Request the pid of another container if current pid is not longer valid

[harche]

When using CRI-O as the container runtime, the infrastructure container can be dropped. In this scenario, CRI-O provides cadvisor with the process ID (PID) of one of the containers within the pod to collect network statistics. However, if the particular container exits, cadvisor is left with an invalid PID. To address this issue, this PR makes cadvisor to switch to another PID from the same pod, ensuring the continuous collection of network statistics.

Pod used to validate this change,

apiVersion: v1
kind: Pod
metadata:
  name: init-sleep-pod
spec:
  initContainers:
  - name: init-sleep
    image: busybox
    command: ["sh", "-c", "sleep 60"]
  containers:
  - name: forever-sleep
    image: busybox
    command: ["sh", "-c", "while true; do sleep 1; done"]

[k8s-ci-robot]

Hi @harche. 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.

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.

[harche]

/cc @haircommander @rphillips

[rphillips]

/lgtm
/assign @bobbypage

[haircommander]

I think we should double check h.needNet() here, or else we may needlessly recall into cri-o because a container didn't provide the network stats

[haircommander]

we can just combine this check as else if len(...) below

[harche]

@haircommander Thanks, good point. Updated.

[haircommander]

I would prefer these two conditions are switched so we don't have to call needsNet twice:

if !h.needNet() {
....
} else if len(...) {
....
}

[harche]

@haircommander Done. Thanks.

[haircommander]

/lgtm

thanks!

[rphillips]

/assign @bobbypage @dims

[dims]

LGTM

[dims]

/ok-to-test

[dims]

/approve
/lgtm

[bobbypage]

lgtm, thanks!

[bobbypage]

Backports:

• v0.47 - [0.47] Backport of PR 3301 to 0.47 #3313 - released in https://github.com/google/cadvisor/releases/tag/v0.47.2
• v0.46 - [0.46] Backport of PR 3301 to 0.46 #3320 - released in https://github.com/google/cadvisor/releases/tag/v0.46.1
• v0.45 - [0.45] Request the pid of another container if current pid is not longer valid #3321 - released in https://github.com/google/cadvisor/releases/tag/v0.45.1
• v0.44 - [0.44] Request the pid of another container if current pid is not longer valid #3322 - released in https://github.com/google/cadvisor/releases/tag/v0.44.2