Kubernetes Pod Keeps Restarting Even Though CPU and Memory Are Below Limits
A practical playbook for diagnosing Pod restarts using exit codes, previous logs, probes, OOM evidence, node pressure, eviction, storage, and rollout history.

A Kubernetes Pod can restart repeatedly even while CPU and memory remain below limits. The cause may be an application exit, probe failure, node pressure, eviction, storage, configuration, dependencies, or controller-driven replacement.
A Kubernetes Pod can restart repeatedly even while CPU and memory remain below limits. The cause may be an application exit, probe failure, node pressure, eviction, storage, configuration, dependencies, or controller-driven replacement.
Core principle
Restart count is a symptom. Use the last termination state, previous logs, events, node conditions, and controller history to prove who stopped the container and why.
Each investigation step should retain timestamps, Pod UID, container ID, termination state, events, and logs.
Each investigation step should retain timestamps, Pod UID, container ID, termination state, events, and logs.
Each investigation step should retain timestamps, Pod UID, container ID, termination state, events, and logs.
Each investigation step should retain timestamps, Pod UID, container ID, termination state, events, and logs.
Each investigation step should retain timestamps, Pod UID, container ID, termination state, events, and logs.






