Containers, Kubernetes & Platform Engineering

A ConfigMap change can update mounted files but not environment variables, or an application may read configuration only at startup. A partial rollout can leave pods using mixed settings. What to check Inspect ConfigMap versions, pod environment and mounts, rollout status, application reload behavior, and logs. Determine whether all replicas restarted...

Traffic can reach pods that are not ready if probes are absent or misconfigured, readiness gates are ignored, or a proxy uses a separate endpoint list. Users may see intermittent errors during startup. What to check Inspect readiness probes, EndpointSlices, pod conditions, rollout events, service mesh routing, and actual request logs. Check whether a...

A namespace change alters service discovery names, policies, secrets, role bindings, and resource quotas. Moving a workload can leave references pointing to the old namespace. What to check Search configuration and manifests for old namespace references, inspect service names, NetworkPolicies, RBAC, secrets, ingress, and monitoring labels. Compare...

A pod may resolve a service in its own namespace but fail to reach a service elsewhere because of an incomplete DNS name, network policy, service mesh, or authorization rule. What to check Check fully qualified service name, namespace, endpoints, network policies in both namespaces, mesh authorization, and pod events. Test DNS and connection from an...

Secret rotation can temporarily leave pods with different credentials or applications using stale connection pools. A rolling update may be incomplete or reference the wrong key. What to check Check secret version, pod rollout, mounted file refresh, environment variable behavior, connection pool lifetime, and authentication errors. Compare healthy and...

A changed ConfigMap or Secret may not update an application’s in-memory settings, while a bad value can cause startup failures and repeated restarts. What to check Inspect pod events, previous container logs, rollout history, probes, mounted configuration, and application validation messages. Compare the current revision with the last healthy...

Users may see intermittent failures if readiness probes report success before an application can serve requests, or if endpoints do not reflect the expected pod state. What to check Inspect readiness probe path and timing, service endpoints, pod conditions, startup duration, and application logs during rollout. Compare healthy and failing replicas....

A Job can exit successfully while processing no records because of an empty input, wrong namespace or selector, permissive exit handling, or output written to an unexpected destination. What to check Review job arguments, input volume, environment, exit code, logs, service account, and output path. Confirm the job ran against the intended dataset and...