We have an application support person who manually monitors their server and application via Perfmon. They are having an issue that seems to crop up whenever the Virtual Bytes metric for a particular process hits 1.9GB in Perfmon. I was asked to monitor this and create an alert in SAM.
I started out looking at SNMP (HOST_RESOURCE_MIB) since that is our default polling method for servers, but this process (w3wp.exe) can be spawned and shutdown over and over again, each time with a new process ID. The OID to poll the process directly is dynamic in that the process ID becomes the last branch of the OID. Plus, after digging into that aspect of it, I realized that the HOST-RESOURCE-MIB does not seem to have an OID for Virtual Memory usage per process.
I switched to WMI since it does do Virtual Memory usage per process. After setting that up, I realized that the Perfmon Virtual Bytes metric is not the same as what WMI is giving me for virtual memory usage.
Has anyone already dealt with polling Perfmon statistics and have a good process? Any suggestions on a better way to approach this?
Thanks,
Romeo