Working out why something is not running.
WHAT TO CHECK FIRST
The pod's status.
WHAT COMMON STATUSES MEAN
- Pending: not yet placed on a node
- Container creating: pulling the image or attaching storage
- Crash looping: starting and exiting repeatedly
- Image pull failure: the image cannot be fetched
- Terminated for memory: it exceeded its limit
WHAT TO READ NEXT
The events for that pod.
WHY EVENTS
They explain scheduling, pulling and mounting failures that logs never show.
WHAT PENDING USUALLY MEANS
Insufficient resources, or storage that cannot be attached.
WHAT IMAGE PULL FAILURE USUALLY MEANS
A wrong name or tag, or missing registry credentials.
WHAT CRASH LOOPING REQUIRES
Reading the logs of the previous attempt.
WHY THE PREVIOUS ATTEMPT
The current one may not have produced output yet.
WHAT TO CHECK FOR A SERVICE RECEIVING NO TRAFFIC
Whether it has endpoints Whether labels match Whether pods are ready
WHAT TO CHECK FOR NETWORK PROBLEMS
Resolution from inside a pod Whether network policies block it
WHAT TO USE FOR INTERACTIVE DIAGNOSIS
A shell in the pod, or a temporary debug pod.
WHAT TO CHECK ABOUT NODES
Whether they are ready, and under what pressure.
WHAT TO DO IN ORDER
Status, events, logs, then inside the container.
