How to fix 504 Gateway Timeout in Kubernetes

Run kubectl logs -n ingress-nginx <controller-pod-name> to see if timeouts are the issue. Look for "upstream timed out" messages. Replace <controller-pod-name> with the actual NGINX Ingress controller pod running in the ingress-nginx namespace.

Run kubectl get pods -o wide to see all pods and their status. Check readiness probes: kubectl describe pod <pod-name>. Test direct pod connectivity with kubectl exec -it <pod-name> -- curl http://localhost:<port>/<path> to ensure the pod responds properly.

Edit the Ingress NGINX ConfigMap: kubectl edit configmap nginx-configuration -n ingress-nginx. Add or update these values (in seconds, no "s" suffix):

proxy-read-timeout: "300"
proxy-connect-timeout: "10"
proxy-send-timeout: "300"
upstream-fail-timeout: "10s"

Save and the controller will automatically reload. Test with kubectl rollout restart deployment -n ingress-nginx.

For specific Ingress resources, add annotations without touching the global config:

apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  annotations:
    nginx.ingress.kubernetes.io/proxy-read-timeout: "300"
    nginx.ingress.kubernetes.io/proxy-connect-timeout: "10"
    nginx.ingress.kubernetes.io/proxy-send-timeout: "300"
spec:
  # ... rest of Ingress config

Apply with kubectl apply -f ingress.yaml.

For Classic Load Balancer, add to Service annotation:

metadata:
  annotations:
    service.beta.kubernetes.io/aws-load-balancer-connection-idle-timeout: "300"

For Application Load Balancer (ALB), add to Ingress:

metadata:
  annotations:
    alb.ingress.kubernetes.io/load-balancer-attributes: idle_timeout.timeout_seconds=300

Apply changes with kubectl apply -f and verify with AWS console.

If your application has slow startup, increase probe timeouts: kubectl edit deployment <deployment-name>. Update the probe configuration:

readinessProbe:
  httpGet:
    path: /health
    port: 8080
  initialDelaySeconds: 30
  timeoutSeconds: 5
  periodSeconds: 10
  failureThreshold: 3

Increase initialDelaySeconds and periodSeconds to allow proper pod startup detection.

Use Prometheus metrics or application logs to identify slow endpoints. Run kubectl exec <pod> -- curl -w "@curl-format.txt" http://localhost:8080/endpoint to measure response times. Optimize your application code or database queries if endpoints consistently approach timeout limits.