Why should you care?

PostgresVerse APIs scale pods horizontally; database is shared tier with migration job per deploy not manual psql in prod.

See it live — copy this example

Run in pgAdmin or psql.

-- Migration job pattern (SQL)
CREATE TABLE schema_migrations (
  version text PRIMARY KEY,
  applied_at timestamptz DEFAULT now()
);
INSERT INTO schema_migrations (version) VALUES ('20260719_orders_idx');

-- App health: SELECT 1 with short timeout
SELECT 1 AS ready;

What happened?

• schema_migrations tracks applied Flyway/Liquibase versions.
• Health probe SELECT 1 verifies pool connection.
• K8s liveness separate from DB readiness when using degraded mode.

Practice next

1. Store DATABASE_URL in K8s secret from vault.
2. Run migration Job before Deployment rollout.
3. Configure readinessProbe calling /health/db endpoint.
4. Init container waits for migrate job completion.
5. Read replica URL for GET endpoints in ConfigMap.

Remember

Migrations as CI/K8s Job once per release. Pooler required for many pods. Secrets rotated without image rebuild.

PostgresVerse K8s rollout

Helm chart runs migrate then rolls 50 API pods; zero failed health during deploy.

Outcome: Continuous delivery ships daily without schema drift.