| 14669 | %CS\ %tsc\ %ac/%fc/%bc/%sc/%rc\ %sq/%bq\ %hr\ %hs\ %{+Q}r |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 14670 | |
William Lallemand | bddd4fd | 2012-02-27 11:23:10 +0100 | [diff] [blame] | 14671 | the default CLF format is defined this way : |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 14672 | |
Willy Tarreau | 2beef58 | 2012-12-20 17:22:52 +0100 | [diff] [blame] | 14673 | log-format %{+Q}o\ %{-Q}ci\ -\ -\ [%T]\ %r\ %ST\ %B\ \"\"\ \"\"\ %cp\ \ |
Willy Tarreau | 773d65f | 2012-10-12 14:56:11 +0200 | [diff] [blame] | 14674 | %ms\ %ft\ %b\ %s\ \%Tq\ %Tw\ %Tc\ %Tr\ %Tt\ %tsc\ %ac\ %fc\ \ |
Willy Tarreau | 2beef58 | 2012-12-20 17:22:52 +0100 | [diff] [blame] | 14675 | %bc\ %sc\ %rc\ %sq\ %bq\ %CC\ %CS\ \%hrl\ %hsl |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 14676 | |
William Lallemand | bddd4fd | 2012-02-27 11:23:10 +0100 | [diff] [blame] | 14677 | and the default TCP format is defined this way : |
| 14678 | |
Willy Tarreau | 2beef58 | 2012-12-20 17:22:52 +0100 | [diff] [blame] | 14679 | log-format %ci:%cp\ [%t]\ %ft\ %b/%s\ %Tw/%Tc/%Tt\ %B\ %ts\ \ |
William Lallemand | bddd4fd | 2012-02-27 11:23:10 +0100 | [diff] [blame] | 14680 | %ac/%fc/%bc/%sc/%rc\ %sq/%bq |
| 14681 | |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 14682 | Please refer to the table below for currently defined variables : |
| 14683 | |
William Lallemand | bddd4fd | 2012-02-27 11:23:10 +0100 | [diff] [blame] | 14684 | +---+------+-----------------------------------------------+-------------+ |
Willy Tarreau | ffc3fcd | 2012-10-12 20:17:54 +0200 | [diff] [blame] | 14685 | | R | var | field name (8.2.2 and 8.2.3 for description) | type | |
William Lallemand | bddd4fd | 2012-02-27 11:23:10 +0100 | [diff] [blame] | 14686 | +---+------+-----------------------------------------------+-------------+ |
| 14687 | | | %o | special variable, apply flags on all next var | | |
| 14688 | +---+------+-----------------------------------------------+-------------+ |
Willy Tarreau | 2beef58 | 2012-12-20 17:22:52 +0100 | [diff] [blame] | 14689 | | | %B | bytes_read (from server to client) | numeric | |
| 14690 | | H | %CC | captured_request_cookie | string | |
| 14691 | | H | %CS | captured_response_cookie | string | |
William Lallemand | 5f23240 | 2012-04-05 18:02:55 +0200 | [diff] [blame] | 14692 | | | %H | hostname | string | |
Andrew Hayworth | 0ebc55f | 2015-04-27 21:37:03 +0000 | [diff] [blame] | 14693 | | H | %HM | HTTP method (ex: POST) | string | |
| 14694 | | H | %HP | HTTP request URI without query string (path) | string | |
Andrew Hayworth | e63ac87 | 2015-07-31 16:14:16 +0000 | [diff] [blame] | 14695 | | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string | |
Andrew Hayworth | 0ebc55f | 2015-04-27 21:37:03 +0000 | [diff] [blame] | 14696 | | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string | |
| 14697 | | H | %HV | HTTP version (ex: HTTP/1.0) | string | |
William Lallemand | a73203e | 2012-03-12 12:48:57 +0100 | [diff] [blame] | 14698 | | | %ID | unique-id | string | |
Willy Tarreau | 4bf9963 | 2014-06-13 12:21:40 +0200 | [diff] [blame] | 14699 | | | %ST | status_code | numeric | |
William Lallemand | 5f23240 | 2012-04-05 18:02:55 +0200 | [diff] [blame] | 14700 | | | %T | gmt_date_time | date | |
William Lallemand | bddd4fd | 2012-02-27 11:23:10 +0100 | [diff] [blame] | 14701 | | | %Tc | Tc | numeric | |
Yuxans Yao | 4e25b01 | 2012-10-19 10:36:09 +0800 | [diff] [blame] | 14702 | | | %Tl | local_date_time | date | |
Willy Tarreau | ffc3fcd | 2012-10-12 20:17:54 +0200 | [diff] [blame] | 14703 | | H | %Tq | Tq | numeric | |
| 14704 | | H | %Tr | Tr | numeric | |
William Lallemand | 5f23240 | 2012-04-05 18:02:55 +0200 | [diff] [blame] | 14705 | | | %Ts | timestamp | numeric | |
William Lallemand | bddd4fd | 2012-02-27 11:23:10 +0100 | [diff] [blame] | 14706 | | | %Tt | Tt | numeric | |
| 14707 | | | %Tw | Tw | numeric | |
Willy Tarreau | 2beef58 | 2012-12-20 17:22:52 +0100 | [diff] [blame] | 14708 | | | %U | bytes_uploaded (from client to server) | numeric | |
William Lallemand | bddd4fd | 2012-02-27 11:23:10 +0100 | [diff] [blame] | 14709 | | | %ac | actconn | numeric | |
| 14710 | | | %b | backend_name | string | |
Willy Tarreau | 2beef58 | 2012-12-20 17:22:52 +0100 | [diff] [blame] | 14711 | | | %bc | beconn (backend concurrent connections) | numeric | |
| 14712 | | | %bi | backend_source_ip (connecting address) | IP | |
| 14713 | | | %bp | backend_source_port (connecting address) | numeric | |
William Lallemand | bddd4fd | 2012-02-27 11:23:10 +0100 | [diff] [blame] | 14714 | | | %bq | backend_queue | numeric | |
Willy Tarreau | 2beef58 | 2012-12-20 17:22:52 +0100 | [diff] [blame] | 14715 | | | %ci | client_ip (accepted address) | IP | |
| 14716 | | | %cp | client_port (accepted address) | numeric | |
William Lallemand | bddd4fd | 2012-02-27 11:23:10 +0100 | [diff] [blame] | 14717 | | | %f | frontend_name | string | |
Willy Tarreau | 2beef58 | 2012-12-20 17:22:52 +0100 | [diff] [blame] | 14718 | | | %fc | feconn (frontend concurrent connections) | numeric | |
| 14719 | | | %fi | frontend_ip (accepting address) | IP | |
| 14720 | | | %fp | frontend_port (accepting address) | numeric | |
Willy Tarreau | 773d65f | 2012-10-12 14:56:11 +0200 | [diff] [blame] | 14721 | | | %ft | frontend_name_transport ('~' suffix for SSL) | string | |
Willy Tarreau | 7346acb | 2014-08-28 15:03:15 +0200 | [diff] [blame] | 14722 | | | %lc | frontend_log_counter | numeric | |
Willy Tarreau | d9ed3d2 | 2014-06-13 12:23:06 +0200 | [diff] [blame] | 14723 | | | %hr | captured_request_headers default style | string | |
| 14724 | | | %hrl | captured_request_headers CLF style | string list | |
| 14725 | | | %hs | captured_response_headers default style | string | |
| 14726 | | | %hsl | captured_response_headers CLF style | string list | |
Willy Tarreau | 812c88e | 2015-08-09 10:56:35 +0200 | [diff] [blame] | 14727 | | | %ms | accept date milliseconds (left-padded with 0) | numeric | |
William Lallemand | 5f23240 | 2012-04-05 18:02:55 +0200 | [diff] [blame] | 14728 | | | %pid | PID | numeric | |
Willy Tarreau | ffc3fcd | 2012-10-12 20:17:54 +0200 | [diff] [blame] | 14729 | | H | %r | http_request | string | |
William Lallemand | bddd4fd | 2012-02-27 11:23:10 +0100 | [diff] [blame] | 14730 | | | %rc | retries | numeric | |
Willy Tarreau | 1f0da24 | 2014-01-25 11:01:50 +0100 | [diff] [blame] | 14731 | | | %rt | request_counter (HTTP req or TCP session) | numeric | |
William Lallemand | bddd4fd | 2012-02-27 11:23:10 +0100 | [diff] [blame] | 14732 | | | %s | server_name | string | |
Willy Tarreau | 2beef58 | 2012-12-20 17:22:52 +0100 | [diff] [blame] | 14733 | | | %sc | srv_conn (server concurrent connections) | numeric | |
| 14734 | | | %si | server_IP (target address) | IP | |
| 14735 | | | %sp | server_port (target address) | numeric | |
William Lallemand | bddd4fd | 2012-02-27 11:23:10 +0100 | [diff] [blame] | 14736 | | | %sq | srv_queue | numeric | |
Willy Tarreau | ffc3fcd | 2012-10-12 20:17:54 +0200 | [diff] [blame] | 14737 | | S | %sslc| ssl_ciphers (ex: AES-SHA) | string | |
| 14738 | | S | %sslv| ssl_version (ex: TLSv1) | string | |
Willy Tarreau | 2beef58 | 2012-12-20 17:22:52 +0100 | [diff] [blame] | 14739 | | | %t | date_time (with millisecond resolution) | date | |
William Lallemand | bddd4fd | 2012-02-27 11:23:10 +0100 | [diff] [blame] | 14740 | | | %ts | termination_state | string | |
Willy Tarreau | ffc3fcd | 2012-10-12 20:17:54 +0200 | [diff] [blame] | 14741 | | H | %tsc | termination_state with cookie status | string | |
William Lallemand | bddd4fd | 2012-02-27 11:23:10 +0100 | [diff] [blame] | 14742 | +---+------+-----------------------------------------------+-------------+ |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 14743 | |
Willy Tarreau | ffc3fcd | 2012-10-12 20:17:54 +0200 | [diff] [blame] | 14744 | R = Restrictions : H = mode http only ; S = SSL only |
William Lallemand | 4894040 | 2012-01-30 16:47:22 +0100 | [diff] [blame] | 14745 | |
Willy Tarreau | 5f51e1a | 2012-12-03 18:40:10 +0100 | [diff] [blame] | 14746 | |
| 14747 | 8.2.5. Error log format |
| 14748 | ----------------------- |
| 14749 | |
| 14750 | When an incoming connection fails due to an SSL handshake or an invalid PROXY |
| 14751 | protocol header, haproxy will log the event using a shorter, fixed line format. |
| 14752 | By default, logs are emitted at the LOG_INFO level, unless the option |
| 14753 | "log-separate-errors" is set in the backend, in which case the LOG_ERR level |
| 14754 | will be used. Connections on which no data are exchanged (eg: probes) are not |
| 14755 | logged if the "dontlognull" option is set. |
| 14756 | |
| 14757 | The format looks like this : |
| 14758 | |
| 14759 | >>> Dec 3 18:27:14 localhost \ |
| 14760 | haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \ |
| 14761 | Connection error during SSL handshake |
| 14762 | |
| 14763 | Field Format Extract from the example above |
| 14764 | 1 process_name '[' pid ']:' haproxy[6103]: |
| 14765 | 2 client_ip ':' client_port 127.0.0.1:56059 |
| 14766 | 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380] |
| 14767 | 4 frontend_name "/" bind_name ":" frt/f1: |
| 14768 | 5 message Connection error during SSL handshake |
| 14769 | |
| 14770 | These fields just provide minimal information to help debugging connection |
| 14771 | failures. |
| 14772 | |
| 14773 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14774 | 8.3. Advanced logging options |
| 14775 | ----------------------------- |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14776 | |
| 14777 | Some advanced logging options are often looked for but are not easy to find out |
| 14778 | just by looking at the various options. Here is an entry point for the few |
| 14779 | options which can enable better logging. Please refer to the keywords reference |
| 14780 | for more information about their usage. |
| 14781 | |
| 14782 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14783 | 8.3.1. Disabling logging of external tests |
| 14784 | ------------------------------------------ |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14785 | |
| 14786 | It is quite common to have some monitoring tools perform health checks on |
| 14787 | haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any |
| 14788 | commercial load-balancer, and sometimes it will simply be a more complete |
| 14789 | monitoring system such as Nagios. When the tests are very frequent, users often |
| 14790 | ask how to disable logging for those checks. There are three possibilities : |
| 14791 | |
| 14792 | - if connections come from everywhere and are just TCP probes, it is often |
| 14793 | desired to simply disable logging of connections without data exchange, by |
| 14794 | setting "option dontlognull" in the frontend. It also disables logging of |
| 14795 | port scans, which may or may not be desired. |
| 14796 | |
| 14797 | - if the connection come from a known source network, use "monitor-net" to |
| 14798 | declare this network as monitoring only. Any host in this network will then |
| 14799 | only be able to perform health checks, and their requests will not be |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 14800 | logged. This is generally appropriate to designate a list of equipment |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14801 | such as other load-balancers. |
| 14802 | |
| 14803 | - if the tests are performed on a known URI, use "monitor-uri" to declare |
| 14804 | this URI as dedicated to monitoring. Any host sending this request will |
| 14805 | only get the result of a health-check, and the request will not be logged. |
| 14806 | |
| 14807 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14808 | 8.3.2. Logging before waiting for the session to terminate |
| 14809 | ---------------------------------------------------------- |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14810 | |
| 14811 | The problem with logging at end of connection is that you have no clue about |
| 14812 | what is happening during very long sessions, such as remote terminal sessions |
| 14813 | or large file downloads. This problem can be worked around by specifying |
| 14814 | "option logasap" in the frontend. Haproxy will then log as soon as possible, |
| 14815 | just before data transfer begins. This means that in case of TCP, it will still |
| 14816 | log the connection status to the server, and in case of HTTP, it will log just |
| 14817 | after processing the server headers. In this case, the number of bytes reported |
| 14818 | is the number of header bytes sent to the client. In order to avoid confusion |
| 14819 | with normal logs, the total time field and the number of bytes are prefixed |
| 14820 | with a '+' sign which means that real numbers are certainly larger. |
| 14821 | |
| 14822 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14823 | 8.3.3. Raising log level upon errors |
| 14824 | ------------------------------------ |
Willy Tarreau | c9bd0cc | 2009-05-10 11:57:02 +0200 | [diff] [blame] | 14825 | |
| 14826 | Sometimes it is more convenient to separate normal traffic from errors logs, |
| 14827 | for instance in order to ease error monitoring from log files. When the option |
| 14828 | "log-separate-errors" is used, connections which experience errors, timeouts, |
| 14829 | retries, redispatches or HTTP status codes 5xx will see their syslog level |
| 14830 | raised from "info" to "err". This will help a syslog daemon store the log in |
| 14831 | a separate file. It is very important to keep the errors in the normal traffic |
| 14832 | file too, so that log ordering is not altered. You should also be careful if |
| 14833 | you already have configured your syslog daemon to store all logs higher than |
| 14834 | "notice" in an "admin" file, because the "err" level is higher than "notice". |
| 14835 | |
| 14836 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14837 | 8.3.4. Disabling logging of successful connections |
| 14838 | -------------------------------------------------- |
Willy Tarreau | c9bd0cc | 2009-05-10 11:57:02 +0200 | [diff] [blame] | 14839 | |
| 14840 | Although this may sound strange at first, some large sites have to deal with |
| 14841 | multiple thousands of logs per second and are experiencing difficulties keeping |
| 14842 | them intact for a long time or detecting errors within them. If the option |
| 14843 | "dontlog-normal" is set on the frontend, all normal connections will not be |
| 14844 | logged. In this regard, a normal connection is defined as one without any |
| 14845 | error, timeout, retry nor redispatch. In HTTP, the status code is checked too, |
| 14846 | and a response with a status 5xx is not considered normal and will be logged |
| 14847 | too. Of course, doing is is really discouraged as it will remove most of the |
| 14848 | useful information from the logs. Do this only if you have no other |
| 14849 | alternative. |
| 14850 | |
| 14851 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14852 | 8.4. Timing events |
| 14853 | ------------------ |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14854 | |
| 14855 | Timers provide a great help in troubleshooting network problems. All values are |
| 14856 | reported in milliseconds (ms). These timers should be used in conjunction with |
| 14857 | the session termination flags. In TCP mode with "option tcplog" set on the |
| 14858 | frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP |
| 14859 | mode, 5 control points are reported under the form "Tq/Tw/Tc/Tr/Tt" : |
| 14860 | |
| 14861 | - Tq: total time to get the client request (HTTP mode only). It's the time |
| 14862 | elapsed between the moment the client connection was accepted and the |
| 14863 | moment the proxy received the last HTTP header. The value "-1" indicates |
| 14864 | that the end of headers (empty line) has never been seen. This happens when |
| 14865 | the client closes prematurely or times out. |
| 14866 | |
| 14867 | - Tw: total time spent in the queues waiting for a connection slot. It |
| 14868 | accounts for backend queue as well as the server queues, and depends on the |
| 14869 | queue size, and the time needed for the server to complete previous |
| 14870 | requests. The value "-1" means that the request was killed before reaching |
| 14871 | the queue, which is generally what happens with invalid or denied requests. |
| 14872 | |
| 14873 | - Tc: total time to establish the TCP connection to the server. It's the time |
| 14874 | elapsed between the moment the proxy sent the connection request, and the |
| 14875 | moment it was acknowledged by the server, or between the TCP SYN packet and |
| 14876 | the matching SYN/ACK packet in return. The value "-1" means that the |
| 14877 | connection never established. |
| 14878 | |
| 14879 | - Tr: server response time (HTTP mode only). It's the time elapsed between |
| 14880 | the moment the TCP connection was established to the server and the moment |
| 14881 | the server sent its complete response headers. It purely shows its request |
| 14882 | processing time, without the network overhead due to the data transmission. |
| 14883 | It is worth noting that when the client has data to send to the server, for |
| 14884 | instance during a POST request, the time already runs, and this can distort |
| 14885 | apparent response time. For this reason, it's generally wise not to trust |
| 14886 | too much this field for POST requests initiated from clients behind an |
| 14887 | untrusted network. A value of "-1" here means that the last the response |
| 14888 | header (empty line) was never seen, most likely because the server timeout |
| 14889 | stroke before the server managed to process the request. |
| 14890 | |
| 14891 | - Tt: total session duration time, between the moment the proxy accepted it |
| 14892 | and the moment both ends were closed. The exception is when the "logasap" |
| 14893 | option is specified. In this case, it only equals (Tq+Tw+Tc+Tr), and is |
| 14894 | prefixed with a '+' sign. From this field, we can deduce "Td", the data |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 14895 | transmission time, by subtracting other timers when valid : |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14896 | |
| 14897 | Td = Tt - (Tq + Tw + Tc + Tr) |
| 14898 | |
| 14899 | Timers with "-1" values have to be excluded from this equation. In TCP |
| 14900 | mode, "Tq" and "Tr" have to be excluded too. Note that "Tt" can never be |
| 14901 | negative. |
| 14902 | |
| 14903 | These timers provide precious indications on trouble causes. Since the TCP |
| 14904 | protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure |
| 14905 | that timers close to multiples of 3s are nearly always related to lost packets |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 14906 | due to network problems (wires, negotiation, congestion). Moreover, if "Tt" is |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14907 | close to a timeout value specified in the configuration, it often means that a |
| 14908 | session has been aborted on timeout. |
| 14909 | |
| 14910 | Most common cases : |
| 14911 | |
| 14912 | - If "Tq" is close to 3000, a packet has probably been lost between the |
| 14913 | client and the proxy. This is very rare on local networks but might happen |
| 14914 | when clients are on far remote networks and send large requests. It may |
| 14915 | happen that values larger than usual appear here without any network cause. |
| 14916 | Sometimes, during an attack or just after a resource starvation has ended, |
| 14917 | haproxy may accept thousands of connections in a few milliseconds. The time |
| 14918 | spent accepting these connections will inevitably slightly delay processing |
| 14919 | of other connections, and it can happen that request times in the order of |
| 14920 | a few tens of milliseconds are measured after a few thousands of new |
Lukas Tribus | f01a9cd | 2016-02-03 18:09:37 +0100 | [diff] [blame] | 14921 | connections have been accepted at once. Using one of the keep-alive modes |
Patrick Mezard | 105faca | 2010-06-12 17:02:46 +0200 | [diff] [blame] | 14922 | may display larger request times since "Tq" also measures the time spent |
| 14923 | waiting for additional requests. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14924 | |
| 14925 | - If "Tc" is close to 3000, a packet has probably been lost between the |
| 14926 | server and the proxy during the server connection phase. This value should |
| 14927 | always be very low, such as 1 ms on local networks and less than a few tens |
| 14928 | of ms on remote networks. |
| 14929 | |
Willy Tarreau | 55165fe | 2009-05-10 12:02:55 +0200 | [diff] [blame] | 14930 | - If "Tr" is nearly always lower than 3000 except some rare values which seem |
| 14931 | to be the average majored by 3000, there are probably some packets lost |
| 14932 | between the proxy and the server. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14933 | |
| 14934 | - If "Tt" is large even for small byte counts, it generally is because |
| 14935 | neither the client nor the server decides to close the connection, for |
| 14936 | instance because both have agreed on a keep-alive connection mode. In order |
| 14937 | to solve this issue, it will be needed to specify "option httpclose" on |
| 14938 | either the frontend or the backend. If the problem persists, it means that |
| 14939 | the server ignores the "close" connection mode and expects the client to |
| 14940 | close. Then it will be required to use "option forceclose". Having the |
| 14941 | smallest possible 'Tt' is important when connection regulation is used with |
| 14942 | the "maxconn" option on the servers, since no new connection will be sent |
| 14943 | to the server until another one is released. |
| 14944 | |
| 14945 | Other noticeable HTTP log cases ('xx' means any value to be ignored) : |
| 14946 | |
| 14947 | Tq/Tw/Tc/Tr/+Tt The "option logasap" is present on the frontend and the log |
| 14948 | was emitted before the data phase. All the timers are valid |
| 14949 | except "Tt" which is shorter than reality. |
| 14950 | |
| 14951 | -1/xx/xx/xx/Tt The client was not able to send a complete request in time |
| 14952 | or it aborted too early. Check the session termination flags |
| 14953 | then "timeout http-request" and "timeout client" settings. |
| 14954 | |
| 14955 | Tq/-1/xx/xx/Tt It was not possible to process the request, maybe because |
| 14956 | servers were out of order, because the request was invalid |
| 14957 | or forbidden by ACL rules. Check the session termination |
| 14958 | flags. |
| 14959 | |
| 14960 | Tq/Tw/-1/xx/Tt The connection could not establish on the server. Either it |
| 14961 | actively refused it or it timed out after Tt-(Tq+Tw) ms. |
| 14962 | Check the session termination flags, then check the |
| 14963 | "timeout connect" setting. Note that the tarpit action might |
| 14964 | return similar-looking patterns, with "Tw" equal to the time |
| 14965 | the client connection was maintained open. |
| 14966 | |
| 14967 | Tq/Tw/Tc/-1/Tt The server has accepted the connection but did not return |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 14968 | a complete response in time, or it closed its connection |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14969 | unexpectedly after Tt-(Tq+Tw+Tc) ms. Check the session |
| 14970 | termination flags, then check the "timeout server" setting. |
| 14971 | |
| 14972 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 14973 | 8.5. Session state at disconnection |
| 14974 | ----------------------------------- |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14975 | |
| 14976 | TCP and HTTP logs provide a session termination indicator in the |
| 14977 | "termination_state" field, just before the number of active connections. It is |
| 14978 | 2-characters long in TCP mode, and is extended to 4 characters in HTTP mode, |
| 14979 | each of which has a special meaning : |
| 14980 | |
| 14981 | - On the first character, a code reporting the first event which caused the |
| 14982 | session to terminate : |
| 14983 | |
| 14984 | C : the TCP session was unexpectedly aborted by the client. |
| 14985 | |
| 14986 | S : the TCP session was unexpectedly aborted by the server, or the |
| 14987 | server explicitly refused it. |
| 14988 | |
| 14989 | P : the session was prematurely aborted by the proxy, because of a |
| 14990 | connection limit enforcement, because a DENY filter was matched, |
| 14991 | because of a security check which detected and blocked a dangerous |
| 14992 | error in server response which might have caused information leak |
Willy Tarreau | 570f221 | 2013-06-10 16:42:09 +0200 | [diff] [blame] | 14993 | (eg: cacheable cookie). |
| 14994 | |
| 14995 | L : the session was locally processed by haproxy and was not passed to |
| 14996 | a server. This is what happens for stats and redirects. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 14997 | |
| 14998 | R : a resource on the proxy has been exhausted (memory, sockets, source |
| 14999 | ports, ...). Usually, this appears during the connection phase, and |
| 15000 | system logs should contain a copy of the precise error. If this |
| 15001 | happens, it must be considered as a very serious anomaly which |
| 15002 | should be fixed as soon as possible by any means. |
| 15003 | |
| 15004 | I : an internal error was identified by the proxy during a self-check. |
| 15005 | This should NEVER happen, and you are encouraged to report any log |
| 15006 | containing this, because this would almost certainly be a bug. It |
| 15007 | would be wise to preventively restart the process after such an |
| 15008 | event too, in case it would be caused by memory corruption. |
| 15009 | |
Simon Horman | 752dc4a | 2011-06-21 14:34:59 +0900 | [diff] [blame] | 15010 | D : the session was killed by haproxy because the server was detected |
| 15011 | as down and was configured to kill all connections when going down. |
| 15012 | |
Justin Karneges | eb2c24a | 2012-05-24 15:28:52 -0700 | [diff] [blame] | 15013 | U : the session was killed by haproxy on this backup server because an |
| 15014 | active server was detected as up and was configured to kill all |
| 15015 | backup connections when going up. |
| 15016 | |
Willy Tarreau | a2a64e9 | 2011-09-07 23:01:56 +0200 | [diff] [blame] | 15017 | K : the session was actively killed by an admin operating on haproxy. |
| 15018 | |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15019 | c : the client-side timeout expired while waiting for the client to |
| 15020 | send or receive data. |
| 15021 | |
| 15022 | s : the server-side timeout expired while waiting for the server to |
| 15023 | send or receive data. |
| 15024 | |
| 15025 | - : normal session completion, both the client and the server closed |
| 15026 | with nothing left in the buffers. |
| 15027 | |
| 15028 | - on the second character, the TCP or HTTP session state when it was closed : |
| 15029 | |
Willy Tarreau | f7b30a9 | 2010-12-06 22:59:17 +0100 | [diff] [blame] | 15030 | R : the proxy was waiting for a complete, valid REQUEST from the client |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15031 | (HTTP mode only). Nothing was sent to any server. |
| 15032 | |
| 15033 | Q : the proxy was waiting in the QUEUE for a connection slot. This can |
| 15034 | only happen when servers have a 'maxconn' parameter set. It can |
| 15035 | also happen in the global queue after a redispatch consecutive to |
| 15036 | a failed attempt to connect to a dying server. If no redispatch is |
| 15037 | reported, then no connection attempt was made to any server. |
| 15038 | |
| 15039 | C : the proxy was waiting for the CONNECTION to establish on the |
| 15040 | server. The server might at most have noticed a connection attempt. |
| 15041 | |
| 15042 | H : the proxy was waiting for complete, valid response HEADERS from the |
| 15043 | server (HTTP only). |
| 15044 | |
| 15045 | D : the session was in the DATA phase. |
| 15046 | |
| 15047 | L : the proxy was still transmitting LAST data to the client while the |
| 15048 | server had already finished. This one is very rare as it can only |
| 15049 | happen when the client dies while receiving the last packets. |
| 15050 | |
| 15051 | T : the request was tarpitted. It has been held open with the client |
| 15052 | during the whole "timeout tarpit" duration or until the client |
| 15053 | closed, both of which will be reported in the "Tw" timer. |
| 15054 | |
| 15055 | - : normal session completion after end of data transfer. |
| 15056 | |
| 15057 | - the third character tells whether the persistence cookie was provided by |
| 15058 | the client (only in HTTP mode) : |
| 15059 | |
| 15060 | N : the client provided NO cookie. This is usually the case for new |
| 15061 | visitors, so counting the number of occurrences of this flag in the |
| 15062 | logs generally indicate a valid trend for the site frequentation. |
| 15063 | |
| 15064 | I : the client provided an INVALID cookie matching no known server. |
| 15065 | This might be caused by a recent configuration change, mixed |
Cyril Bonté | a8e7bbc | 2010-04-25 22:29:29 +0200 | [diff] [blame] | 15066 | cookies between HTTP/HTTPS sites, persistence conditionally |
| 15067 | ignored, or an attack. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15068 | |
| 15069 | D : the client provided a cookie designating a server which was DOWN, |
| 15070 | so either "option persist" was used and the client was sent to |
| 15071 | this server, or it was not set and the client was redispatched to |
| 15072 | another server. |
| 15073 | |
Willy Tarreau | 996a92c | 2010-10-13 19:30:47 +0200 | [diff] [blame] | 15074 | V : the client provided a VALID cookie, and was sent to the associated |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15075 | server. |
| 15076 | |
Willy Tarreau | 996a92c | 2010-10-13 19:30:47 +0200 | [diff] [blame] | 15077 | E : the client provided a valid cookie, but with a last date which was |
| 15078 | older than what is allowed by the "maxidle" cookie parameter, so |
| 15079 | the cookie is consider EXPIRED and is ignored. The request will be |
| 15080 | redispatched just as if there was no cookie. |
| 15081 | |
| 15082 | O : the client provided a valid cookie, but with a first date which was |
| 15083 | older than what is allowed by the "maxlife" cookie parameter, so |
| 15084 | the cookie is consider too OLD and is ignored. The request will be |
| 15085 | redispatched just as if there was no cookie. |
| 15086 | |
Willy Tarreau | c89ccb6 | 2012-04-05 21:18:22 +0200 | [diff] [blame] | 15087 | U : a cookie was present but was not used to select the server because |
| 15088 | some other server selection mechanism was used instead (typically a |
| 15089 | "use-server" rule). |
| 15090 | |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15091 | - : does not apply (no cookie set in configuration). |
| 15092 | |
| 15093 | - the last character reports what operations were performed on the persistence |
| 15094 | cookie returned by the server (only in HTTP mode) : |
| 15095 | |
| 15096 | N : NO cookie was provided by the server, and none was inserted either. |
| 15097 | |
| 15098 | I : no cookie was provided by the server, and the proxy INSERTED one. |
| 15099 | Note that in "cookie insert" mode, if the server provides a cookie, |
| 15100 | it will still be overwritten and reported as "I" here. |
| 15101 | |
Willy Tarreau | 996a92c | 2010-10-13 19:30:47 +0200 | [diff] [blame] | 15102 | U : the proxy UPDATED the last date in the cookie that was presented by |
| 15103 | the client. This can only happen in insert mode with "maxidle". It |
Jarno Huuskonen | 0e82b92 | 2014-04-12 18:22:19 +0300 | [diff] [blame] | 15104 | happens every time there is activity at a different date than the |
Willy Tarreau | 996a92c | 2010-10-13 19:30:47 +0200 | [diff] [blame] | 15105 | date indicated in the cookie. If any other change happens, such as |
| 15106 | a redispatch, then the cookie will be marked as inserted instead. |
| 15107 | |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15108 | P : a cookie was PROVIDED by the server and transmitted as-is. |
| 15109 | |
| 15110 | R : the cookie provided by the server was REWRITTEN by the proxy, which |
| 15111 | happens in "cookie rewrite" or "cookie prefix" modes. |
| 15112 | |
| 15113 | D : the cookie provided by the server was DELETED by the proxy. |
| 15114 | |
| 15115 | - : does not apply (no cookie set in configuration). |
| 15116 | |
Willy Tarreau | 996a92c | 2010-10-13 19:30:47 +0200 | [diff] [blame] | 15117 | The combination of the two first flags gives a lot of information about what |
| 15118 | was happening when the session terminated, and why it did terminate. It can be |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15119 | helpful to detect server saturation, network troubles, local system resource |
| 15120 | starvation, attacks, etc... |
| 15121 | |
| 15122 | The most common termination flags combinations are indicated below. They are |
| 15123 | alphabetically sorted, with the lowercase set just after the upper case for |
| 15124 | easier finding and understanding. |
| 15125 | |
| 15126 | Flags Reason |
| 15127 | |
| 15128 | -- Normal termination. |
| 15129 | |
| 15130 | CC The client aborted before the connection could be established to the |
| 15131 | server. This can happen when haproxy tries to connect to a recently |
| 15132 | dead (or unchecked) server, and the client aborts while haproxy is |
| 15133 | waiting for the server to respond or for "timeout connect" to expire. |
| 15134 | |
| 15135 | CD The client unexpectedly aborted during data transfer. This can be |
| 15136 | caused by a browser crash, by an intermediate equipment between the |
| 15137 | client and haproxy which decided to actively break the connection, |
| 15138 | by network routing issues between the client and haproxy, or by a |
| 15139 | keep-alive session between the server and the client terminated first |
| 15140 | by the client. |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 15141 | |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15142 | cD The client did not send nor acknowledge any data for as long as the |
| 15143 | "timeout client" delay. This is often caused by network failures on |
Cyril Bonté | dc4d903 | 2012-04-08 21:57:39 +0200 | [diff] [blame] | 15144 | the client side, or the client simply leaving the net uncleanly. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15145 | |
| 15146 | CH The client aborted while waiting for the server to start responding. |
| 15147 | It might be the server taking too long to respond or the client |
| 15148 | clicking the 'Stop' button too fast. |
| 15149 | |
| 15150 | cH The "timeout client" stroke while waiting for client data during a |
| 15151 | POST request. This is sometimes caused by too large TCP MSS values |
| 15152 | for PPPoE networks which cannot transport full-sized packets. It can |
| 15153 | also happen when client timeout is smaller than server timeout and |
| 15154 | the server takes too long to respond. |
| 15155 | |
| 15156 | CQ The client aborted while its session was queued, waiting for a server |
| 15157 | with enough empty slots to accept it. It might be that either all the |
| 15158 | servers were saturated or that the assigned server was taking too |
| 15159 | long a time to respond. |
| 15160 | |
| 15161 | CR The client aborted before sending a full HTTP request. Most likely |
| 15162 | the request was typed by hand using a telnet client, and aborted |
| 15163 | too early. The HTTP status code is likely a 400 here. Sometimes this |
| 15164 | might also be caused by an IDS killing the connection between haproxy |
Willy Tarreau | 0f228a0 | 2015-05-01 15:37:53 +0200 | [diff] [blame] | 15165 | and the client. "option http-ignore-probes" can be used to ignore |
| 15166 | connections without any data transfer. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15167 | |
| 15168 | cR The "timeout http-request" stroke before the client sent a full HTTP |
| 15169 | request. This is sometimes caused by too large TCP MSS values on the |
| 15170 | client side for PPPoE networks which cannot transport full-sized |
| 15171 | packets, or by clients sending requests by hand and not typing fast |
| 15172 | enough, or forgetting to enter the empty line at the end of the |
Willy Tarreau | 2705a61 | 2014-05-23 17:38:34 +0200 | [diff] [blame] | 15173 | request. The HTTP status code is likely a 408 here. Note: recently, |
Willy Tarreau | 0f228a0 | 2015-05-01 15:37:53 +0200 | [diff] [blame] | 15174 | some browsers started to implement a "pre-connect" feature consisting |
| 15175 | in speculatively connecting to some recently visited web sites just |
| 15176 | in case the user would like to visit them. This results in many |
| 15177 | connections being established to web sites, which end up in 408 |
| 15178 | Request Timeout if the timeout strikes first, or 400 Bad Request when |
| 15179 | the browser decides to close them first. These ones pollute the log |
| 15180 | and feed the error counters. Some versions of some browsers have even |
| 15181 | been reported to display the error code. It is possible to work |
| 15182 | around the undesirable effects of this behaviour by adding "option |
| 15183 | http-ignore-probes" in the frontend, resulting in connections with |
| 15184 | zero data transfer to be totally ignored. This will definitely hide |
| 15185 | the errors of people experiencing connectivity issues though. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15186 | |
| 15187 | CT The client aborted while its session was tarpitted. It is important to |
| 15188 | check if this happens on valid requests, in order to be sure that no |
Willy Tarreau | 55165fe | 2009-05-10 12:02:55 +0200 | [diff] [blame] | 15189 | wrong tarpit rules have been written. If a lot of them happen, it |
| 15190 | might make sense to lower the "timeout tarpit" value to something |
| 15191 | closer to the average reported "Tw" timer, in order not to consume |
| 15192 | resources for just a few attackers. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15193 | |
Willy Tarreau | 570f221 | 2013-06-10 16:42:09 +0200 | [diff] [blame] | 15194 | LR The request was intercepted and locally handled by haproxy. Generally |
| 15195 | it means that this was a redirect or a stats request. |
| 15196 | |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 15197 | SC The server or an equipment between it and haproxy explicitly refused |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15198 | the TCP connection (the proxy received a TCP RST or an ICMP message |
| 15199 | in return). Under some circumstances, it can also be the network |
| 15200 | stack telling the proxy that the server is unreachable (eg: no route, |
| 15201 | or no ARP response on local network). When this happens in HTTP mode, |
| 15202 | the status code is likely a 502 or 503 here. |
| 15203 | |
| 15204 | sC The "timeout connect" stroke before a connection to the server could |
| 15205 | complete. When this happens in HTTP mode, the status code is likely a |
| 15206 | 503 or 504 here. |
| 15207 | |
| 15208 | SD The connection to the server died with an error during the data |
| 15209 | transfer. This usually means that haproxy has received an RST from |
| 15210 | the server or an ICMP message from an intermediate equipment while |
| 15211 | exchanging data with the server. This can be caused by a server crash |
| 15212 | or by a network issue on an intermediate equipment. |
| 15213 | |
| 15214 | sD The server did not send nor acknowledge any data for as long as the |
| 15215 | "timeout server" setting during the data phase. This is often caused |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 15216 | by too short timeouts on L4 equipments before the server (firewalls, |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15217 | load-balancers, ...), as well as keep-alive sessions maintained |
| 15218 | between the client and the server expiring first on haproxy. |
| 15219 | |
| 15220 | SH The server aborted before sending its full HTTP response headers, or |
| 15221 | it crashed while processing the request. Since a server aborting at |
| 15222 | this moment is very rare, it would be wise to inspect its logs to |
| 15223 | control whether it crashed and why. The logged request may indicate a |
| 15224 | small set of faulty requests, demonstrating bugs in the application. |
| 15225 | Sometimes this might also be caused by an IDS killing the connection |
| 15226 | between haproxy and the server. |
| 15227 | |
| 15228 | sH The "timeout server" stroke before the server could return its |
| 15229 | response headers. This is the most common anomaly, indicating too |
| 15230 | long transactions, probably caused by server or database saturation. |
| 15231 | The immediate workaround consists in increasing the "timeout server" |
| 15232 | setting, but it is important to keep in mind that the user experience |
| 15233 | will suffer from these long response times. The only long term |
| 15234 | solution is to fix the application. |
| 15235 | |
| 15236 | sQ The session spent too much time in queue and has been expired. See |
| 15237 | the "timeout queue" and "timeout connect" settings to find out how to |
| 15238 | fix this if it happens too often. If it often happens massively in |
| 15239 | short periods, it may indicate general problems on the affected |
| 15240 | servers due to I/O or database congestion, or saturation caused by |
| 15241 | external attacks. |
| 15242 | |
| 15243 | PC The proxy refused to establish a connection to the server because the |
| 15244 | process' socket limit has been reached while attempting to connect. |
Cyril Bonté | dc4d903 | 2012-04-08 21:57:39 +0200 | [diff] [blame] | 15245 | The global "maxconn" parameter may be increased in the configuration |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15246 | so that it does not happen anymore. This status is very rare and |
| 15247 | might happen when the global "ulimit-n" parameter is forced by hand. |
| 15248 | |
Willy Tarreau | ed2fd2d | 2010-12-29 11:23:27 +0100 | [diff] [blame] | 15249 | PD The proxy blocked an incorrectly formatted chunked encoded message in |
| 15250 | a request or a response, after the server has emitted its headers. In |
| 15251 | most cases, this will indicate an invalid message from the server to |
Willy Tarreau | f3a3e13 | 2013-08-31 08:16:26 +0200 | [diff] [blame] | 15252 | the client. Haproxy supports chunk sizes of up to 2GB - 1 (2147483647 |
| 15253 | bytes). Any larger size will be considered as an error. |
Willy Tarreau | ed2fd2d | 2010-12-29 11:23:27 +0100 | [diff] [blame] | 15254 | |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15255 | PH The proxy blocked the server's response, because it was invalid, |
| 15256 | incomplete, dangerous (cache control), or matched a security filter. |
| 15257 | In any case, an HTTP 502 error is sent to the client. One possible |
| 15258 | cause for this error is an invalid syntax in an HTTP header name |
Willy Tarreau | ed2fd2d | 2010-12-29 11:23:27 +0100 | [diff] [blame] | 15259 | containing unauthorized characters. It is also possible but quite |
| 15260 | rare, that the proxy blocked a chunked-encoding request from the |
| 15261 | client due to an invalid syntax, before the server responded. In this |
| 15262 | case, an HTTP 400 error is sent to the client and reported in the |
| 15263 | logs. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15264 | |
| 15265 | PR The proxy blocked the client's HTTP request, either because of an |
| 15266 | invalid HTTP syntax, in which case it returned an HTTP 400 error to |
| 15267 | the client, or because a deny filter matched, in which case it |
| 15268 | returned an HTTP 403 error. |
| 15269 | |
| 15270 | PT The proxy blocked the client's request and has tarpitted its |
| 15271 | connection before returning it a 500 server error. Nothing was sent |
| 15272 | to the server. The connection was maintained open for as long as |
| 15273 | reported by the "Tw" timer field. |
| 15274 | |
| 15275 | RC A local resource has been exhausted (memory, sockets, source ports) |
| 15276 | preventing the connection to the server from establishing. The error |
| 15277 | logs will tell precisely what was missing. This is very rare and can |
| 15278 | only be solved by proper system tuning. |
| 15279 | |
Willy Tarreau | 996a92c | 2010-10-13 19:30:47 +0200 | [diff] [blame] | 15280 | The combination of the two last flags gives a lot of information about how |
| 15281 | persistence was handled by the client, the server and by haproxy. This is very |
| 15282 | important to troubleshoot disconnections, when users complain they have to |
| 15283 | re-authenticate. The commonly encountered flags are : |
| 15284 | |
| 15285 | -- Persistence cookie is not enabled. |
| 15286 | |
| 15287 | NN No cookie was provided by the client, none was inserted in the |
| 15288 | response. For instance, this can be in insert mode with "postonly" |
| 15289 | set on a GET request. |
| 15290 | |
| 15291 | II A cookie designating an invalid server was provided by the client, |
| 15292 | a valid one was inserted in the response. This typically happens when |
Jamie Gloudon | aaa2100 | 2012-08-25 00:18:33 -0400 | [diff] [blame] | 15293 | a "server" entry is removed from the configuration, since its cookie |
Willy Tarreau | 996a92c | 2010-10-13 19:30:47 +0200 | [diff] [blame] | 15294 | value can be presented by a client when no other server knows it. |
| 15295 | |
| 15296 | NI No cookie was provided by the client, one was inserted in the |
| 15297 | response. This typically happens for first requests from every user |
| 15298 | in "insert" mode, which makes it an easy way to count real users. |
| 15299 | |
| 15300 | VN A cookie was provided by the client, none was inserted in the |
| 15301 | response. This happens for most responses for which the client has |
| 15302 | already got a cookie. |
| 15303 | |
| 15304 | VU A cookie was provided by the client, with a last visit date which is |
| 15305 | not completely up-to-date, so an updated cookie was provided in |
| 15306 | response. This can also happen if there was no date at all, or if |
| 15307 | there was a date but the "maxidle" parameter was not set, so that the |
| 15308 | cookie can be switched to unlimited time. |
| 15309 | |
| 15310 | EI A cookie was provided by the client, with a last visit date which is |
| 15311 | too old for the "maxidle" parameter, so the cookie was ignored and a |
| 15312 | new cookie was inserted in the response. |
| 15313 | |
| 15314 | OI A cookie was provided by the client, with a first visit date which is |
| 15315 | too old for the "maxlife" parameter, so the cookie was ignored and a |
| 15316 | new cookie was inserted in the response. |
| 15317 | |
| 15318 | DI The server designated by the cookie was down, a new server was |
| 15319 | selected and a new cookie was emitted in the response. |
| 15320 | |
| 15321 | VI The server designated by the cookie was not marked dead but could not |
| 15322 | be reached. A redispatch happened and selected another one, which was |
| 15323 | then advertised in the response. |
| 15324 | |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15325 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 15326 | 8.6. Non-printable characters |
| 15327 | ----------------------------- |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15328 | |
| 15329 | In order not to cause trouble to log analysis tools or terminals during log |
| 15330 | consulting, non-printable characters are not sent as-is into log files, but are |
| 15331 | converted to the two-digits hexadecimal representation of their ASCII code, |
| 15332 | prefixed by the character '#'. The only characters that can be logged without |
| 15333 | being escaped are comprised between 32 and 126 (inclusive). Obviously, the |
| 15334 | escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It |
| 15335 | is the same for the character '"' which becomes "#22", as well as '{', '|' and |
| 15336 | '}' when logging headers. |
| 15337 | |
| 15338 | Note that the space character (' ') is not encoded in headers, which can cause |
| 15339 | issues for tools relying on space count to locate fields. A typical header |
| 15340 | containing spaces is "User-Agent". |
| 15341 | |
| 15342 | Last, it has been observed that some syslog daemons such as syslog-ng escape |
| 15343 | the quote ('"') with a backslash ('\'). The reverse operation can safely be |
| 15344 | performed since no quote may appear anywhere else in the logs. |
| 15345 | |
| 15346 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 15347 | 8.7. Capturing HTTP cookies |
| 15348 | --------------------------- |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15349 | |
| 15350 | Cookie capture simplifies the tracking a complete user session. This can be |
| 15351 | achieved using the "capture cookie" statement in the frontend. Please refer to |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 15352 | section 4.2 for more details. Only one cookie can be captured, and the same |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15353 | cookie will simultaneously be checked in the request ("Cookie:" header) and in |
| 15354 | the response ("Set-Cookie:" header). The respective values will be reported in |
| 15355 | the HTTP logs at the "captured_request_cookie" and "captured_response_cookie" |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 15356 | locations (see section 8.2.3 about HTTP log format). When either cookie is |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15357 | not seen, a dash ('-') replaces the value. This way, it's easy to detect when a |
| 15358 | user switches to a new session for example, because the server will reassign it |
| 15359 | a new cookie. It is also possible to detect if a server unexpectedly sets a |
| 15360 | wrong cookie to a client, leading to session crossing. |
| 15361 | |
| 15362 | Examples : |
| 15363 | # capture the first cookie whose name starts with "ASPSESSION" |
| 15364 | capture cookie ASPSESSION len 32 |
| 15365 | |
| 15366 | # capture the first cookie whose name is exactly "vgnvisitor" |
| 15367 | capture cookie vgnvisitor= len 32 |
| 15368 | |
| 15369 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 15370 | 8.8. Capturing HTTP headers |
| 15371 | --------------------------- |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15372 | |
| 15373 | Header captures are useful to track unique request identifiers set by an upper |
| 15374 | proxy, virtual host names, user-agents, POST content-length, referrers, etc. In |
| 15375 | the response, one can search for information about the response length, how the |
| 15376 | server asked the cache to behave, or an object location during a redirection. |
| 15377 | |
| 15378 | Header captures are performed using the "capture request header" and "capture |
| 15379 | response header" statements in the frontend. Please consult their definition in |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 15380 | section 4.2 for more details. |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15381 | |
| 15382 | It is possible to include both request headers and response headers at the same |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 15383 | time. Non-existent headers are logged as empty strings, and if one header |
| 15384 | appears more than once, only its last occurrence will be logged. Request headers |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15385 | are grouped within braces '{' and '}' in the same order as they were declared, |
| 15386 | and delimited with a vertical bar '|' without any space. Response headers |
| 15387 | follow the same representation, but are displayed after a space following the |
| 15388 | request headers block. These blocks are displayed just before the HTTP request |
| 15389 | in the logs. |
| 15390 | |
Willy Tarreau | d9ed3d2 | 2014-06-13 12:23:06 +0200 | [diff] [blame] | 15391 | As a special case, it is possible to specify an HTTP header capture in a TCP |
| 15392 | frontend. The purpose is to enable logging of headers which will be parsed in |
| 15393 | an HTTP backend if the request is then switched to this HTTP backend. |
| 15394 | |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15395 | Example : |
| 15396 | # This instance chains to the outgoing proxy |
| 15397 | listen proxy-out |
| 15398 | mode http |
| 15399 | option httplog |
| 15400 | option logasap |
| 15401 | log global |
| 15402 | server cache1 192.168.1.1:3128 |
| 15403 | |
| 15404 | # log the name of the virtual server |
| 15405 | capture request header Host len 20 |
| 15406 | |
| 15407 | # log the amount of data uploaded during a POST |
| 15408 | capture request header Content-Length len 10 |
| 15409 | |
| 15410 | # log the beginning of the referrer |
| 15411 | capture request header Referer len 20 |
| 15412 | |
| 15413 | # server name (useful for outgoing proxies only) |
| 15414 | capture response header Server len 20 |
| 15415 | |
| 15416 | # logging the content-length is useful with "option logasap" |
| 15417 | capture response header Content-Length len 10 |
| 15418 | |
| 15419 | # log the expected cache behaviour on the response |
| 15420 | capture response header Cache-Control len 8 |
| 15421 | |
| 15422 | # the Via header will report the next proxy's name |
| 15423 | capture response header Via len 20 |
| 15424 | |
| 15425 | # log the URL location during a redirection |
| 15426 | capture response header Location len 20 |
| 15427 | |
| 15428 | >>> Aug 9 20:26:09 localhost \ |
| 15429 | haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \ |
| 15430 | proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \ |
| 15431 | {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \ |
| 15432 | "GET http://fr.adserver.yahoo.com/" |
| 15433 | |
| 15434 | >>> Aug 9 20:30:46 localhost \ |
| 15435 | haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \ |
| 15436 | proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \ |
| 15437 | {w.ods.org||} {Formilux/0.1.8|3495|||} \ |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 15438 | "GET http://trafic.1wt.eu/ HTTP/1.1" |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15439 | |
| 15440 | >>> Aug 9 20:30:46 localhost \ |
| 15441 | haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \ |
| 15442 | proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \ |
| 15443 | {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \ |
| 15444 | {Apache|230|||http://www.sytadin.} \ |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 15445 | "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1" |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15446 | |
| 15447 | |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 15448 | 8.9. Examples of logs |
| 15449 | --------------------- |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15450 | |
| 15451 | These are real-world examples of logs accompanied with an explanation. Some of |
| 15452 | them have been made up by hand. The syslog part has been removed for better |
| 15453 | reading. Their sole purpose is to explain how to decipher them. |
| 15454 | |
| 15455 | >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \ |
| 15456 | px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \ |
| 15457 | "HEAD / HTTP/1.0" |
| 15458 | |
| 15459 | => long request (6.5s) entered by hand through 'telnet'. The server replied |
| 15460 | in 147 ms, and the session ended normally ('----') |
| 15461 | |
| 15462 | >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \ |
| 15463 | px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \ |
| 15464 | 0/9 "HEAD / HTTP/1.0" |
| 15465 | |
| 15466 | => Idem, but the request was queued in the global queue behind 9 other |
| 15467 | requests, and waited there for 1230 ms. |
| 15468 | |
| 15469 | >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \ |
| 15470 | px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \ |
| 15471 | "GET /image.iso HTTP/1.0" |
| 15472 | |
| 15473 | => request for a long data transfer. The "logasap" option was specified, so |
Krzysztof Piotr Oledzki | f864533 | 2009-12-13 21:55:50 +0100 | [diff] [blame] | 15474 | the log was produced just before transferring data. The server replied in |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15475 | 14 ms, 243 bytes of headers were sent to the client, and total time from |
| 15476 | accept to first data byte is 30 ms. |
| 15477 | |
| 15478 | >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \ |
| 15479 | px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \ |
| 15480 | "GET /cgi-bin/bug.cgi? HTTP/1.0" |
| 15481 | |
| 15482 | => the proxy blocked a server response either because of an "rspdeny" or |
| 15483 | "rspideny" filter, or because the response was improperly formatted and |
Willy Tarreau | 3c92c5f | 2011-08-28 09:45:47 +0200 | [diff] [blame] | 15484 | not HTTP-compliant, or because it blocked sensitive information which |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15485 | risked being cached. In this case, the response is replaced with a "502 |
| 15486 | bad gateway". The flags ("PH--") tell us that it was haproxy who decided |
| 15487 | to return the 502 and not the server. |
| 15488 | |
| 15489 | >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \ |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 15490 | px-http/<NOSRV> -1/-1/-1/-1/8490 -1 0 - - CR-- 2/2/2/0/0 0/0 "" |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15491 | |
| 15492 | => the client never completed its request and aborted itself ("C---") after |
| 15493 | 8.5s, while the proxy was waiting for the request headers ("-R--"). |
| 15494 | Nothing was sent to any server. |
| 15495 | |
| 15496 | >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \ |
| 15497 | px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 "" |
| 15498 | |
| 15499 | => The client never completed its request, which was aborted by the |
| 15500 | time-out ("c---") after 50s, while the proxy was waiting for the request |
| 15501 | headers ("-R--"). Nothing was sent to any server, but the proxy could |
| 15502 | send a 408 return code to the client. |
| 15503 | |
| 15504 | >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \ |
| 15505 | px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0 |
| 15506 | |
| 15507 | => This log was produced with "option tcplog". The client timed out after |
| 15508 | 5 seconds ("c----"). |
| 15509 | |
| 15510 | >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \ |
| 15511 | px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \ |
Willy Tarreau | d72758d | 2010-01-12 10:42:19 +0100 | [diff] [blame] | 15512 | 0/0 "HEAD / HTTP/1.0" |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15513 | |
| 15514 | => The request took 3s to complete (probably a network problem), and the |
Willy Tarreau | c57f0e2 | 2009-05-10 13:12:33 +0200 | [diff] [blame] | 15515 | connection to the server failed ('SC--') after 4 attempts of 2 seconds |
Willy Tarreau | cc6c891 | 2009-02-22 10:53:55 +0100 | [diff] [blame] | 15516 | (config says 'retries 3'), and no redispatch (otherwise we would have |
| 15517 | seen "/+3"). Status code 503 was returned to the client. There were 115 |
| 15518 | connections on this server, 202 connections on this proxy, and 205 on |
| 15519 | the global process. It is possible that the server refused the |
| 15520 | connection because of too many already established. |
Willy Tarreau | 844e3c5 | 2008-01-11 16:28:18 +0100 | [diff] [blame] | 15521 | |
Willy Tarreau | 52b2d22 | 2011-09-07 23:48:48 +0200 | [diff] [blame] | 15522 | |
Christopher Faulet | c3fe533 | 2016-04-07 15:30:10 +0200 | [diff] [blame] | 15523 | 9. Supported filters |
| 15524 | -------------------- |
| 15525 | |
| 15526 | Here are listed officially supported filters with the list of parameters they |
| 15527 | accept. Depending on compile options, some of these filters might be |
| 15528 | unavailable. The list of available filters is reported in haproxy -vv. |
| 15529 | |
| 15530 | See also : "filter" |
| 15531 | |
| 15532 | 9.1. Trace |
| 15533 | ---------- |
| 15534 | |
| 15535 | filter trace [name <name>] [random-parsing] [random-forwarding] |
| 15536 | |
| 15537 | Arguments: |
| 15538 | <name> is an arbitrary name that will be reported in |
| 15539 | messages. If no name is provided, "TRACE" is used. |
| 15540 | |
| 15541 | <random-parsing> enables the random parsing of data exchanged between |
| 15542 | the client and the server. By default, this filter |
| 15543 | parses all available data. With this parameter, it |
| 15544 | only parses a random amount of the available data. |
| 15545 | |
| 15546 | <random-forwarding> enables the random forwading of parsed data. By |
| 15547 | default, this filter forwards all previously parsed |
| 15548 | data. With this parameter, it only forwards a random |
| 15549 | amount of the parsed data. |
| 15550 | |
| 15551 | This filter can be used as a base to develop new filters. It defines all |
| 15552 | callbacks and print a message on the standard error stream (stderr) with useful |
| 15553 | information for all of them. It may be useful to debug the activity of other |
| 15554 | filters or, quite simply, HAProxy's activity. |
| 15555 | |
| 15556 | Using <random-parsing> and/or <random-forwarding> parameters is a good way to |
| 15557 | tests the behavior of a filter that parses data exchanged between a client and |
| 15558 | a server by adding some latencies in the processing. |
| 15559 | |
| 15560 | |
| 15561 | 9.2. HTTP compression |
| 15562 | --------------------- |
| 15563 | |
| 15564 | filter compression |
| 15565 | |
| 15566 | The HTTP compression has been moved in a filter in HAProxy 1.7. "compression" |
| 15567 | keyword must still be used to enable and configure the HTTP compression. And |
| 15568 | when no other filter is used, it is enough. But it is mandatory to explicitly |
| 15569 | use a filter line to enable the HTTP compression when two or more filters are |
| 15570 | used for the same listener/frontend/backend. This is important to know the |
| 15571 | filters evaluation order. |
| 15572 | |
| 15573 | See also : "compression" |
| 15574 | |
| 15575 | |
Willy Tarreau | 0ba2750 | 2007-12-24 16:55:16 +0100 | [diff] [blame] | 15576 | /* |
| 15577 | * Local variables: |
| 15578 | * fill-column: 79 |
| 15579 | * End: |
| 15580 | */ |