derailed/k9s · info

pod is terminating

Error message

pod is terminating

What it means

Pod.diagnose terminal gate: the pod passed every readiness check but its phase is Terminating (deletionTimestamp set). Purely informational — the pod is healthy and being deleted, so k9s still marks it non-quiet so operators notice in-flight removals.

Source

Thrown at internal/render/pod.go:256

	return p.diagnose(phase, cr, ct, ready, rgr, rgt)
}

func (*Pod) diagnose(phase string, cr, ct int, ready bool, rgr, rgt int) error {
	if phase == Completed {
		return nil
	}
	if cr != ct || ct == 0 {
		return fmt.Errorf("container ready check failed: %d of %d", cr, ct)
	}
	if rgt > 0 && rgr != rgt {
		return fmt.Errorf("readiness gate check failed: %d of %d", rgr, rgt)
	}
	if !ready {
		return fmt.Errorf("pod condition ready is false")
	}
	if phase == Terminating {
		return fmt.Errorf("pod is terminating")
	}

	return nil
}

// ----------------------------------------------------------------------------
// Helpers...

func asNominated(n string) string {
	if n == "" {
		return MissingValue
	}
	return n
}

func asReadinessGate(spec *v1.PodSpec, st *v1.PodStatus) string {
	if len(spec.ReadinessGates) == 0 {
		return MissingValue

View on GitHub (pinned to 2d3ccc6ba2)

Solutions

1. If rolling out: nothing — wait for replacement pods to go Ready
2. If stuck Terminating: kubectl get pod <name> -o jsonpath='{.metadata.finalizers}' and clear orphaned finalizers
3. Shorten terminationGracePeriodSeconds / preStop hooks if shutdowns drag
4. Verify endpoints controllers have removed the pod from Services

Example fix

# before: 10-minute graceful shutdowns stall rollouts
terminationGracePeriodSeconds: 600
# after
terminationGracePeriodSeconds: 60

if pod.DeletionTimestamp == nil {
	// not terminating; other gates apply
}

When it happens

Trigger: kubectl delete pod in progress; Deployment/DaemonSet rolling replacement deleting old pods; eviction draining a node; gracefulTerminationPeriod waiting on a slow preStop hook.

Common situations: Rollouts where old pods linger during long termination grace (stateful apps flushing, long preStop sleeps); node drains; stuck finalizers keeping pods Terminating forever.

Related errors

• expecting Deployment resource
• expecting Pod resource
• no node assigned
• failed to locate pod %q: %w
• user is not authorized to view pod logs

AI-assisted analysis of derailed/k9s@2d3ccc6ba2 (2026-08-15). Data as JSON: /api/errors/55678d641de0b44a. Report an issue: GitHub.