How to Fix CrashLoopBackOff Pod Status in Kubernetes

✨ Quick Answer

The CrashLoopBackOff status in Kubernetes indicates that a container inside a Pod fails to start, crashes immediately, and enters a continuous loop of automatic restarts where Kubernetes exponentially increases the back-off delay between restart attempts.

Quick Diagnostics

Cause

Pod stuck in CrashLoopBackOff status with high RESTARTS counter: Application crash, missing environment variables, Out of Memory (OOMKilled), or failing livenessProbe

Solution

Inspect previous container logs with kubectl logs --previous and check events with kubectl describe pod

Step-by-Step Solution

1. 1

   Step 1: Inspect the failure reason with kubectl describe

   The primary command for gathering lifecycle details about the failing Pod is kubectl describe:

   BASH Copy

   # Fetch detailed status of the target Pod
   kubectl describe pod <pod-name> -n <namespace>
   

   In the Containers > State and Last State sections, pay close attention to the following indicators:

   • Exit Code 1 or 127: Application failure (unhandled runtime exception, missing file, or broken CMD path).
   • Exit Code 137 (OOMKilled): The container exceeded its strict memory limit (limits.memory) configured in the YAML manifest.
   • Liveness probe failed: The configured health check failed to respond within the specified timeout.
2. 2

   Step 2: Extract logs from the crashed container

   To inspect standard output (stdout/stderr) right before the container crashed on its previous attempt, add the --previous flag:

   BASH Copy

   # View logs from the previous container instance
   kubectl logs <pod-name> --previous -n <namespace>
   
   # If the Pod has multiple containers, specify the container name:
   kubectl logs <pod-name> -c <container-name> --previous -n <namespace>
   

3. 3

   Step 3: Adjust container memory limits (OOMKilled)

   If the container was killed due to memory exhaustion (exit code 137), update your deployment manifest to allocate additional memory resources:

   YAML Copy

   apiVersion: apps/v1
   kind: Deployment
   metadata:
     name: my-service
   spec:
     template:
       spec:
         containers:
         - name: my-app
           image: my-image:latest
           resources:
             requests:
               memory: "256Mi"
               cpu: "250m"
             limits:
               memory: "512Mi" # Increase limit if OOMKilled occurs
               cpu: "500m"
   

   Apply the changes:

   BASH Copy

   kubectl apply -f deployment.yaml
   

4. 4

   Step 4: Adjust or delay Liveness and Readiness Probes

   If your application takes a long time to boot (such as performing database migrations on startup), the livenessProbe might kill the Pod prematurely. Increase initialDelaySeconds:

   YAML Copy

   livenessProbe:
     httpGet:
       path: /healthz
       port: 8080
     initialDelaySeconds: 30 # Give the application time to complete initialization
     periodSeconds: 10
     failureThreshold: 3