API changes: admin limit overrides, probe timeouts, db_engine, replicas

New in the API: per-app limit overrides for admins, a configurable health-check timeout, the database engine field, and replica counts.

A batch of API changes shipped together.

For everyone

  • healthCheck.timeoutSeconds (1 to 90) for slow-starting apps. Blank keeps the defaults, and the probe period is raised to at least the timeout.
  • resources.db_engine (mysql or pg), also returned as db_engine on the app.
  • replicas is now visible on the app.

For admins

PATCH /admin/apps/{id}/resources overrides limits per app: CPU, memory, scratch disk, database connections, database and storage quotas, image size, and outbound bandwidth (1 to 160 Mbit/s). Raising a quota re-checks usage right away, so a read-only database or bucket is restored without waiting. The response says whether the change rolls the pods. Replica count (1 to 3) has its own admin endpoint and is quota-aware. The admin UI has a matching limits dialog.