How to fix Pod stuck in Terminating in Kubernetes

Inspect the pod for finalizers and status:

kubectl get pod <pod-name> -o yaml | grep -A 5 finalizers
kubectl describe pod <pod-name>

Look for:
- finalizers: section in metadata
- Events showing termination progress
- Node status if pod is stuck on a specific node

Also check the node:

kubectl get nodes
kubectl describe node <node-name>

Finalizers block deletion until removed. If the finalizer controller isn't running or can't complete:

kubectl patch pod <pod-name> -p '{"metadata":{"finalizers":null}}'

Or edit directly:

kubectl edit pod <pod-name>
# Remove the finalizers section, save and exit

Caution: Only remove finalizers if you understand what cleanup they were supposed to perform.

If the pod remains stuck after removing finalizers:

kubectl delete pod <pod-name> --grace-period=0 --force

This tells the API server to remove the pod object immediately without waiting for confirmation from the kubelet.

Warning: Force delete doesn't guarantee the container stopped. The process may still be running on the node. Only use when the node is unreachable or you've verified the container is gone.

If the node is reachable but kubelet isn't responding:

# SSH to the node
ssh <node-ip>

# Check kubelet status
systemctl status kubelet

# View kubelet logs for errors
journalctl -u kubelet -n 100

# Restart if necessary
sudo systemctl restart kubelet

After kubelet restarts, it will sync state with the API server and complete pending terminations.

Stuck volume unmounts can block termination:

# Check PV/PVC status
kubectl get pv
kubectl get pvc

# Look for volume attachment issues
kubectl describe volumeattachment

For cloud providers, check if the volume is stuck in detaching state in the cloud console. You may need to force-detach the volume.

On the node:

# Check mounted volumes
mount | grep kubernetes
lsof +D /var/lib/kubelet/pods/<pod-uid>/volumes

Ensure applications handle SIGTERM properly:

import signal
import sys

def shutdown(signum, frame):
    # Cleanup code
    sys.exit(0)

signal.signal(signal.SIGTERM, shutdown)

Set appropriate grace periods:

spec:
  terminationGracePeriodSeconds: 30  # Adjust based on app needs

Use preStop hooks carefully—ensure they complete within the grace period.