Monitoring a SQL Cluster

[peakyblinder]

Hi,

I'm looking to monitor our SQL Cluster estate. I have seen that its not straight forward with regards to NSClient++. I was wondering if this guide (http://nsclient.org/nscp/wiki/guides/clusters) is still valid and if so, can it be used for Windows 2008/SQL 2005/2008.

Is anyone successfully monitoring their SQL clusters with NSClient/Nagios?

Have also asked this question over in the NSClient forums for full coverage.

Thanks,

[eloyd]

What do you think is lacking with regards to NSclient++?

[peakyblinder]

What do you think is lacking with regards to NSclient++?

I dont think its lacking, I just cant work it out.

I was asking who had managed to get it working, hoping someone could give some tips.

Thanks,

[tmcdonald]

I can't say I have monitored a SQL cluster with nsclient, and looking at the page you linked I see it was last updated 3 years ago. I would suggest following the documentation and seeing if it does work, unless someone can specifically say one way or the other. Did you hear anything from the nsclient forums? It might be worth trying to talk to the author directly via email if you can't get anything from either forum.

[cbeattie]

The way we do it is to monitor both hosts individually and the cluster itself as a third host. It is not necessary to set NSClient++ as a clustered service this way. The standard set of service checks are applied to the individual hosts, plus the SQL Server checks. The cluster virtual host gets checked for SQL Server and drive space checks for the clustered physical drives, but not for the standard checks a regular host gets (CPU, memory, running antivirus, etc.) since those are covered by the individual hosts.

It's true that for SQL clusters, there will always be a service on one host in a critical state (like those docs said), but doing it this way alerts the DBAs that the clustered SQL service has failed over from one host to the other. The service is still okay, according to the SQL check on the cluster virtual host, but the DBAs know they need to investigate why the failover occurred.

[tmcdonald]

@peakyblinder, we haven't heard from you in a bit. Did cbeattie's post help at all?

[peakyblinder]

Sorry, only just got back to work from holiday... Will look into it and update later today/tomorrow.

Thanks,

[tmcdonald]

Haven't heard from you in a bit. Did this end up working out?