v0.47: vendor: bump runc to 1.1.9

[haircommander]

to fix an issue with cgroupv2 memory accounting not being accurate

[k8s-ci-robot]

Hi @haircommander. Thanks for your PR.

I'm waiting for a google member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Details

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.

[iwankgb]

I'm afraid we have to stick to 1.7 until Kubernetes updates its dependency version.

[iwankgb]

Pending update in Kubernetes.

[haircommander]

I don't think I understand why. does cadvisor depend on kubernetes go packages at all? I would think we'd update in cadvisor, then pull cadvisor and run changes into kube together. Otherwise, we'd do a runc bump in kube, one here, and then one for cadvisor, increasing the total amount of PRs needed

[haircommander]

WDYT @iwankgb ?

[iwankgb]

Kubernetes depends on cAdvisor and I"m hesitant to introduce changes that would affect their go.mod.

[iwankgb]

This is why i decided to block the PR: i should have been clearer explaining my reasoning.

[haircommander]

ah I see, but I am trying to bump it there too kubernetes/kubernetes#119909 I figured this would be a good first step

[haircommander]

@iwankgb @bobbypage WDYT?

[bobbypage]

Thanks @haircommander for the PR! Why do we need backport this PR to v0.47? Can we just merge this in master and use the next cut version of cAdvisor in the next k8s?

[haircommander]

I believe it's worth bumping the older versions to get kubernetes/kubernetes#118916 fixes in the k8s release branches to more accurately describe the memory usage on v2. 1.1.9 has the fix opencontainers/runc#3933 which should fix this without code changes in k8s or cadvisor

[haircommander]

WDYT @bobbypage

[haircommander]

@iwankgb @bobbypage PTAL

[bobbypage]

Thanks @haircommander. Where we plan to merge the fix for kubernetes/kubernetes#118916? If it's only master branch, then we don't need to backport. Or do we plan to merge it to both 1.27 + 1.28 and master? cAdvisor v0.47 is used in k/k in v1.27 and v1.28, so if we backport it here, we should ensure make the corresponding changes in k/k in v1.27 and v1.28.

[haircommander]

yeah I'd like to pull this back to 1.27 and 1.28

[bobbypage]

sg, let's merge this than, and I will followup to create a release.

[ritazh]

sg, let's merge this than, and I will followup to create a release.

Can we also get this into k8s 1.25 and 1.26? Thank you!

[RDPzero]

I would appreciate to see this on k8s, since it solves a problem I'm facing for a long time.

Are you going to cut a patch release from 0.47 @bobbypage ?

[aritraghosh]

@bobbypage When is the next release scheduled for, We would love to have it to merge kubernetes/kubernetes#119909

@haircommander

[aritraghosh]

@haircommander Can this change also be cherry picked to 1.26, 1.27 and 1.28? This change significantly reduces the memory reported on cgroup v2 to match v1

[ritazh]

sg, let's merge this than, and I will followup to create a release.

@bobbypage thanks for merging the PR! Can you please share when the next release is scheduled for so we can back port this change to kubernetes? Thanks in advance!

xref: kubernetes/kubernetes#121637 (comment)

[bobbypage]

Thanks @ritazh for the ping.

Before we backport this change, I would like to ensure we have the full clarity on the fix. Specifically there are still some open questions in opencontainers/runc#3933 (comment) if this is the right fix.