Website container lifecycle not owned by ah reconciler

[dennisonbertram]

Problem

The landing page website container was deployed via direct Docker API calls rather than through the ah API. As a result:

• The reconciler detects it as a "split-brain" situation
• When the container exits, ah does not restart it (no ownership)
• Manual intervention required to bring it back up

Expected behavior

All running containers should be deployed via the ah API so the reconciler fully owns their lifecycle (restart, health checks, GC decisions).

Fix

Re-deploy the website through the proper ah API flow so it's tracked in the database with a valid container ID. Ensure the reconciler loop handles the "restart crashed service" case correctly.

Labels

bug, reliability