The container starts, exits, and Kubernetes backs off before restarting it. The pod is a symptom the cause is almost always in the previous container log or the probe config.
kubectl get pods shows CrashLoopBackOff with a rising restart count. Backoff doubles up to five minutes, so late in an incident the pod looks frozen rather than crashing.
Exit 1 — the application threw during startup. Read the previous log.
Separate liveness from readiness, and never point liveness at an endpoint that touches the database.