Support Azure CNI with Kured reboot

[tamilmani1989]

Component (Azure NPM or Azure CNI):

Azure CNI

Describe in detail the feature/behavior/change you'd like to see:
With kured reboot, we witnessed files under /var/run persist and cni trying to rehydrate state. We need to detect kured reboot and handle effectively in CNI

Orchestrator(e.g. Kubernetes, Docker):
Kubernetes

Operating System (Linux/Windows):
Linux

Anything else you would like to add:
[Miscellaneous information that will assist in solving the issue.]

[a-rt-works]

Other semi related issues with Kured on AKS

Azure/AKS#1320

[palma21]

Why do you need to treat a kured reboot differently than a regular soft VM reboot? It should just be a basic systemctl reboot. Does that issue not happen if you manually do that on the vm?

[palma21]

cc @aanandr

[tamilmani1989]

We have seen this only with kured setup so far. We saw node wasn't rebooted but reboot time got updated somehow. We couldn't repro this on our own. It would be good if you could provide us repro steps or repro setup so that we can investigate

[palma21]

If node wasn't rebooted, then that's a bug in either kured or customer config. Do you have those logs somewhere? we can follow up offline

[tamilmani1989]

@palma21 on further investigation, found that uptime -s (last reboot time) was returning erratic values even if vm hasn't rebooted. Not sure if this happening in newer azure linux kernel. Nevertheless we don't need this code path to be invoked for cni. If cni, we skip getting reboot time and do any action based on that. That would fix this issue.

[pranaypathik]

Hello
I was wondering on the status of this issue.

Pranay

[github-actions]

This issue is stale because it has been open 60 days with no activity. Remove stale label or comment or this will be closed in 14 days

[github-actions]

Issue closed due to inactivity.