Failed guest OS does not trigger any action

Matt Schlarb 1

A failing guest OS (eg. kernel panic) does not trigger any action. I know from Hyper-V that you could monitor the guest OS heartbeat and define an action when heartbeat is lost. Is there any possibility in Azure to restart a frozen VM? Other cloud providers will restart the VM when it is not responsive...

jakaruna-MSFT 601 Reputation points Microsoft Employee

2019-12-13T10:42:15.137+00:00

I reached out to the internal teams to get info about this question. I will update here once i have some information.
Ronen Ariely 15,196 Reputation points

2019-12-30T09:54:38.887+00:00

Hi,
did you tried to restart the machine using PowerShell?
If yes, then what is the exact error you get
theodorbrander 56 Reputation points

2019-12-30T12:47:19.167+00:00

I have not deep dived in this, but if I remember correctly this can be performed by Azure Monitor & an automation runbook. I googled quickly and you can probably use this for reference.

https://blogs.technet.microsoft.com/systemcenterpt/2018/03/21/heartbeat-metric-alerts-for-log-analytics/
Matt Schlarb 6 Reputation points Microsoft Employee

2020-01-08T16:25:04.94+00:00

I already searched for stuff like this. Thing is, there's no "Heartbeat" metric (step 5 in your guide) when I select a VM (guest OS doesn't matter in this case). Probably you need Operations Management Suite for this, but I'm searching for something Azure native.

2 answers

Matt Schlarb 6 Reputation points Microsoft Employee

2020-01-10T18:31:28.643+00:00

Ok I got it now. It was a little bit of a hassle because the metric "Hearbeat" is not available under standard VM metrics (waagent), but only in connection with Log Analytics (omsagent). Connect VM to Log Analytics Workspace where your create a monitoring rule with a condition that monitors the VM's Hearbeat signal. As action, create a PowerShell automation runbook which reboots the VM. Basically it's like @theodorbrander mentioned, only the Log Analytics part was missing.
Please sign in to rate this answer.

1 person found this answer helpful.
Matt Schlarb 6 Reputation points Microsoft Employee

2020-01-10T18:32:18.32+00:00

The only thing that baffles me was that sometimes it takes up to 15 minutes until the VM gets rebooted. Maybe it's the whole runbook automation framework and its authentication methods or whatsoever...
Sign in to comment

Use comments to ask for clarification, additional information, or improvements to the question.
Iwan van de Wetering 1 Reputation point

2019-12-30T15:59:03.63+00:00

Restart-AzureRmVM -ResourceGroupName "ResourceGroup" -Name "VirtualMachine"
From Azure CLI
regards

Iwan
Please sign in to rate this answer.
Iwan van de Wetering 1 Reputation point

2019-12-30T16:02:16.167+00:00

az vm restart -g ResourceGroup -n VirtualMachine

Matt Schlarb 6 Reputation points Microsoft Employee

2020-01-08T15:03:54.443+00:00

Hello Iwanvandewetering,

I'm talking about an automatic restart of a failed VM, not a manual VM restart.

Regards
Sign in to comment

Use comments to ask for clarification, additional information, or improvements to the question.

Share via

Failed guest OS does not trigger any action

2 answers

Your answer