[MEDIUM] differentiate between generic LB params and map-specific ones
Since the introduction of server weights, all load balancing algorithms
relied on a pre-computed map. Incidently, quite a bunch of map-specific
parameters were used at random places in order to get the number of
servers or their total weight. It was not architecturally acceptable
that optimizations for the map computation had impact on external parts.
For instance, during this cleanup it was found that a backend weight was
seen as 1 when only the first backup server is used, whatever its weight.
This cleanup consists in differentiating between LB-generic parameters,
such as total weights, number of servers, etc... and map-specific ones.
The struct proxy has been enhanced in order to make it easier to later
support other algorithms. The recount_servers() function now also
updates generic values such as total weights so that it's not needed
anymore to call recalc_server_map() when weights are needed. This
permitted to simplify some code which does not need to know about map
internals anymore.
diff --git a/include/proto/backend.h b/include/proto/backend.h
index fc71717..7414ede 100644
--- a/include/proto/backend.h
+++ b/include/proto/backend.h
@@ -46,7 +46,7 @@
/*
* This function tries to find a running server with free connection slots for
* the proxy <px> following the round-robin method.
- * If any server is found, it will be returned and px->srv_rr_idx will be updated
+ * If any server is found, it will be returned and px->lbprm.map.rr_idx will be updated
* to point to the next server. If no valid server is found, NULL is returned.
*/
static inline struct server *get_server_rr_with_conns(struct proxy *px)
@@ -54,25 +54,25 @@
int newidx;
struct server *srv;
- if (px->map_state & PR_MAP_RECALC)
- recalc_server_map(px);
-
- if (px->srv_map_sz == 0)
+ if (px->lbprm.tot_weight == 0)
return NULL;
+ if (px->lbprm.map.state & PR_MAP_RECALC)
+ recalc_server_map(px);
+
- if (px->srv_rr_idx < 0 || px->srv_rr_idx >= px->srv_map_sz)
- px->srv_rr_idx = 0;
- newidx = px->srv_rr_idx;
+ if (px->lbprm.map.rr_idx < 0 || px->lbprm.map.rr_idx >= px->lbprm.tot_weight)
+ px->lbprm.map.rr_idx = 0;
+ newidx = px->lbprm.map.rr_idx;
do {
- srv = px->srv_map[newidx++];
+ srv = px->lbprm.map.srv[newidx++];
if (!srv->maxconn || srv->cur_sess < srv_dynamic_maxconn(srv)) {
- px->srv_rr_idx = newidx;
+ px->lbprm.map.rr_idx = newidx;
return srv;
}
- if (newidx == px->srv_map_sz)
+ if (newidx == px->lbprm.tot_weight)
newidx = 0;
- } while (newidx != px->srv_rr_idx);
+ } while (newidx != px->lbprm.map.rr_idx);
return NULL;
}
@@ -81,20 +81,20 @@
/*
* This function tries to find a running server for the proxy <px> following
* the round-robin method.
- * If any server is found, it will be returned and px->srv_rr_idx will be updated
+ * If any server is found, it will be returned and px->lbprm.map.rr_idx will be updated
* to point to the next server. If no valid server is found, NULL is returned.
*/
static inline struct server *get_server_rr(struct proxy *px)
{
- if (px->map_state & PR_MAP_RECALC)
- recalc_server_map(px);
-
- if (px->srv_map_sz == 0)
+ if (px->lbprm.tot_weight == 0)
return NULL;
+ if (px->lbprm.map.state & PR_MAP_RECALC)
+ recalc_server_map(px);
+
- if (px->srv_rr_idx < 0 || px->srv_rr_idx >= px->srv_map_sz)
- px->srv_rr_idx = 0;
- return px->srv_map[px->srv_rr_idx++];
+ if (px->lbprm.map.rr_idx < 0 || px->lbprm.map.rr_idx >= px->lbprm.tot_weight)
+ px->lbprm.map.rr_idx = 0;
+ return px->lbprm.map.srv[px->lbprm.map.rr_idx++];
}
@@ -110,21 +110,21 @@
{
unsigned int h, l;
- if (px->map_state & PR_MAP_RECALC)
- recalc_server_map(px);
-
- if (px->srv_map_sz == 0)
+ if (px->lbprm.tot_weight == 0)
return NULL;
+ if (px->lbprm.map.state & PR_MAP_RECALC)
+ recalc_server_map(px);
+
l = h = 0;
if (px->srv_act > 1 || (px->srv_act == 0 && px->srv_bck > 1)) {
while ((l + sizeof (int)) <= len) {
h ^= ntohl(*(unsigned int *)(&addr[l]));
l += sizeof (int);
}
- h %= px->srv_map_sz;
+ h %= px->lbprm.tot_weight;
}
- return px->srv_map[h];
+ return px->lbprm.map.srv[h];
}
/*
@@ -144,12 +144,12 @@
unsigned long hash = 0;
int c;
- if (px->map_state & PR_MAP_RECALC)
- recalc_server_map(px);
-
- if (px->srv_map_sz == 0)
+ if (px->lbprm.tot_weight == 0)
return NULL;
+ if (px->lbprm.map.state & PR_MAP_RECALC)
+ recalc_server_map(px);
+
while (uri_len--) {
c = *uri++;
if (c == '?')
@@ -157,7 +157,7 @@
hash = c + (hash << 6) + (hash << 16) - hash;
}
- return px->srv_map[hash % px->srv_map_sz];
+ return px->lbprm.map.srv[hash % px->lbprm.tot_weight];
}
diff --git a/include/types/proxy.h b/include/types/proxy.h
index ecc4d9c..03f9601 100644
--- a/include/types/proxy.h
+++ b/include/types/proxy.h
@@ -54,7 +54,7 @@
#define PR_MODE_HTTP 1
#define PR_MODE_HEALTH 2
-/* values for proxy->map_state */
+/* values for proxy->lbprm.map.state */
#define PR_MAP_RECALC (1 << 0)
/* flag values for proxy->cap. This is a bitmask of capabilities supported by the proxy */
@@ -78,13 +78,21 @@
struct list acl; /* ACL declared on this proxy */
struct list block_cond; /* early blocking conditions (chained) */
struct list switching_rules; /* content switching rules (chained) */
- int map_state; /* PR_MAP_RECALC */
struct server *srv; /* known servers */
int srv_act, srv_bck; /* # of running servers */
- int tot_wact, tot_wbck; /* total weights of active and backup servers */
- struct server **srv_map; /* the server map used to apply weights */
- int srv_map_sz; /* the size of the effective server map */
- int srv_rr_idx; /* next server to be elected in round robin mode */
+
+ struct {
+ int tot_wact, tot_wbck; /* total effective weights of active and backup servers */
+ int tot_weight; /* total effective weight of servers participating to LB */
+ int tot_used; /* total number of servers used for LB */
+ int wmult; /* ratio between user weight and effective weight */
+ struct {
+ struct server **srv; /* the server map used to apply weights */
+ int rr_idx; /* next server to be elected in round robin mode */
+ int state; /* PR_MAP_RECALC */
+ } map; /* LB parameters for map-based algorithms */
+ } lbprm; /* LB parameters for all algorithms */
+
char *cookie_name; /* name of the cookie to look for */
int cookie_len; /* strlen(cookie_name), computed only once */
char *url_param_name; /* name of the URL parameter used for hashing */
diff --git a/src/backend.c b/src/backend.c
index 2ae25c3..96de0ba 100644
--- a/src/backend.c
+++ b/src/backend.c
@@ -54,50 +54,77 @@
void recount_servers(struct proxy *px)
{
struct server *srv;
+ int first_bkw = 0;
- px->srv_act = 0; px->srv_bck = px->tot_wact = px->tot_wbck = 0;
+ px->srv_act = px->srv_bck = 0;
+ px->lbprm.tot_wact = px->lbprm.tot_wbck = 0;
for (srv = px->srv; srv != NULL; srv = srv->next) {
if (srv->state & SRV_RUNNING) {
if (srv->state & SRV_BACKUP) {
px->srv_bck++;
- px->tot_wbck += srv->eweight;
+ px->lbprm.tot_wbck += srv->eweight;
+ if (px->srv_bck == 1)
+ first_bkw = srv->eweight;
} else {
px->srv_act++;
- px->tot_wact += srv->eweight;
+ px->lbprm.tot_wact += srv->eweight;
}
}
}
+
+ if (px->srv_act) {
+ px->lbprm.tot_weight = px->lbprm.tot_wact;
+ px->lbprm.tot_used = px->srv_act;
+ }
+ else if (px->srv_bck) {
+ if (px->options & PR_O_USE_ALL_BK) {
+ px->lbprm.tot_weight = px->lbprm.tot_wbck;
+ px->lbprm.tot_used = px->srv_bck;
+ }
+ else { /* the first backup server is enough */
+ px->lbprm.tot_weight = first_bkw;
+ px->lbprm.tot_used = 1;
+ }
+ }
+ else {
+ px->lbprm.tot_weight = 0;
+ px->lbprm.tot_used = 0;
+ }
+
}
-/* This function recomputes the server map for proxy px. It
- * relies on px->tot_wact and px->tot_wbck, so it must be
- * called after recount_servers(). It also expects px->srv_map
- * to be initialized to the largest value needed.
+/* This function recomputes the server map for proxy px. It relies on
+ * px->lbprm.tot_wact, tot_wbck, tot_used, tot_weight, so it must be
+ * called after recount_servers(). It also expects px->lbprm.map.srv
+ * to be allocated with the largest size needed. It updates tot_weight.
*/
void recalc_server_map(struct proxy *px)
{
int o, tot, flag;
struct server *cur, *best;
- if (px->srv_act) {
- flag = SRV_RUNNING;
- tot = px->tot_wact;
- } else if (px->srv_bck) {
- flag = SRV_RUNNING | SRV_BACKUP;
- if (px->options & PR_O_USE_ALL_BK)
- tot = px->tot_wbck;
- else
- tot = 1; /* the first server is enough */
- } else {
- px->srv_map_sz = 0;
- px->map_state &= ~PR_MAP_RECALC;
+ switch (px->lbprm.tot_used) {
+ case 0: /* no server */
+ px->lbprm.map.state &= ~PR_MAP_RECALC;
return;
+ case 1: /* only one server, just fill first entry */
+ tot = 1;
+ break;
+ default:
+ tot = px->lbprm.tot_weight;
+ break;
}
+ /* here we *know* that we have some servers */
+ if (px->srv_act)
+ flag = SRV_RUNNING;
+ else
+ flag = SRV_RUNNING | SRV_BACKUP;
+
/* this algorithm gives priority to the first server, which means that
* it will respect the declaration order for equivalent weights, and
* that whatever the weights, the first server called will always be
- * the first declard. This is an important asumption for the backup
+ * the first declared. This is an important asumption for the backup
* case, where we want the first server only.
*/
for (cur = px->srv; cur; cur = cur->next)
@@ -116,6 +143,7 @@
*/
if (tot == 1) {
best = cur;
+ /* note that best->wscore will be wrong but we don't care */
break;
}
@@ -127,11 +155,10 @@
}
}
}
- px->srv_map[o] = best;
+ px->lbprm.map.srv[o] = best;
best->wscore -= tot;
}
- px->srv_map_sz = tot;
- px->map_state &= ~PR_MAP_RECALC;
+ px->lbprm.map.state &= ~PR_MAP_RECALC;
}
/*
@@ -150,12 +177,12 @@
char *p;
int plen;
- if (px->map_state & PR_MAP_RECALC)
- recalc_server_map(px);
-
- if (px->srv_map_sz == 0)
+ if (px->lbprm.tot_weight == 0)
return NULL;
+ if (px->lbprm.map.state & PR_MAP_RECALC)
+ recalc_server_map(px);
+
p = memchr(uri, '?', uri_len);
if (!p)
return NULL;
@@ -183,7 +210,7 @@
uri_len--;
p++;
}
- return px->srv_map[hash % px->srv_map_sz];
+ return px->lbprm.map.srv[hash % px->lbprm.tot_weight];
}
}
diff --git a/src/cfgparse.c b/src/cfgparse.c
index cd8bdbb..05864d8 100644
--- a/src/cfgparse.c
+++ b/src/cfgparse.c
@@ -2652,6 +2652,8 @@
curproxy->srv = next;
}
+ curproxy->lbprm.wmult = 1; /* default weight multiplier */
+
/* now, newsrv == curproxy->srv */
if (newsrv) {
struct server *srv;
@@ -2673,6 +2675,12 @@
}
}
+ /* It is sometimes useful to know what factor to apply
+ * to the backend's effective weight to know its real
+ * weight.
+ */
+ curproxy->lbprm.wmult = pgcd;
+
act = bck = 0;
for (srv = newsrv; srv; srv = srv->next) {
srv->eweight = srv->uweight / pgcd;
@@ -2686,9 +2694,9 @@
if (act < bck)
act = bck;
- curproxy->srv_map = (struct server **)calloc(act, sizeof(struct server *));
+ curproxy->lbprm.map.srv = (struct server **)calloc(act, sizeof(struct server *));
/* recounts servers and their weights */
- curproxy->map_state = PR_MAP_RECALC;
+ curproxy->lbprm.map.state = PR_MAP_RECALC;
recount_servers(curproxy);
recalc_server_map(curproxy);
}
diff --git a/src/checks.c b/src/checks.c
index 77e9b51..3713210 100644
--- a/src/checks.c
+++ b/src/checks.c
@@ -63,7 +63,7 @@
s->state &= ~SRV_RUNNING;
recount_servers(s->proxy);
- s->proxy->map_state |= PR_MAP_RECALC;
+ s->proxy->lbprm.map.state |= PR_MAP_RECALC;
/* we might have sessions queued on this server and waiting for
* a connection. Those which are redispatchable will be queued
@@ -479,7 +479,7 @@
}
recount_servers(s->proxy);
- s->proxy->map_state |= PR_MAP_RECALC;
+ s->proxy->lbprm.map.state |= PR_MAP_RECALC;
/* check if we can handle some connections queued at the proxy. We
* will take as many as we can handle.
diff --git a/src/dumpstats.c b/src/dumpstats.c
index fd16a63..2c91b43 100644
--- a/src/dumpstats.c
+++ b/src/dumpstats.c
@@ -842,18 +842,6 @@
case DATA_ST_PX_BE:
/* print the backend */
if (px->cap & PR_CAP_BE) {
- int gcd = 1;
-
- if (px->map_state & PR_MAP_RECALC)
- recalc_server_map(px);
-
- /* The GCD which was computed causes the total effective
- * weight to appear lower than all weights. Let's
- * recompute it.
- */
- if (px->srv && px->srv->eweight)
- gcd = px->srv->uweight / px->srv->eweight;
-
if (flags & STAT_FMT_HTML) {
chunk_printf(&msg, sizeof(trash),
/* name */
@@ -883,8 +871,9 @@
px->failed_conns, px->failed_resp,
px->retries, px->redispatches,
human_time(now.tv_sec - px->last_change, 1),
- (px->srv_map_sz > 0 || !px->srv) ? "UP" : "DOWN",
- px->srv_map_sz * gcd, px->srv_act, px->srv_bck);
+ (px->lbprm.tot_weight > 0 || !px->srv) ? "UP" : "DOWN",
+ px->lbprm.tot_weight * px->lbprm.wmult,
+ px->srv_act, px->srv_bck);
chunk_printf(&msg, sizeof(trash),
/* rest of backend: nothing, down transformations, total downtime */
@@ -928,10 +917,11 @@
px->denied_req, px->denied_resp,
px->failed_conns, px->failed_resp,
px->retries, px->redispatches,
- (px->srv_map_sz > 0 || !px->srv) ? "UP" : "DOWN",
- px->srv_map_sz * gcd, px->srv_act, px->srv_bck,
+ (px->lbprm.tot_weight > 0 || !px->srv) ? "UP" : "DOWN",
+ px->lbprm.tot_weight * px->lbprm.wmult,
+ px->srv_act, px->srv_bck,
px->down_trans, now.tv_sec - px->last_change,
- px->srv?be_downtime(px):0,
+ px->srv?be_downtime(px):0,
relative_pid, px->uuid);
}
if (buffer_write_chunk(rep, &msg) != 0)