Kubernetes Exit Code 0

Step-by-Step Troubleshooting

1. Confirm the Exit Code

kubectl describe pod <pod-name>

Look for:

Last State:     Terminated
  Reason:       Completed
  Exit Code:    0

The Completed reason with exit code 0 means the process finished successfully.

2. Check the Container Logs

kubectl logs <pod-name> --previous

The logs should show no errors — the application simply completed its work and exited. This confirms the issue is about the workload type, not a bug.

3. Determine the Workload Type

Is this supposed to be a long-running service or a one-shot task?

Long-running service (web server, API, worker):

• Should run indefinitely as a Deployment or StatefulSet
• Must have a foreground process that stays alive

One-shot task (migration, data import, cleanup):

• Should run once and exit as a Job
• Exit code 0 means success

4. Fix: Convert to a Job (For One-Shot Tasks)

If the workload is meant to run once and exit:

apiVersion: batch/v1
kind: Job
metadata:
  name: db-migration
spec:
  backoffLimit: 3
  activeDeadlineSeconds: 600
  ttlSecondsAfterFinished: 3600
  template:
    spec:
      restartPolicy: OnFailure
      containers:
        - name: migrate
          image: myapp:v1
          command: ["./migrate.sh"]

# Delete the Deployment
kubectl delete deployment <deploy-name>

# Create the Job
kubectl apply -f job.yaml

# Monitor the Job
kubectl get jobs -w

5. Fix: Keep the Process in the Foreground (For Services)

If the workload should be a long-running service, ensure the main process runs in the foreground.

Problem: Background process

# BAD: Application starts in background, script exits
CMD ["sh", "-c", "myapp &"]

Fix: Foreground process

# GOOD: Application runs in foreground as PID 1
CMD ["myapp"]

Problem: Wrapper script without exec

#!/bin/sh
# BAD: Shell starts app as child process, then shell exits
myapp --config /etc/app/config.yaml

#!/bin/sh
# GOOD: exec replaces the shell with the app process
exec myapp --config /etc/app/config.yaml

6. Fix: Shell Script Entrypoint

A common pattern is a shell script that sets up environment and then runs the app. Always use exec for the final command.

#!/bin/sh
set -e

# Setup tasks
echo "Configuring application..."
envsubst < /etc/app/template.yaml > /etc/app/config.yaml

# Start the application (MUST use exec)
exec java -jar /app/application.jar

Without exec, the shell is PID 1 and the Java process is a child. When the shell finishes executing (after the java command is backgrounded or after the script logic is complete), the container exits.

7. Debug the Container Locally

Test the container to see if it stays running.

# Run the container and check if it exits
docker run --rm <image>

# Check the exit code
echo $?

# If it exits immediately, check what the entrypoint does
docker run --rm --entrypoint sh <image> -c "cat /entrypoint.sh"

8. Check for Missing Foreground Flags

Some applications need specific flags to run in the foreground:

# nginx needs daemon off
command: ["nginx", "-g", "daemon off;"]

# Apache httpd needs -D FOREGROUND
command: ["httpd", "-D", "FOREGROUND"]

# Redis uses foreground by default but check config
command: ["redis-server", "--daemonize", "no"]

9. Verify the Fix

# Watch the pod
kubectl get pods -w

# Verify it stays running
kubectl get pod <pod-name> -o jsonpath='{.status.containerStatuses[0].restartCount}'
# Should be 0 or not increasing

# Check the container state
kubectl get pod <pod-name> -o jsonpath='{.status.containerStatuses[0].state}'
# Should show: {"running":{"startedAt":"..."}}

The pod should reach Running state and stay there without restarting.