feat: add annotation to exclude components from triggering restart po…

[Syspretor]

…licy

Add ComponentNoTriggerRestartPolicyAnnotationKey to allow specific components in customComponentsPattern to be excluded from triggering the role's restart policy. When this annotation is set to "true" on a component's pod template, pod restart/delete events from this component will NOT trigger RecreateRoleInstanceOnPodRestart or RecreateRBGOnPodRestart actions.

This is useful for auxiliary components (e.g., monitoring, logging sidecars) whose failures should not affect the main workload.

Usage:
template:
metadata:
annotations:
rbg.workloads.x-k8s.io/component-no-trigger-restart: "true"

Ⅰ. Motivation

Ⅱ. Modifications

Ⅲ. Does this pull request fix one issue?

fixes #XXXX

Ⅳ. List the added test cases (unit test/integration test) if any, please explain if no tests are needed.

Ⅴ. Describe how to verify it

VI. Special notes for reviews

Checklist

• Format your code make fmt.
• Add unit tests or integration tests.
• Update the documentation related to the change.

[gemini-code-assist]

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!

[Copilot]

Pull request overview

This PR introduces a component-level annotation intended to prevent certain Pods (e.g., auxiliary/monitoring components in customComponentsPattern) from triggering restart-policy actions when they restart or are deleted.

Changes:

• Add a new annotation key/value (rbg.workloads.x-k8s.io/restart-trigger-policy: "Ignore") and constants for it.
• Gate Pod-event → RBG reconciliation enqueueing in the pod controller when the annotation is set to "Ignore".
• Update failure-handling documentation and a customComponentsPattern example to describe/illustrate the feature.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 8 comments.

File	Description
internal/controller/workloads/pod_controller.go	Skips enqueueing RBG reconcile on pod restart/delete when the new annotation is set to Ignore.
api/workloads/constants/annotation.go	Defines the new annotation key and allowed values (Inherit / Ignore).
doc/features/failure-handling.md	Documents excluding components from restart-policy triggers via the new annotation.
examples/basic/rbg/patterns/custom-components-pattern.yaml	Adds explanatory comments and an example component annotated to not trigger restart policy.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[cheyang]

The annotation check only covers the RecreateRBGOnPodRestart path. The RecreateRoleInstanceOnPodRestart path (which is the default and more common policy) is not covered — see inline comment for details.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

[Copilot]

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated 3 comments.

Comments suppressed due to low confidence (1)

pkg/reconciler/roleinstance/sync/instance_scale.go:287

• containerRestarted only checks pod.Status.ContainerStatuses and ignores InitContainerStatuses (and potentially EphemeralContainerStatuses). If an init container restarts, the instance recreation policy won’t trigger even though the doc comment says “any container” and LWS semantics include init-container restarts. Consider checking init-container restart counts as well (and decide explicitly whether to include ephemeral containers).

// containerRestarted checks if any container in the pod has been restarted.
func containerRestarted(pod *v1.Pod) bool {
	for i := range pod.Status.ContainerStatuses {
		if pod.Status.ContainerStatuses[i].RestartCount > 0 {
			return true
		}
	}
	return false

[cheyang]

Two issues to fix before merge:

1. containerRestarted() needs DeletionTimestamp + phase guards to match the adjacent PodFailed check — without them, Succeeded pods with RestartCount > 0 and terminating pods incorrectly trigger instance recreation.

2. Docs overstate pattern scope — the annotation only takes effect for customComponentsPattern and standalonePattern. LeaderWorkerPattern delegates restart handling to the LWS controller (lws_reconciler.go:286) which doesn't check this annotation.

The annotation design is solid, test coverage is comprehensive, and the previous review feedback about RecreateRoleInstanceOnPodRestart coverage has been properly addressed.

[cheyang]

containerRestarted(p) runs on all pods including Succeeded ones (via the inactivePods list) and pods being deleted. The sibling PodFailed check above has a DeletionTimestamp == nil guard but this doesn't.

A Succeeded pod that had a container restart during its lifetime (e.g. init crash-loop before eventual completion) would incorrectly trigger instance recreation. Same for a pod mid-termination.

Suggest:

if p.DeletionTimestamp == nil && p.Status.Phase != v1.PodSucceeded && containerRestarted(p) {
    return true
}

[cheyang]

This claims the annotation works with "any deployment pattern" including LeaderWorkerPattern, but LeaderWorkerPattern maps RecreateRoleInstanceOnPodRestart to LWS RecreateGroupOnPodRestart and delegates entirely to the LWS controller (lws_reconciler.go:286). The LWS controller doesn't check this annotation.

Should be scoped to customComponentsPattern and standalonePattern.