Worker cannot connect to broker due to network policy or DNS issues

Workers have correct credentials but cannot reach the message broker. Common causes include network policies blocking traffic, pods in different namespaces, missing service endpoints, or service name typos. Results in connection refused errors.

Test connectivity from worker pod using nc -zv. Verify service exists and has endpoints using kubectl get svc and kubectl get endpoints. For cross-namespace access, use full DNS name (service.namespace.svc.cluster.local). Check if broker pods are running and service selector matches pod labels.