Demo / hello (push) Successful in 18s
With a 150m CPU limit, pg_isready routinely exceeds the 5s probe timeout under load (e.g. Gitea's startup migration check), so the kubelet kills postgres mid-query. This is the cause of the long-running restart loop on gitea-postgresql-0. Raise the limit to one core (request 250m). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>