I am attempting to monitor a PeopleSoft process that is running on only the Active server in the cluster of 2 servers. The problem I am having that when I when I attempt to monitor the cluster for the process the process is checked on both computers in the cluster instead of just the active one.

Any ideas on how to isolate just the active computer?

Thanks, I'm a total noob at this SCOM stuff.

please refer to this blog

http://thoughtsonopsmgr.blogspot.hk/2011/09/intelligent-service-monitoring-part-i.html

Hope this help you

Roger

That example is for a "service" and I need to do a "process". I see the logic and the general idea is similar for both, however the entire check is based on the Automatic state of the service, which is not an option for a process that either running or not. Is there a guide out there for doing the same thing for a monitor?

I attempted to extrapolate the relevant data from the article and apply it to a monitor check, but the meat of the article is too in-directly related to be relevant.

Hi Adam,

Does this helps: http://social.technet.microsoft.com/Forums/en-US/operationsmanagerauthoring/thread/406449cb-5cf5-4ad4-90d5-bda2766b396b  ?

Hi Adam,

Does this helps: http://social.technet.microsoft.com/Forums/en-US/operationsmanagerauthoring/thread/406449cb-5cf5-4ad4-90d5-bda2766b396b&n

Alex,

I don't believe that I need to write my own management pack for this. It seems as if I should be able to create a group that dynamically populates based on which server is the Active Cluster Node. And then link the Process Monitor to that group. I am currently stuck trying to figure out how to build that group to either prove or disprove the theory.

Does that sound like it might work or am I wasting my time?

Cheers!

There are two issues here:

First is the way a process monitoring work. The Microsoft.SystemCenter.Process.DiscoveryDataSource doesn't evaluate any conditions, it just returns an object instance (a process). So when you target the discovery to a server - it returns an object. And it will return this object for all nodes. I don't believe it will work for a clustered processes.

Second, you need a mechanism that will 'follow' a failover.

You can try to use an existing object from Windows Cluster MP, Resource Groups for example and target a single custom monitor to a resource group. You can use a built-in Microsoft.SystemCenter.Process.ProcessInstanceCountMonitorType or create your own.

And, yes - I can't see any good way to achieve your goals without developing a custom MP.

Hope this will help you

http://social.technet.microsoft.com/Forums/en-US/operationsmanagerauthoring/thread/564d41d7-19ab-47d6-a0e1-f9bc8a108f15/

Roger

There are two issues here:

First is the way a process monitoring work. The Microsoft.SystemCenter.Process.DiscoveryDataSource doesn't evaluate any conditions, it just returns an object instance (a process). So when you target the discovery to a server - it returns an object. And it will return this object for all nodes. I don't believe it will work for a clustered processes.

Second, you need a mechanism that will 'follow' a failover.

You can try to use an existing object from Windows Cluster MP, Resource Groups for example and target a single custom monitor to a resource group. You can use a built-in Microsoft.SystemCenter.Process.ProcessInstanceCountMonitorType or create your own.

And, yes - I can't see any good way to achieve your goals without developing a custom MP.

Hope this will help you

http://social.technet.microsoft.com/Forums/en-US/operationsmanagerauthoring/thread/564d41d7-19ab-47d6-a0e1-f9bc8a108f15/

Roger

Hi Adam.

I came across this recently, and according to TechNet this is possible.  However, you need to do a little work.  Essentially, the process monitor 'Unit Monitor' as configureb by the UI is set to non-remotable.  You can modify t his by exporting the Management Pack containing the configuration and manually editing this.  I've written an article about this here

HTH

Cheers

Shaun