node-problem-detector does not discover OOMs in newer Linux kernels

[tosi3k]

What happened:
In a cluster with NPD addon installed and Linux 5.1+ as the kernel OS of the nodes OOMs are not being captured by the NPD.

What you expected to happen:
NPD should inform about OOMs by emitting appropriate K8s events.

How to reproduce it (as minimally and precisely as possible):
Create a K8s cluster with NPD addon and Linux 5.1+ installed on the nodes. Schedule a pod which leaks memory. Try to list events with reason: OOMKilling field after an OOM occurs to see there's none.

Anything else we need to know?:
Fix is already merged into newest (v0.8.5) NPD release: kubernetes/node-problem-detector#481.

We're currently waiting for the release of Docker image of NPD v0.8.5 - we can proceed with fixing this in k/k afterwards. After it gets merged to the master branch, it should be then cherry-picked to older releases as well.

Environment:
Any Kubernetes version with NPD addon installed and Linux as the node OS with the kernel version 5.1 or newer.

[k8s-ci-robot]

@tosi3k: This issue is currently awaiting triage.

If a SIG or subproject determines this is a relevant issue, they will accept it by applying the triage/accepted label and provide further guidance.

The triage/accepted label can be added by org members by writing /triage accepted in a comment.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[tosi3k]

/sig scalability
/sig node

I opened a fixing PR in #96716 but it's on hold since the Docker image of NPD v0.8.5 doesn't exist yet.