Work-around for missing memory metrics on CRI-O exited containers

[joelsmith]

What type of PR is this?
/kind bug

What this PR does / why we need it:

HPA needs metrics for exited init containers before it will take action. By setting memory and CPU usage to zero for any containers that cAdvisor didn't provide statistics for, we are assured that HPA will be able to correctly calculate pod resource usage.

This PR essentially makes the kubelet's cAdvisor stats provider behave the same as its CRI stats provider (see #74336).

Which issue(s) this PR fixes:
No issue filed.

Special notes for your reviewer:

N/A

Does this PR introduce a user-facing change?:

NONE

Additional documentation e.g., KEPs (Kubernetes Enhancement Proposals), usage docs, etc.:
N/A

[joelsmith]

/cc @dashpole
This is the issue that I brought up in the sig-node meeting this week.

[dashpole]

Should we do this for all metrics (e.g. disk as well) or do you think we should only do so for cpu and memory?

[dashpole]

Should we default other fields to 0 to have consistent behavior?

[joelsmith]

Perhaps. I was just trying to match the behavior in the CRI stats provider, here: https://github.com/kubernetes/kubernetes/blob/master/pkg/kubelet/stats/cri_stats_provider.go#L537

[joelsmith]

As to the question of populating other missing metrics like disk and network, I'm not sure. It's certainly worth thinking about.

[dashpole]

/priority important-soon
This meets my bar for a bug-fix. Feel free to follow-up with other changes discussed above if you are interested.
/lgtm
/approve

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: dashpole, joelsmith

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• pkg/kubelet/stats/OWNERS [dashpole]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[rphillips]

@dashpole @derekwaynecarr thoughts on getting this into 1.18?

[dashpole]

@rphillips I am in favor.

[derekwaynecarr]

its a bug fix that impacts users.

/milestone v1.18

[joelsmith]

/retest

[k8s-ci-robot]

@joelsmith: The following test failed, say /retest to rerun all failed tests:

Test name	Commit	Details	Rerun command
pull-kubernetes-conformance-kind-ga-only-parallel	da98829	link	/test pull-kubernetes-conformance-kind-ga-only-parallel

Full PR test history. Your PR dashboard. Please help us cut down on flakes by linking to an open issue when you hit one in your PR.

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. I understand the commands that are listed here.

[matthias50]

@joelsmith, @dashpole @derekwaynecarr, we are running into this and are currently about to upgrade from 1.15 to 1.6. Can we get this cherry-picked back to 1.17 and 1.16?

[joelsmith]

I've opened #90900 (1.16) and #90901 (1.17).
Edit: both of these have merged, so in addition to being fixed in 1.18.0, it should also be in 1.16.9 and 1.17.6.