Process Manager(LSF) alerts configuration through Environment Manager(EVM)

[sathya66]

Hi All,

I setup the alerts for all SAS services in EVM and all are working fine.

How can I setup the alerts for Process manager (lsf). I couldn't find any services related  to lsf in EVM. Please can you suggest  how to set them up in EVM.

Thanks,

SS

[gwootton · Accepted solution]

Hi Sathya!

SAS Environment Manager includes a SAS Grid Manager plug-in that interfaces with Platform Web Services, an LSF component, to gather metrics on LSF jobs, but there is not a pre-existing resource that directly monitors the LSF or Process Manager daemons.

It is possible to create Platform Resources in Environment Manager. Select the host under Resources -> Platforms where you would like to create the monitoring resource and then under the Tools Menu select "New Platform Service". 

Give it a unique name, e.g. hostname - Process Manager and select "Process" from the Service Type drop down and click "Ok".

This will take you to a new screen with a link to modify the configuration properties of your new service. Click the "Configuration Properties" link and in the process.query field enter a query for your process. For example, in UNIX to monitor a process called "jfd" this would be State.Name.eq=jfd. In Windows, you can specify the service name using Pid.Service.eq to define the Windows Service name (IBM Spectrum LSF Process Manager for example).

This new resource will gather metrics on the process/service, and you can then define alerts against these as defined in the documentation.

SAS Environment Manager User's Guide - Working with Resource Alerts

This could similarly be done for the other LSF daemons like gabd, lim, res and sbatchd.

[sathya66]

I am getting the below Error.
The configuration has not been set for this resource due to : Invalid configuration: org.hyperic.hq.measurement.monitor.LiveMeasurementException: Error retrieving value: Failed to invoke getProcMem[State.Name.eq=jfd]: Query did not match any processes: State.Name.eq=jfd
Thanks,
SS

[gwootton]

Thanks Sathya!

That error indicates that a process named "jfd" is not running on the Platform where you created the resource.

Is the jfd process running on the host where you created the query? The command "ps -ef | grep jfd | grep -v grep" should list the process as a result.