* Idem for zabbix-agent.

svn path=/nixos/trunk/; revision=20040

modules/services/monitoring/zabbix-agent.nix

@@ -76,20 +76,29 @@ in
           ''
             mkdir -m 0755 -p ${stateDir} ${logDir}
             chown zabbix ${stateDir} ${logDir}
-
-            # Grrr, zabbix_agentd cannot be properly monitored by
-            # Upstart.  Upstart's "expect fork/daemon" feature doesn't
-            # work because zabbix_agentd runs some programs on
-            # startup, and zabbix_agentd doesn't have a flag to
-            # prevent daemonizing.
-            export PATH=${pkgs.nettools}/bin:$PATH
-            ${pkgs.zabbix.agent}/sbin/zabbix_agentd --config ${configFile}
           '';
 
+        # Zabbix doesn't have an option not to daemonize, and doesn't
+        # daemonize in a way that allows Upstart to track it.  So to
+        # make sure that we notice when it goes down, we start Zabbix
+        # with an open connection to a fifo, with a `cat' on the other
+        # side.  If Zabbix dies, then `cat' will exit as well, so we
+        # just monitor `cat'.
+        script =
+          ''
+            export PATH=${pkgs.nettools}/bin:$PATH
+            rm -f ${stateDir}/dummy2
+            mkfifo ${stateDir}/dummy2
+            cat ${stateDir}/dummy2 &
+            pid=$!
+            ${pkgs.zabbix.agent}/sbin/zabbix_agentd --config ${configFile} 100>${stateDir}/dummy2
+            wait "$pid"
+          '';
+        
         postStop =
           ''
-            pid=$(cat ${pidFile})
+            pid=$(cat ${pidFile} 2> /dev/null || true)
-            test -n "$pid" && kill "$pid"
+            (test -n "$pid" && kill "$pid") || true
             # Wait until they're really gone.
             while ${pkgs.procps}/bin/pgrep -u zabbix zabbix_agentd > /dev/null; do sleep 1; done
           '';