[zhan9san]

Issue #, if available: Fixes #1288

Description of changes:

What

Adds a PodDetails field to the webhook template data, with each pod's namespace, phase, controller kind, labels and annotations.

Why

The webhook fires while the affected pods can still be inspected, but Pods only carries names. Without namespace, phase, owner, labels or annotations, a receiver can't tell which workloads an interruption actually hit, or route the notification to the team that owns them. Details are in #1288.

Changes

• pkg/node: add a PodDetail type, FetchPodDetails and a PodNames helper. FetchPodDetails uses the same fetchAllPods List call as FetchPodNameList, so no extra API request is made.
• pkg/interruptionevent/draincordon: the handler fetches pod details once and sets both Pods (unchanged) and PodDetails from that result.
• pkg/monitor: add PodDetails []node.PodDetail to InterruptionEvent, tagged json:"-". The event is logged with Interface("event", ...) when handling fails, so this keeps pod annotations out of the logs. Templates still see the field.
• OwnerKind is the kind of the pod's controller (metav1.GetControllerOf), or "" for a bare pod. Expose pod details (namespace, phase, owner, labels, annotations) to the webhook template #1288 proposed the first owner reference; the controller reference is more precise when a pod has non-controller owners.
• Pods, LogPods and the default template are unchanged, so existing templates render the same as before.

Example: running workload pods only.

{{ range .PodDetails }}
{{- if and (eq .Phase "Running") (ne .OwnerKind "DaemonSet") }}
{{ .Namespace }}/{{ .Name }} {{ index .Annotations "example.com/owner" }}
{{ end }}{{ end }}

How you tested your changes:

• New unit tests:
  • TestFetchPodDetails covers field mapping, a controller owner and a non-controller owner.
  • TestPodNames.
  • TestHandleEvent_PopulatesPodDetails checks that the handler sets Pods and PodDetails.
  • TestPostPodDetails filters by phase and owner and reads labels and annotations in a template.
  • TestInterruptionEventJSONOmitsPodDetails.
• go test ./... passes.
• e2e: test/e2e/webhook-test now ranges over .PodDetails in its template. If the field couldn't be rendered, template execution would fail and the test's Webhook Success assertion would not pass. I checked the escaped --set value with helm template, but I didn't run the kind-based e2e suite locally.

Environment (Linux / Windows): macOS (darwin/arm64), unit tests only
Kubernetes Version: client-go fake clientset (k8s.io/client-go v0.30.4)

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[zhan9san]

@LikithaVemulapalli

Could you plz help review this PR?

[LikithaVemulapalli]

Overall looks good, added some minor comments, will enable the PR tests to see if the change runs successfully on all the k8s versions that we have in our e2e tests. Thanks for owning the fix :)

[LikithaVemulapalli]

Exposing full label/annotation maps is fine since the template opts in and the default template is unchanged, but annotations can hold sensitive data (e.g. last-applied-configuration). Add a docs caveat to select specific keys rather than dumping whole maps. The json:"-" keeping them out of logs is good.

[zhan9san]

Added a "Webhook Template" section to the chart README listing the .PodDetails fields, with a caveat to select specific keys via index .Annotations "key" rather than rendering whole maps, since annotations like last-applied-configuration can carry sensitive data. Also noted it on the Annotations field.

[LikithaVemulapalli]

FetchPodNameList now has no in-repo caller, remove it or reimplement as PodNames(FetchPodDetails(...)).

[zhan9san]

Removed it along with its test; the handler derives names via PodNames(FetchPodDetails(...)), and TestFetchPodDetails/TestPodNames cover that path.

[zhan9san]

@LikithaVemulapalli Thanks for the review! Both comments are addressed in 7bb07c5. Could you approve the workflow run for the new commit, and add the stalebot-ignore label so the PR is not closed by the stale bot while it waits for review? Thanks!

[LikithaVemulapalli]

/lgtm

[zhan9san]

@LikithaVemulapalli thanks for your help

May I ask when we can have a new release for this?