How to fix Pod grace period exceeded, force killing pod in Kubernetes

What this error means

Kubernetes sends SIGTERM to pods during shutdown (termination). The grace period (default 30 seconds) is the time allowed for graceful shutdown. If the pod does not terminate within this window, Kubernetes force-kills it (SIGKILL). This error indicates your application is not handling SIGTERM properly or takes longer than the grace period to shut down cleanly.

Typical symptoms

• Pods show "Terminating" status for extended periods
• Error logs: "Pod grace period exceeded, force killing"
• Application logs show no graceful shutdown sequence
• Kubernetes events show force deletion warnings
• Database connections lost abruptly without cleanup
• Deployment rollouts stall during termination
• Job pods terminate ungracefully without saving state

Common causes

1. Application not handling SIGTERM signal (only responding to SIGKILL)
2. Grace period too short for application shutdown (database flushes, file closes, etc.)
3. Infinite loops or deadlocks in shutdown handlers
4. Long-running requests not cancelling during shutdown
5. init processes in containers masking signals
6. finalizers on objects blocking pod deletion
7. Sidecar containers (logging, monitoring) hanging during shutdown

How to fix "Pod grace period exceeded, force killing pod"

Advanced notes

Frequently asked questions

What does "Pod grace period exceeded, force killing pod" mean?

Kubernetes sends SIGTERM to pods during shutdown (termination). The grace period (default 30 seconds) is the time allowed for graceful shutdown. If the pod does not terminate within this window, Kubernetes force-kills it (SIGKILL).

How do I fix "Pod grace period exceeded, force killing pod"?

Pod termination exceeds the grace period timeout when containers ignore SIGTERM, processes hang during shutdown, or finalization takes too long. Pods force-terminate without graceful cleanup.

What causes "Pod grace period exceeded, force killing pod"?

Application not handling SIGTERM signal (only responding to SIGKILL) Grace period too short for application shutdown (database flushes, file closes, etc.) Infinite loops or deadlocks in shutdown handlers

Is the "Pod grace period exceeded, force killing pod" error serious?

Yes. This is a high-severity error that usually blocks work until it is fixed, but the steps above resolve it.

Related errors

• Dashboard provisioning failed

  How to fix "grafana dashboard error" in Kubernetes

  MEDIUM

• Workload Identity Federation failed

  How to fix "gke workload identity failed" in Kubernetes

  MEDIUM

• Grafana datasource connection failed

  How to fix "grafana datasource error" in Kubernetes

  MEDIUM

• Error authenticating with Workload Identity: unable to generate access token

  How to fix "GKE Workload Identity error" in Kubernetes

  HIGH

• upstream connect error or disconnect/reset before headers. reset reason: overflow

  How to fix "upstream connect error or disconnect/reset before headers" in Kubernetes

  HIGH

• Waiting for rollout to finish

  How to fix "Waiting for rollout to finish" in Kubernetes

  HIGH

Official resources & further reading

• 📚Kubernetes Official Documentation
• 📖
  Kubernetes API Reference(Official Docs)
• 📖
  Kubernetes Tasks: Run Applications(Official Docs)
• 💬
  Stack Overflow: Kubernetes grace period exceeded(Stack Overflow)
• 🐙
  Kubernetes GitHub Issues(GitHub)