cpu/memory manager containerMap memory leak

[Dingshujie]

What type of PR is this?

/kind bug

What this PR does / why we need it:

if cpu manager/ memory manager policy set to none, no one remove container info from containerMap, finaly lead memory leak.

this pr call cpu manager/memory manager RemoveContainer in PostStopContainer, to remove container info from containerMap

Which issue(s) this PR fixes:

Fixes #

Special notes for your reviewer:

Does this PR introduce a user-facing change?

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

[Dingshujie]

/sig node

[Dingshujie]

/cc @liggitt @rphillips @gjkim42 @SergeyKanzhelev

[derekwaynecarr]

/assign

@klueska as well.

[klueska]

We cannot do what you propose here. We used to do what you are doing, but it causes issues when containers are restarted (within a still running pod). We need to maintain the containerMap across these restarts.

We may indeed have a memory leak for the case of the none policy, but we will need to solve it another way.

[Dingshujie]

We cannot do what you propose here. We used to do what you are doing, but it causes issues when containers are restarted (within a still running pod). We need to maintain the containerMap across these restarts.

We may indeed have a memory leak for the case of the none policy, but we will need to solve it another way.

got restart scene, i will try another method to solve it

[Dingshujie]

@klueska find another method to slove this leak, Can you take a look at this PR again? thanks a lot.

[ehashman]

/release-note-none
/priority important-soon
/triage accepted

[Dingshujie]

Can we add this to v1.24 milestone? @rphillips @liggitt

[liggitt]

It appears we have to wait for the tree to thaw for 1.25 for this fix

Bugfixes can be merged during code freeze if impact/risk is justified.

Is this a regression? (looks like not, since there are cherry-picks open)

What is the risk of the fix? Is it changing feature-gated or optionally executed code? If so, what feature gate or opt-in configuration exposes the bug?

[liggitt]

I also wouldn't expect us to take a change without accompanying tests

[Dingshujie]

sorry， careless close pr, reopen now

[Dingshujie]

It appears we have to wait for the tree to thaw for 1.25 for this fix

Bugfixes can be merged during code freeze if impact/risk is justified.

Is this a regression? (looks like not, since there are cherry-picks open)

What is the risk of the fix? Is it changing feature-gated or optionally executed code? If so, what feature gate or opt-in configuration exposes the bug?

cpu manager default policy is none, if pod repeated create and delete, kubelet memory will keep growing.
also cpu manager policy set to static， none exclusive pod container id also will leak in containermaps, kubelet memory will keep growing.

[k8s-triage-robot]

The Kubernetes project has merge-blocking tests that are currently too flaky to consistently pass.

This bot retests PRs for certain kubernetes repos according to the following rules:

• The PR does have any do-not-merge/* labels
• The PR does not have the needs-ok-to-test label
• The PR is mergeable (does not have a needs-rebase label)
• The PR is approved (has cncf-cla: yes, lgtm, approved labels)
• The PR is failing tests required for merge

You can:

• Review the full test history for this PR
• Prevent this bot from retesting with /lgtm cancel or /hold
• Help make our tests less flaky by following our Flaky Tests Guide

/retest

[rphillips]

@Dingshujie Since the project is in a frozen state, the criteria is does the PR fix a regression within the 1.24 release. Waiting for the thaw minimizes churn/bugs/regressions added during the release process.

[Dingshujie]

@Dingshujie Since the project is in a frozen state, the criteria is does the PR fix a regression within the 1.24 release. Waiting for the thaw minimizes churn/bugs/regressions added during the release process.

ok, let's wait for 1.24 released.

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: Dingshujie, klueska, yanghesong

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/cm/OWNERS [klueska]

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

[pacoxu]

This will be included in 1.25. So a cherry-pick PR for 1.24 will be needed as well. (It is ok to open one after v1.24 is released. https://github.com/kubernetes/kubernetes/tree/release-1.24 is not ready yet.)

[Dingshujie]

This will be included in 1.25. So a cherry-pick PR for 1.24 will be needed as well. (It is ok to open one after v1.24 is released. https://github.com/kubernetes/kubernetes/tree/release-1.24 is not ready yet.)

i will propose a pr for release-1.24