Module #12 · intermediate Article
CrashLoopBackOff Troubleshooting: Root Cause Analysis for Application Container Crashes
CrashLoopBackOff means the container starts and immediately crashes — get the previous container's logs with `kubectl logs --previous`, then check exit codes...
5 min read+15 XPPublished 2026-07-23
Quick Answer: CrashLoopBackOff means the container starts and immediately crashes — get the previous container's logs with kubectl logs --previous, then check exit codes, env vars, and probe configuration.
Detailed Answer:
# Step 1: Confirm it's CrashLoopBackOff and check restart count
kubectl get pod my-pod
# NAME READY STATUS RESTARTS AGE
# my-pod 0/1 CrashLoopBackOff 8 12m
# High restart count = been crashing repeatedly
# Step 2: Get logs from the crashed container (not the current one)
kubectl logs my-pod --previous
kubectl logs my-pod --previous --tail=50 # Last 50 lines of crashed container
# Step 3: Check the exit code — tells you HOW it died
kubectl describe pod my-pod
# Look for:
# Last State: Terminated
# Reason: Error / OOMKilled / Completed
# Exit Code: 1
# Started: Mon, 01 Jan 2026 10:00:00
# Finished: Mon, 01 Jan 2026 10:00:01 ← crashed after 1 second!
# Exit code meanings:
# 0 = Success (shouldn't crash loop)
# 1 = General application error
# 2 = Misuse of shell command
# 126 = Command not executable
# 127 = Command not found (wrong entrypoint/CMD)
# 128+signal = Killed by signal
# 137 = SIGKILL (OOMKilled or manual kill)
# 143 = SIGTERM (graceful shutdown — may be probe timeout)
Common Root Causes:
# CAUSE 1: Application error on startup
# Symptom: Logs show exception/panic before crash
# Fix: Check env vars, config files, missing dependencies
kubectl exec -it my-pod -- env | grep -i db # Check env vars are set correctly
kubectl get secret db-secret -o yaml # Verify secret exists and has right keys
# CAUSE 2: OOMKilled — memory limit too low
# Symptom: Exit Code 137, Reason: OOMKilled
# Fix: Increase memory limit
kubectl get pod my-pod \
-o jsonpath='{.status.containerStatuses[0].lastState.terminated.reason}'
# Returns: OOMKilled
# Temporary fix: increase memory limit
kubectl patch deployment my-app \
-p '{"spec":{"template":{"spec":{"containers":[{"name":"app","resources":{"limits":{"memory":"1Gi"}}}]}}}}'
# CAUSE 3: Wrong command/entrypoint
# Symptom: Exit Code 127 "command not found"
# Fix: Check your Dockerfile CMD/ENTRYPOINT vs K8s command/args
# K8s command overrides Docker ENTRYPOINT
# K8s args overrides Docker CMD
spec:
containers:
- name: app
image: myapp:v1
command: ["/bin/sh"] # Overrides ENTRYPOINT
args: ["-c", "echo hello"] # Overrides CMD
# CAUSE 4: Liveness probe too aggressive
# Symptom: Pod restarts regularly, logs show normal operation
# Fix: Increase initialDelaySeconds or failureThreshold
livenessProbe:
httpGet:
path: /healthz
port: 8080
initialDelaySeconds: 30 # Was 5 — app needs 20s to initialize
failureThreshold: 5 # Was 3 — give more attempts
# CAUSE 5: Missing ConfigMap or Secret
# Symptom: Pod won't start, describe shows CreateContainerConfigError
kubectl describe pod my-pod | grep -A5 "Warning"
# Warning Failed CreateContainerConfigError:
# couldn't find key DB_HOST in ConfigMap production/app-config
# Fix: Verify the ConfigMap/Secret exists with the correct keys
kubectl get configmap app-config -o yaml
kubectl get secret db-secret -o yaml
# CAUSE 6: Image runs as root but security context forbids it
# Symptom: Exit Code 1 with "permission denied" in logs
# Fix: Either fix the image to use non-root, or adjust runAsUser
securityContext:
runAsUser: 1000 # Must match what the image expects
Debug Trick — Override the entrypoint to get a shell:
# Temporarily override command to keep container alive for inspection
spec:
containers:
- name: app
image: myapp:v1
command: ["sleep", "infinity"] # Override crashing entrypoint
# Then: kubectl exec -it my-pod -- /bin/sh
# Manually run your app command to see what error it produces
Key Takeaway: kubectl logs --previous is the single most important command for CrashLoopBackOff — it shows what the app was doing in its last moments before dying.
Finished reading?
Mark as complete to claim your +15 XP and track progress.