Boot order encoded as three priority numbers

Three programs, one chain: Postgres needs to be up before Redis, Redis before the API. If the API starts first and can't reach the cache, it crashes into BACKOFF and burns retry budget for thirty seconds.

You set priority: 1 on the database, priority: 5 on the cache, priority: 10 on the API. You set boot: true on all three. The container reboots. Postgres starts first, accepts connections. Redis starts second, warms its cache. The API starts last, finds both dependencies healthy, settles into RUNNING.

This is the supervisord priority field, exposed as a JSON property in a POST body. No init containers, no readiness probes, no Helm chart, no entrypoint script that re-implements wait-for-it.sh.