Action "slavelag" for master-slave replication (legacy)

[heuri]

Added the action "slavelag" to determine the lag behind the master of a master slave replication (legacy).

Currently the command rs.printSlaveReplicationInfo() is not supported by pymongo, so we determine the lag by using ('serverStatus', 1), ('repl', 2)

[vkhatri]

@heuri thank you for adding it to the plugin. how about adding perf data to it? this check throws an exception if missing lagSeconds, perhaps print the node type primary/arbi etc. in the message?

patch:

--- check_mongodb.py    2015-05-30 05:52:17.603794569 +0000
+++ check_mongodb.py.me 2015-05-30 05:47:36.707526065 +0000
@@ -253,7 +253,7 @@
     elif action == "replset_quorum":
         return check_replset_quorum(con, perf_data)
     elif action == "slavelag":
-        return check_slavelag(con, warning, critical)
+        return check_slavelag(con, warning, critical, perf_data)
     else:
         return check_connect(host, port, warning, critical, perf_data, user, passwd, conn_time)

@@ -1237,7 +1237,7 @@
         primary_status = 1
     return check_levels(primary_status, warning, critical, message)

-def check_slavelag(con, warning, critical):
+def check_slavelag(con, warning, critical, perf_data):
     warning = warning or 30
     critical = critical or 60

@@ -1251,11 +1251,12 @@
         try:
           seconds = int(data['repl']['sources'][0]['lagSeconds'])
           master = data['repl']['sources'][0]['host']
+          message = "Master Slave Replication " + str(seconds) + " secs behind '" + master + "' (master)."
         except (KeyError, IndexError):
-          raise Exception("Missing lagSeconds-Value, not running with --slave?")
-
-        message = "Master Slave Replication " + str(seconds) + " secs behind '" + master + "' (master)."
+          seconds = 0
+          message = "Missing lagSeconds-Value, not running with --slave or connected to a master/primary node"

+        message += performance_data(perf_data, [(seconds, "slavelag", warning, critical)])
         return check_levels(seconds, warning, critical, message)

     except Exception, e:

[heuri]

@vkhatri great idea, works like a charm. Yes we could determine the node status by checking con.admin.command("replSetGetStatus"), but here we need also then a version switch between >= 2.0 and < 2.0.