One of the processes on a number of remote branch VMs is leaking handles and that eventually causes the server to run out of memory.
Restarting the service that owns (spawned) the process resolves the issue.
There's already a performance counter that is tracking the handle count for that process in SolarWinds - I just can't figure out how to restart the service based on the counter value.
So, the question:
How do I structure the alert, the trigger condition and the trigger action to automate this via SolarWinds?
Trigger condition:
- Objects:
- Component Name = <the name of the service>
- "The actual trigger condition:"
- Component Name = <handle count tracker for a process that is spawned by that service>
- Component Status in Warning or higher status
Obviously, this won't work because the trigger condition never matches the objects - i.e. there are never any objects matching these conditions. If I put <handle count tracker for a process that is spawned by that service> in the "objects", then SolarWinds' standard service restart utility (APMServiceControl.exe ${N=SwisEntity;M=ComponentAlert.ComponentID} -c=RESTART) won't work as it only works on services, not processes.
Is it possible to modify that trigger action so that it restarts the service that own the process?
If not, is there a simple way to resolve this?
Thank you!
P.S.
