2

Does anybody know how to configure Nagios to monitor MS Exchange mail stores?

We have had mailstores unmount (because of diskspace, and other causes), and would like to monitor their state (mounted or not) via nagios.

We use pNSClient.exe on our windows servers, which can monitor processes and services - but apparently Exchange simply uses one process (store.exe) for all mail stores - and so monitoring store.exe will not be helpful.

Improve this question
0

3 Answers 3

Reset to default
2

As you stated, because Exchange uses the store.exe process for all mailbox databases, it is not possible to monitor individual mailbox database that way. There may be ways to do so by talking directley to Exchange, but I don't know of any process for Nagios that can do that.

The store.exe process is the Exchange Information Store service, this can be monitored by Nagios. If the mailbox databases dismounts because of diskspace issues, and many other issues it is likely this server will have stopped as well. Monitoring this should give you a good indication of mailbox issues

This article is an excellent tutorial on setting up monitoring for Exchange services and includes a lot of info on the different things you can check to monitor your Exchange servers health.

[Edit]

Windows Performance Counters does have a counter for the Exchange store, the MSExchangeIS Mailbox, counter so this may also be able to be used.

Improve this answer
4
  • Do you know if the mailstores can be monitored via "Windows Performance Counters"? check_nt supports them, but I'm not sure whether mailstore status can be accessed that way.
    – Brent
    May 4, 2009 at 17:51
  • Sure, see edit above
    – Sam Cogan
    May 4, 2009 at 17:59
  • Thanks Sam - unfortunately the SMTPNTFSStoreDriver counter doesn't appear to have instances for our individual mailstores. MSExchangeIS Mailbox, however, does.
    – Brent
    May 4, 2009 at 23:02
  • Good to know, not used the performance counter for Exchange before so wasn't sure which one did it. Have updated to reflect that.
    – Sam Cogan
    May 4, 2009 at 23:10
2

I have found a way to check how many active connections there are to a particular mailstore using the check_nt -v COUNTER check as follows:

check_nt -H $HOSTNAME$ -p $PORT$ -v COUNTER -l "\\MSExchangeIS Mailbox($ARG1$)\\Active Client Logons","$ARG1$ Sessions: %.f" -w0 -c0

I am assuming that there will be at least 1 Active Client Login from a system process as long as the mailstore is mounted, but that it will drop to 0 if it is unmounted.

I will test this later, and let you know.

edit

Active Client Logins dropped to 0 overnight for some of the mailstores, so this is not the solution - however, I now wonder whether Client Logins will work as described.

Testing again...

edit

Using Client Logins does exactly what I want.

Improve this answer
0

This might be my unpopular opinion but, monitoring fire-fighting-style seems like a lot of work... (but much better than not monitoring at all). Use a monitoring tool that has built-in intelligence from the developers for Exchange or whatever service needs monitoring which can discover problems you cannot define.

The one in the question is most likely 1 metric of several thousand important metrics when properly monitoring an Exchange server... also, running out of disk space seems like the most basic stuff any monitoring suite should warn about - but now I'm being unusally annoying ^^

Improve this answer
2
  • Yes, but this is an actual case that we have run into on a recurring basis. I'm only guessing that it is due to diskspace issues, but we DO monitor diskspace, and those alarms are not tripped. We think it may be a temporary shortage due to intermediate backup caches or something.
    – Brent
    May 19, 2009 at 17:55
  • I see. It could be interesting to run a performance logger over night to see what it is doing during the problematic period.
    – Oskar Duveborn
    May 19, 2009 at 18:29

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .