1

I am using MMS to monitor a 13 node MongoDB Cluster(3 Config Servers, 4 MongoS, 2 replica sets with 2 members + 1 arbiter each). MMS was working properly since its installation, but now it is not monitoring the shards & replica-sets properly.

It is monitoring only 4 nodes(1 Primary 1 Secondary and 2 arbiters) out of 6 nodes and auto-removing the other replica-set(primary + secondary) every time I am adding them manually. Also for the replica-set it is monitoring it is taking the primary secondary and arbiter as independent nodes(even though it is showing correct TYPE for them in the monitoring window).

When I checked Host Mapping it was surprised to see that it was showing the same alias name for both the Primary nodes and same for both the secondaries. I guess this is the reason it is not showing those 2 nodes in the monitoring window.

I manually removed those entries of same alias names from the Host Mapping window but it is auto-updating it after few minutes.

Why is it considering the two different nodes as same and taking the same alias name for them ? Is it a bug ?

MMS is working well for Config-servers and MongoS nodes.

My MMS group is "MMS@Ork"

My Data nodes are as follows:

Server1 -> Primary of Replica-set1 + Secondary of Replica-set2(2 Nodes) Server2 -> Primary of Replica-set2 + Secondary of Replica-set1(2 Nodes) Arbiters of both replica-sets are on separate machines. I have kept one port number common for both the primaries and other port number common for both the secondaries.

Version of all the nodes in the cluster is 2.4.9 and that of MMS monitoring-agent is 2.3.1

enter image description here

enter image description here

2 Answers 2

0

From the MMS Dashboard; what is the MMS Agent Log saying?

enter image description here

Have you looked in the mms-monitoring agent log locally on the server?

/var/log/mongodb-mms/monitoring-agent.log

Update

Please add a screen shot of how you are configuring the host

enter image description here

Update 14:02

Can you ping S143, S144 and mongo-client2 from the server where the mms-agent is installed?

Can you telnet to those servers from the mms-agent machine?

ex

telnet S143 37007
telnet S144 37007
telnet mongo-client2 37007

If you can't, then perhaps there is a network problem between the server with the MMS agent and the three mongods you listed above.

10
  • Yes I checked the monitoring-agent logs. Logs are OK and not showing any error. But when I am adding the missing host through add host option it is auto removing it after some time with the following message in the logs: "[2014/07/30 14:17:45] [monitoring.routinemanager.info] [monitoring-agent/components/bus.go:53] Shutting down monitoring to cluster-datanode1:37010: no longer present [2014/07/30 14:17:45] [monitoring.process-cluster-datanode1:37010.info] [monitoring-agent/components/bus.go:153] Received quit to shutdown all 3 monitors" Commented Jul 30, 2014 at 9:09
  • In the Host Mapping it is automatically assigning the name of this removed node as the alias name to the other node. In this way it is considering the two nodes as same i guess. Commented Jul 30, 2014 at 9:14
  • Hello @user2710961, have you tried reinstalling the latest agent? I know this may not be possible, but if you remove all hosts and mongos and configuration information from MMS, then add back cluster-datanode1:37010? Can you also add a screen shot of the add host configuration? I added an example above Commented Jul 30, 2014 at 9:16
  • Hi Craig, I have updated monitoring agent to the latest version 2.4.0. Then removed all the hosts and mongos from the MMS. Then I added one mongos to the MMS using the add host option. After few minutes MMS auto added another mongos but with alert message as shown in the screenshot I have pasted now with my question. I have waited for more then half an hour but no other node is auto added to the MMS yet. Also the cluster name it is taking is incorrect. Commented Jul 30, 2014 at 10:09
  • @user2710961, Dumb question maybe, but is the mms agent running on the mongo 37007 host? Commented Jul 30, 2014 at 10:15
0

The issue was only one replica-set was being monitored by the MMS and second one was not. The issue is resolved now. The problem was the corresponding nodes of these replica-sets were given more than one host names in /etc/hosts file and out of them one host name was common between them which was also their default host name given in /etc/hostname file. This was making MMS to consider them as a single node, therefore it was using the same alias names for them and showing only one of them in the monitoring window.

These names were assigned at the time of cluster creation and was not having any issue until I updated the version of MMS monitoring agent to 2.3.1 and then 2.4.0. I guess this may be because of some extra features in the new versions.

I have changed the host name which was common in the nodes and the MMS is working fine now.

Thanks @CraigEfrein for your support.

Thanks & Regards

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.