Scaling Replication Controller doesn't work

[ethernetdan]

Under the current configuration if the replica count on the RC is increased and persistent volumes are not being used (at least on GCE, it couldn't get a read-write mount) any additional replicas will create an empty database on startup. Traffic would be load balanced between the instance with an empty database and the one without one.

[hobti01]

You will not be able to share the database data by attaching the same persistent disk / persistent volume to multiple pods. GCE will not allow read-write access by multiple writers, and this carries through to Persistent Volumes.

Is there a specific reason that you've moved to a single container with app + db instead of keeping the db as a separate container in its own Pod? I think that you will discover that auto-scaling Postgres is non-trivial. :-) If you (re)separate the app and db, then you'll be able to increase the app replicas.