is it ok to start an mms agent on all mongo servers

100 views
Skip to first unread message

Grégoire Seux

unread,
May 7, 2013, 4:03:09 AM5/7/13
to mongod...@googlegroups.com
Hello,

I'd like to know if anybody has already started a mms agent on each
server of a cluster (45 nodes) and if there are issues arising.

mms doc states
>only one agent per group or environment will report to MMS at a time, but you can run multiple instances of the agent to provide redundancy.

but it is unclear if a lot of agent will cause trouble

maybe someone working in mms dev team?

The objective is to make all servers alike and thus minimizing custom
"role" in the cluster (all server have a mongoS for instance).

thanks

--
Grégoire

Sam Millman

unread,
May 7, 2013, 4:28:44 AM5/7/13
to mongod...@googlegroups.com
I am not a mms dev or whatever however you shouldn't need to put mms on each replica/shard of a cluster. You should only need to put it on a server that can see the cluster through mongodb and it should just propogate the entire cluster to mms.



--
--
You received this message because you are subscribed to the Google
Groups "mongodb-user" group.
To post to this group, send email to mongod...@googlegroups.com
To unsubscribe from this group, send email to
mongodb-user...@googlegroups.com
See also the IRC channel -- freenode.net#mongodb

---
You received this message because you are subscribed to the Google Groups "mongodb-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mongodb-user...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.



Grégoire Seux

unread,
May 13, 2013, 2:25:24 AM5/13/13
to mongod...@googlegroups.com


On Tuesday, May 7, 2013 10:28:44 AM UTC+2, Sammaye wrote:
I am not a mms dev or whatever however you shouldn't need to put mms on each replica/shard of a cluster. You should only need to put it on a server that can see the cluster through mongodb and it should just propogate the entire cluster to mms.

Thanks for this answer.
I am aware I don't need to have one everywhere but it would drastically simplify the setup to have mms agent on each host.

I'd like an argumentative answer.

-- 
Grégoire

Stephen Steneker

unread,
May 13, 2013, 3:31:22 AM5/13/13
to mongod...@googlegroups.com

On Monday, May 13, 2013 4:25:24 PM UTC+10, Grégoire Seux wrote:
I am aware I don't need to have one everywhere but it would drastically simplify the setup to have mms agent on each host.

Hi Grégoire,

You can certainly have multiple MMS agents, but only one will be actively reporting (the others act as hot standbys):

I'm not aware of any limitations or caveats on multiple agents, but having 45 identical hosts seems excessive.

Within that group of hosts there must be some further roles such as replSet membership that would allow you to limit the number of agents running.  With that many hosts I would expect you are using an automated tool for deployment which would make this more manageable.

Regards,
Stephen

Grégoire Seux

unread,
May 13, 2013, 3:46:34 AM5/13/13
to mongod...@googlegroups.com
Thanks for your answer.
Yes indeed, we use chef to manage our infrstructure and it allows to
specify specific conf for specific hosts. Having specific conf however
increases complexity that I wanted to avoid.
I have solved the issue by deploying mms agent on each mongoconf host
(or each member of replicaset for non sharded cluster).


--
Grégoire

sam

unread,
May 13, 2013, 11:51:17 AM5/13/13
to mongod...@googlegroups.com
I don't believe having multiple MMS agents is fixing your issue.  Can you explain what you are trying to solve for?  You should only need one MMS agent to satisfy monitoring.

Grégoire Seux

unread,
May 13, 2013, 12:00:08 PM5/13/13
to mongod...@googlegroups.com
On Mon, May 13, 2013 at 5:51 PM, sam <samnsa...@gmail.com> wrote:
> aving multiple MMS agents is fixing your issue. Can you explain what you
> are trying to solve for? You should only need one MMS agent to satisfy
> monitoring.


It allows to have some redundancy, in case one agent crash (for
instance because the host it runs crash) or is stopped for
maintenance.

--
Grégoire
Reply all
Reply to author
Forward
0 new messages