Alarms of Guest OS process

[costelj]

Is there a way to setup an alarm for a certain process INSIDE a virtual machine? I suspect there isn't, based on the user guide and going through the gui, but I thought I'd ask anyway. Can it be a feature request? Or is it too far out of scope? Here's my situation:

I have several VMs that have a particular runaway process, occurring several times throughout the week (we are troubleshooting the issue with Microsoft, fwiw). It would be great to be alerted when the runaway process kicks in; as it stands now we have to continually watch each VM.

Thanks!

John

[Alexey D.]

Hello John,

Monitor is agentless application by its design. So it doesn't have any modules installed inside VM to catch such kind of internal processes.
But I'm just wondering, maybe this runaway process has some 'external signs' (raised disk I/O, memory usage, etc.)?
Monitor has plenty of alarms to fine-tune and catch that behaviour. Can you describe your runaway process?

[Vitaliy S.]

John,

Yes, that might be considered as a feature request, but if we implement it then most likely you would need to deploy an agent inside a guest OS to monitor internal processes.

By the way, do you have any EMS? How do you monitor your physical environment? SCOM by any chance?

[Gostev]

Agent inside OS is not a big deal as long as it is tiny, brings value, and users do not need to manage it. For example, Veeam Backup does use tiny agent inside each VM to coordinate application-aware processing stuff and guest indexing. But nobody from users realizes that because they do not need to manage it (pay-to-get/deploy/manage/update/monitor/babysit/troubleshoot). From user's perspective, the solution is "agentless".

Let's discuss this feature request internally.

[costelj]

Hello John,
I'm just wondering, maybe this runaway process has some 'external signs' (raised disk I/O, memory usage, etc.)?
Monitor has plenty of alarms to fine-tune and catch that behaviour. Can you describe your runaway process?

I'm trying this angle, but it's not accurate enough for my liking (well, the customer's liking, actually. I'm the consultant for them in this scenario). I can usually get the process to trigger a spike in active memory on the VM and THAT gets me the alert I want, but we get some false positives when other, legitimate things pike the VMs memory. To answer your question about the process, these are XenApp servers and there is an issue with Outlook using up available memory....caused by the print driver isolation feature of R2 (I am not a citrix guy, but our citrix guy is working with Citrix, Microsoft and the customer on this). The customer does not want to spend money on Citrix Edgesight and since I'm in there as a vSphere consultant, demonstrating the benefits of VEEAM One (for other reasons unrelated to this issue, such as replacing vFoglight), I thought I'd take a crack at getting some reliable alerts for this Citrix issue while I was onsite. Like I mentioned above, it's decent when I do it via active memory, but I still get false positives.

By the way, do you have any EMS? How do you monitor your physical environment? SCOM by any chance?

The customer has no monitoring solutions in-place today except vFoglight. And hopefully VEEAM One soon.

Let's discuss this feature request internally.

Cool. Thanks for listening.


John