I have a multi-master replication setup using Postgres and Bucardo. On top of it, I have a Django application.
The setup does not fully works since I keep getting primary-key collisions. Since the replication is configured to clone the whole database, and thus I'm somewhat confident that Postgres' sequence ids are being updated, I wonder if it is not Django that keeps somewhere in memory the last sequential id number. If so, does anyone have any hint on what could be done to fix this issue?
Thanks in advance