Hello

We have a two node (sql1 and sql2) active/active cluster (Windows 2008 R2) each running two SQL instances. PRTG is installed on both nodes, the master node is sql1. The cluster is named sqlcluster.

Trying to monitor these four SQL 2008 R2 instances, I have added the a "WMI Microsoft SQL Server 2008" sensor and selected all four SQL instances. The sensor is added under a device which points to the sqlcluster DNS name.

Is this correct? The two sensors monitoring the instances on sql1 reads values fine (for example user connections), but the two sensors monitoring the instances on sql2 does not report any errors, but they both return 0 for all the values.

Any idea what is causing this?


Article Comments

Hello,

what happens if you try adding separate devices for each of the individual physical nodes and then setup the sensors there? Do then all work?

best regards.


Jul, 2012 - Permalink

If I add a device pointing at sql2, it receives the correct values using WMI for the instances on that node. It's good, but it fails in case of a failover. Then I would manually have to change the monitored instances on the nodes.

The way I see it, the device point at the cluster (sqlcluster in my case) should query both nodes and use the value from the one responding. Or maybe I'm just missing some thing - any ideas?


Jul, 2012 - Permalink

Well, it was supposed at least for a test to troubleshoot.


Jul, 2012 - Permalink

Anything else I can do to test it? Is this supposed to work, or is it something that is unsupported? Can you please provide some details? The last comment was very short...


Jul, 2012 - Permalink

Well, if you are not trying to troubleshoot, by for example adding the individual physical nodes as devices, and then trying (for a test) to monitor them with individual sensors, it's kind of hard to help.


Jul, 2012 - Permalink

I did that, and I explained it in my previous comment. Addind the physical node makes me able to monitor the instances running on that node. So that works. But if one node fails, I'm not able to monitor the instances that failed over to another node.


Jul, 2012 - Permalink

Hello again. We are wondering if you have any suggestions on how to resolve this issue? We're considering buying the product, but not unless we can monitor our SQL cluster.

We really hope that we can solve this issue.

Thank in advance.


Jul, 2012 - Permalink

Any suggestions on how we can debug it?


Aug, 2012 - Permalink