HAProxy patch against 1.6.7 to control the backup flag of a server using state file, and have the ability to save global server-state-file

haproxy-1.6.7-backup_state_file.patch

# This patch lets you:
# - put server/backend in backup mode, or active mode; "backup" parameter for server isn't needed in the configuration file. 
# - save global server state to 'server-state-file' (absolute path required)

# see the UI here: http://i.imgur.com/n2O5V6P.png

# Instructions
$ wget http://www.haproxy.org/download/1.6/src/haproxy-1.6.7.tar.gz
$ tar xf haproxy-1.6.7.tar.gz
$ cd haproxy-1.6.7
$ cat below_patch | patch -p1
$ make USE_PCRE=1 DEBUG="" ARCH=$(uname -m) TARGET=linux26
$ sudo make install

/etc/haproxy/haproxy.cfg:
global
  ...
  user haproxy
  group haproxy
  stats socket /etc/haproxy/haproxy.sock level admin
  server-state-file /etc/haproxy/haproxy.state
  ...

defaults
  ...
  load-server-state-from-file global
  ...

# make sure the state file is there and it's owned by user haproxy
$ sudo touch /etc/haproxy/haproxy.state
$ sudo chown haproxy.haproxy /etc/haproxy/haproxy.state

# put myserver in mybackend into backup mode
$ echo "set server mybackend/myserver state backup" | socat unix-connect:/etc/haproxy/haproxy.sock stdio
# or make it active again
$ echo "set server mybackend/myserver state active" | socat unix-connect:/etc/haproxy/haproxy.sock stdio
# save haproxy state
$ echo "show servers state" | socat unix-connect:/etc/haproxy/haproxy.sock stdio > /etc/haproxy/haproxy.state

# -----------------------------------------------------------------------------------------------------------------#

diff -urN a/doc/management.txt b/doc/management.txt
--- a/doc/management.txt	2016-07-13 17:57:01.000000000 +0000
+++ b/doc/management.txt	2016-07-26 23:04:28.323646629 +0000
@@ -1586,6 +1586,7 @@
                                   configuration.
      srv_f_forced_id:             Flag to know if the server's ID is forced by
                                   configuration.
+     srv_is_backup:               Flag to know if the server's in backup mode

 show sess
   Dump all known sessions. Avoid doing this on slow connections as this can
diff -urN a/include/types/proto_http.h b/include/types/proto_http.h
--- a/include/types/proto_http.h	2016-07-13 17:57:01.000000000 +0000
+++ b/include/types/proto_http.h	2016-07-29 00:11:51.373705211 +0000
@@ -263,6 +263,8 @@
 	STAT_STATUS_NONE,	/* nothing happened (no action chosen or servers state didn't change) */
 	STAT_STATUS_PART,	/* the action is partially successful */
 	STAT_STATUS_UNKN,	/* an unknown error occured, shouldn't happen */
+	STAT_STATUS_SAVE_OK,	/* global save-state save successful */
+	STAT_STATUS_SAVE_ERR,	/* global save-state save failed */
 	STAT_STATUS_SIZE
 };

diff -urN a/include/types/server.h b/include/types/server.h
--- a/include/types/server.h	2016-07-13 17:57:01.000000000 +0000
+++ b/include/types/server.h	2016-07-26 23:04:28.323646629 +0000
@@ -86,7 +86,7 @@
 #define SRV_STATE_FILE_VERSION 1
 #define SRV_STATE_FILE_VERSION_MIN 1
 #define SRV_STATE_FILE_VERSION_MAX 1
-#define SRV_STATE_FILE_FIELD_NAMES "be_id be_name srv_id srv_name srv_addr srv_op_state srv_admin_state srv_uweight srv_iweight srv_time_since_last_change srv_check_status srv_check_result srv_check_health srv_check_state srv_agent_state bk_f_forced_id srv_f_forced_id"
+#define SRV_STATE_FILE_FIELD_NAMES "be_id be_name srv_id srv_name srv_addr srv_op_state srv_admin_state srv_uweight srv_iweight srv_time_since_last_change srv_check_status srv_check_result srv_check_health srv_check_state srv_agent_state bk_f_forced_id srv_f_forced_id srv_is_backup"
 #define SRV_STATE_FILE_MAX_FIELDS 18
 #define SRV_STATE_FILE_NB_FIELDS_VERSION_1 18
 #define SRV_STATE_LINE_MAXLEN 512
diff -urN a/src/dumpstats.c b/src/dumpstats.c
--- a/src/dumpstats.c	2016-07-13 17:57:01.000000000 +0000
+++ b/src/dumpstats.c	2016-07-29 18:32:56.605201609 +0000
@@ -23,6 +23,7 @@
 #include <sys/socket.h>
 #include <sys/stat.h>
 #include <sys/types.h>
+#include <sys/file.h>

 #include <common/cfgparse.h>
 #include <common/compat.h>
@@ -117,6 +118,10 @@
 	ST_ADM_ACTION_ARUNN,
 	ST_ADM_ACTION_ADOWN,

+	/* backup/active states */
+	ST_ADM_ACTION_BACKUP,
+	ST_ADM_ACTION_ACTIVE,
+
 	/* set admin state */
 	ST_ADM_ACTION_READY,
 	ST_ADM_ACTION_DRAIN,
@@ -1529,6 +1534,20 @@
 					srv_adm_set_drain(sv);
 				else if (strcmp(args[4], "maint") == 0)
 					srv_adm_set_maint(sv);
+				else if (strcmp(args[4], "backup") == 0) {
+					if (!(sv->flags & SRV_F_BACKUP)) {
+						sv->flags |= SRV_F_BACKUP;
+						sv->proxy->srv_act --;
+						sv->proxy->srv_bck ++;
+					}
+				}
+				else if (strcmp(args[4], "active") == 0) {
+					if (sv->flags & SRV_F_BACKUP) {
+						sv->flags &= ~SRV_F_BACKUP;
+						sv->proxy->srv_act ++;
+						sv->proxy->srv_bck --;
+					}
+				}
 				else {
 					appctx->ctx.cli.msg = "'set server <srv> state' expects 'ready', 'drain' and 'maint'.\n";
 					appctx->st0 = STAT_CLI_PRINT;
@@ -2749,6 +2768,7 @@
 	char srv_addr[INET6_ADDRSTRLEN + 1];
 	time_t srv_time_since_last_change;
 	int bk_f_forced_id, srv_f_forced_id;
+	int srv_is_backup;


 	/* we don't want to report any state if the backend is not enabled on this process */
@@ -2764,6 +2784,7 @@
 		srv_time_since_last_change = 0;
 		bk_f_forced_id = 0;
 		srv_f_forced_id = 0;
+		srv_is_backup = 0;

 		switch (srv->addr.ss_family) {
 			case AF_INET:
@@ -2778,19 +2799,20 @@
 		srv_time_since_last_change = now.tv_sec - srv->last_change;
 		bk_f_forced_id = appctx->ctx.server_state.px->options & PR_O_FORCED_ID ? 1 : 0;
 		srv_f_forced_id = srv->flags & SRV_F_FORCED_ID ? 1 : 0;
+		srv_is_backup = srv->flags & SRV_F_BACKUP ? 1 : 0;

 		chunk_appendf(buf,
 				"%d %s "
 				"%d %s %s "
 				"%d %d %d %d %ld "
 				"%d %d %d %d %d "
-				"%d %d"
+				"%d %d %d"
 				"\n",
 				appctx->ctx.server_state.px->uuid, appctx->ctx.server_state.px->id,
 				srv->puid, srv->id, srv_addr,
 				srv->state, srv->admin, srv->uweight, srv->iweight, (long int)srv_time_since_last_change,
 				srv->check.status, srv->check.result, srv->check.health, srv->check.state, srv->agent.state,
-				bk_f_forced_id, srv_f_forced_id);
+				bk_f_forced_id, srv_f_forced_id, srv_is_backup);
 		if (bi_putchk(si_ic(si), &trash) == -1) {
 			si_applet_cant_put(si);
 			return 0;
@@ -2885,6 +2907,67 @@
 	return 1;
 }

+/* Parses backend list and dumps the state to global server state file
+ */
+static int stats_save_state_to_file(struct stream_interface *si)
+{
+	FILE *fp;
+	struct appctx *appctx = __objt_appctx(si->end);
+	char globalfilepath[MAXPATHLEN + 1];
+	int len, errno, globalfilepathlen;
+	extern struct proxy *proxy;
+	struct proxy *curproxy;
+	struct chunk *mybuf;
+
+	if (!global.server_state_file) {
+		return 0;
+	}
+
+	if (global.server_state_file[0] == '/') {
+		len = strlen(global.server_state_file);
+		if (len > MAXPATHLEN) {
+			return 0;
+		}
+		memcpy(globalfilepath, global.server_state_file, len);
+		globalfilepath[len] = '\0';
+		globalfilepathlen = len;
+	} else {
+		return 0;
+	}
+
+	fp = fopen(globalfilepath, "w");
+	if(errno || !fp) {
+		return 0;
+	}
+
+	if(flock(fileno(fp), LOCK_EX) == -1) {
+		fclose(fp);
+		return 0;
+	}
+
+        mybuf = get_trash_chunk();
+	chunk_reset(mybuf);
+
+	chunk_appendf(mybuf, "%d\n# %s\n", SRV_STATE_FILE_VERSION, SRV_STATE_FILE_FIELD_NAMES);
+
+	appctx->ctx.server_state.px = proxy;
+
+	for (; appctx->ctx.server_state.px != NULL; appctx->ctx.server_state.px = curproxy->next) {
+		curproxy = appctx->ctx.server_state.px;
+		if (curproxy->cap & PR_CAP_BE) {
+			if (!dump_servers_state(si, mybuf))
+				return 0;
+		}
+	}
+
+	fprintf(fp, "%s", mybuf->str);
+
+	flock(fileno(fp), LOCK_UN);
+	fclose(fp);
+
+	return 1;
+}
+
 /* This function dumps memory usage information onto the stream interface's
  * read buffer. It returns 0 as long as it does not complete, non-zero upon
  * completion. No state is used.
@@ -4081,19 +4164,31 @@
 			      "Choose the action to perform on the checked servers : "
 			      "<select name=action>"
 			      "<option value=\"\"></option>"
-			      "<option value=\"ready\">Set state to READY</option>"
-			      "<option value=\"drain\">Set state to DRAIN</option>"
-			      "<option value=\"maint\">Set state to MAINT</option>"
-			      "<option value=\"dhlth\">Health: disable checks</option>"
-			      "<option value=\"ehlth\">Health: enable checks</option>"
-			      "<option value=\"hrunn\">Health: force UP</option>"
-			      "<option value=\"hnolb\">Health: force NOLB</option>"
-			      "<option value=\"hdown\">Health: force DOWN</option>"
-			      "<option value=\"dagent\">Agent: disable checks</option>"
-			      "<option value=\"eagent\">Agent: enable checks</option>"
-			      "<option value=\"arunn\">Agent: force UP</option>"
-			      "<option value=\"adown\">Agent: force DOWN</option>"
+			      "<optgroup label=\"OP states\">"
+			      "<option value=\"ready\">Set OP state to READY</option>"
+			      "<option value=\"drain\">Set OP state to DRAIN</option>"
+			      "<option value=\"maint\">Set OP state to MAINT</option>"
+			      "</optgroup>"
+			      "<optgroup label=\"BACKUP states\">"
+			      "<option value=\"backup\">Set BACKUP state to ON</option>"
+			      "<option value=\"active\">Set BACKUP state to OFF</option>"
+			      "</optgroup>"
+			      "<optgroup label=\"HEALTH actions\">"
+			      "<option value=\"dhlth\">HEALTH: disable checks</option>"
+			      "<option value=\"ehlth\">HEALTH: enable checks</option>"
+			      "<option value=\"hrunn\">HEALTH: force UP</option>"
+			      "<option value=\"hnolb\">HEALTH: force NOLB</option>"
+			      "<option value=\"hdown\">HEALTH: force DOWN</option>"
+			      "</optgroup>"
+			      "<optgroup label=\"AGENT actions\">"
+			      "<option value=\"dagent\">AGENT: disable checks</option>"
+			      "<option value=\"eagent\">AGENT: enable checks</option>"
+			      "<option value=\"arunn\">AGENT: force UP</option>"
+			      "<option value=\"adown\">AGENT: force DOWN</option>"
+			      "</optgroup>"
+			      "<optgroup label=\"OTHER actions\">"
 			      "<option value=\"shutdown\">Kill Sessions</option>"
+			      "</optgroup>"
 			      "</select>"
 			      "<input type=\"hidden\" name=\"b\" value=\"#%d\">"
 			      "&nbsp;<input type=\"submit\" value=\"Apply\">"
@@ -4492,7 +4587,29 @@
 	              "<b>system limits:</b> memmax = %s%s; ulimit-n = %d<br>\n"
 	              "<b>maxsock = </b> %d; <b>maxconn = </b> %d; <b>maxpipes = </b> %d<br>\n"
 	              "current conns = %d; current pipes = %d/%d; conn rate = %d/sec<br>\n"
-	              "Running tasks: %d/%d; idle = %d %%<br>\n"
+	              "Running tasks: %d/%d; idle = %d %%<br>\n",
+	              (uri->flags & ST_HIDEVER) ? "" : (STATS_VERSION_STRING),
+	              pid, (uri->flags & ST_SHNODE) ? " on " : "",
+		      (uri->flags & ST_SHNODE) ? (uri->node ? uri->node : global.node) : "",
+	              (uri->flags & ST_SHDESC) ? ": " : "",
+		      (uri->flags & ST_SHDESC) ? (uri->desc ? uri->desc : global.desc) : "",
+	              pid, relative_pid, global.nbproc,
+	              up / 86400, (up % 86400) / 3600,
+	              (up % 3600) / 60, (up % 60),
+	              global.rlimit_memmax ? ultoa(global.rlimit_memmax) : "unlimited",
+	              global.rlimit_memmax ? " MB" : "",
+	              global.rlimit_nofile,
+	              global.maxsock, global.maxconn, global.maxpipes,
+	              actconn, pipes_used, pipes_used+pipes_free, read_freq_ctr(&global.conn_per_sec),
+	              run_queue_cur, nb_tasks_cur, idle_pct
+	              );
+
+	if(global.server_state_file) {
+		chunk_appendf(&trash,
+		              "<form method=\"POST\"><input type=\"hidden\" name=\"w\" value=\"y\"><input type=\"submit\" value=\"Save global server state to %s\"></form><br>\n", global.server_state_file);
+	}
+
+	chunk_appendf(&trash,
 	              "</td><td align=\"center\" nowrap>\n"
 	              "<table class=\"lgd\"><tr>\n"
 	              "<td class=\"active_up\">&nbsp;</td><td class=\"noborder\">active UP </td>"
@@ -4514,23 +4631,8 @@
 	              "Note: \"NOLB\"/\"DRAIN\" = UP with load-balancing disabled."
 	              "</td>"
 	              "<td align=\"left\" valign=\"top\" nowrap width=\"1%%\">"
-	              "<b>Display option:</b><ul style=\"margin-top: 0.25em;\">"
-	              "",
-	              (uri->flags & ST_HIDEVER) ? "" : (STATS_VERSION_STRING),
-	              pid, (uri->flags & ST_SHNODE) ? " on " : "",
-		      (uri->flags & ST_SHNODE) ? (uri->node ? uri->node : global.node) : "",
-	              (uri->flags & ST_SHDESC) ? ": " : "",
-		      (uri->flags & ST_SHDESC) ? (uri->desc ? uri->desc : global.desc) : "",
-	              pid, relative_pid, global.nbproc,
-	              up / 86400, (up % 86400) / 3600,
-	              (up % 3600) / 60, (up % 60),
-	              global.rlimit_memmax ? ultoa(global.rlimit_memmax) : "unlimited",
-	              global.rlimit_memmax ? " MB" : "",
-	              global.rlimit_nofile,
-	              global.maxsock, global.maxconn, global.maxpipes,
-	              actconn, pipes_used, pipes_used+pipes_free, read_freq_ctr(&global.conn_per_sec),
-	              run_queue_cur, nb_tasks_cur, idle_pct
-	              );
+	              "<b>Options:</b><ul style=\"margin-top: 0.25em;\">");
+

 	/* scope_txt = search query, appctx->ctx.stats.scope_len is always <= STAT_SCOPE_TXT_MAXLEN */
 	memcpy(scope_txt, bo_ptr(si_ob(si)) + appctx->ctx.stats.scope_str, appctx->ctx.stats.scope_len);
@@ -4676,6 +4778,26 @@
 			              (appctx->ctx.stats.flags & STAT_NO_REFRESH) ? ";norefresh" : "",
 			              scope_txt);
 			break;
+		case STAT_STATUS_SAVE_OK:
+			chunk_appendf(&trash,
+			              "<p><div class=active_up>"
+			              "<a class=lfsb href=\"%s%s%s%s\" title=\"Remove this message\">[X]</a> "
+			              "Global server state saved to '%s'"
+			              "</div>\n", uri->uri_prefix,
+			              (appctx->ctx.stats.flags & STAT_HIDE_DOWN) ? ";up" : "",
+			              (appctx->ctx.stats.flags & STAT_NO_REFRESH) ? ";norefresh" : "",
+			              scope_txt, global.server_state_file);
+			break;
+		case STAT_STATUS_SAVE_ERR:
+			chunk_appendf(&trash,
+			              "<p><div class=active_down>"
+			              "<a class=lfsb href=\"%s%s%s%s\" title=\"Remove this message\">[X]</a> "
+			              "Global server state could not be saved to '%s'"
+			              "</div>\n", uri->uri_prefix,
+			              (appctx->ctx.stats.flags & STAT_HIDE_DOWN) ? ";up" : "",
+			              (appctx->ctx.stats.flags & STAT_NO_REFRESH) ? ";norefresh" : "",
+			              scope_txt, global.server_state_file);
+			break;
 		default:
 			chunk_appendf(&trash,
 			              "<p><div class=active_no_check>"
@@ -4873,8 +4995,17 @@
 			if (url_decode(key) < 0 || url_decode(value) < 0)
 				break;

+			/* check if we're asked to save the state */
+			if (strcmp(key, "w") == 0) {
+				if (stats_save_state_to_file(si)) {
+					appctx->ctx.stats.st_code = STAT_STATUS_SAVE_OK;
+				} else {
+					appctx->ctx.stats.st_code = STAT_STATUS_SAVE_ERR;
+				}
+				goto out;
+			}
 			/* Now we can check the key to see what to do */
-			if (!px && (strcmp(key, "b") == 0)) {
+			else if (!px && (strcmp(key, "b") == 0)) {
 				if ((px = proxy_be_by_name(value)) == NULL) {
 					/* the backend name is unknown or ambiguous (duplicate names) */
 					appctx->ctx.stats.st_code = STAT_STATUS_ERRP;
@@ -4891,6 +5022,12 @@
 				else if (strcmp(value, "maint") == 0) {
 					action = ST_ADM_ACTION_MAINT;
 				}
+				else if (strcmp(value, "backup") == 0) {
+					action = ST_ADM_ACTION_BACKUP;
+				}
+				else if (strcmp(value, "active") == 0) {
+					action = ST_ADM_ACTION_ACTIVE;
+				}
 				else if (strcmp(value, "shutdown") == 0) {
 					action = ST_ADM_ACTION_SHUTDOWN;
 				}
@@ -5064,6 +5201,24 @@
 						altered_servers++;
 						total_servers++;
 						break;
+					case ST_ADM_ACTION_BACKUP:
+						if (!(sv->flags & SRV_F_BACKUP)) {
+							sv->flags |= SRV_F_BACKUP;
+							px->srv_act --;
+							px->srv_bck ++;
+							altered_servers ++;
+							total_servers ++;
+						}
+						break;
+					case ST_ADM_ACTION_ACTIVE:
+						if (sv->flags & SRV_F_BACKUP) {
+							sv->flags &= ~SRV_F_BACKUP;
+							px->srv_act ++;
+							px->srv_bck --;
+							altered_servers ++;
+							total_servers ++;
+						}
+						break;
 					case ST_ADM_ACTION_SHUTDOWN:
 						if (px->state != PR_STSTOPPED) {
 							struct stream *sess, *sess_bck;
diff -urN a/src/proto_http.c b/src/proto_http.c
--- a/src/proto_http.c	2016-07-13 17:57:01.000000000 +0000
+++ b/src/proto_http.c	2016-07-29 17:09:13.824000318 +0000
@@ -233,6 +233,8 @@
 	[STAT_STATUS_NONE] = "NONE",
 	[STAT_STATUS_PART] = "PART",
 	[STAT_STATUS_UNKN] = "UNKN",
+	[STAT_STATUS_SAVE_OK] = "SVOK",
+	[STAT_STATUS_SAVE_ERR] = "SVER"
 };


diff -urN a/src/server.c b/src/server.c
--- a/src/server.c	2016-07-13 17:57:01.000000000 +0000
+++ b/src/server.c	2016-07-26 23:04:28.325646629 +0000
@@ -1918,6 +1918,7 @@
 	int srv_check_state, srv_agent_state;
 	int bk_f_forced_id;
 	int srv_f_forced_id;
+	int srv_is_backup;

 	msg = get_trash_chunk();
 	switch (version) {
@@ -1937,6 +1938,7 @@
 			 * srv_agent_state:      params[10]
 			 * bk_f_forced_id:       params[11]
 			 * srv_f_forced_id:      params[12]
+			 * srv_is_backup:        params[13]
 			 */

 			/* validating srv_op_state */
@@ -2045,6 +2047,13 @@
 			if (p == params[12] || errno == EINVAL || errno == ERANGE || !((srv_f_forced_id == 0) || (srv_f_forced_id == 1)))
 				chunk_appendf(msg, ", invalid srv_f_forced_id value '%s'", params[12]);

+			/* validating srv_is_backup */
+			p = NULL;
+			errno = 0;
+			srv_is_backup = strtol(params[13], &p, 10);
+			if (p == params[13] || errno == EINVAL || errno == ERANGE || !((srv_is_backup == 0) || (srv_is_backup == 1)))
+				chunk_appendf(msg, ", invalid srv_is_backup value '%s'", params[13]);
+

 			/* don't apply anything if one error has been detected */
 			if (msg->len)
@@ -2077,6 +2086,20 @@
 			 *   state is different from new configuration state
 			 */
 			/* configuration has changed */
+			if (srv_is_backup) {
+				if (!(srv->flags & SRV_F_BACKUP)) {
+					srv->flags |= SRV_F_BACKUP;
+					srv->proxy->srv_act --;
+					srv->proxy->srv_bck ++;
+				}
+			}
+			else {
+				if (srv->flags & SRV_F_BACKUP) {
+					srv->flags &= ~SRV_F_BACKUP;
+					srv->proxy->srv_act ++;
+					srv->proxy->srv_bck --;
+				}
+			}
 			if ((srv_admin_state & SRV_ADMF_CMAINT) != (srv->admin & SRV_ADMF_CMAINT)) {
 				if (srv->admin & SRV_ADMF_CMAINT)
 					srv_adm_set_maint(srv);