kubectl logs

The first step of troubleshooting is the logs. kubectl logs shows a container's stdout/stderr output; you can select the pod by name or through its owner (deploy/web):

kubectl logs web-7c5b9f6d8-abcde
kubectl logs deploy/web                 # one pod of the deployment
kubectl logs -f deploy/web --since=10m  # live, last 10 minutes
kubectl logs api-xyz -c log-gonderici   # multi-container pod
kubectl logs api-xyz --previous         # the previous, crashed container

The kubelet keeps logs on the node and rotates them. When a pod is deleted or moved to another node, its logs go too. That is why in production logs are sent to a central system (in the advanced section: Fluentd, Elasticsearch, Loki). For short-lived problems, --previous still gives the last log of the previous container.

kubectl logs -l app=api --all-containers --prefix --tail=20
kubectl logs api-xyz --timestamps | grep -i error