Azure Database for PostgreSQL: Unexpected PgBouncer Restarts Outside Maintenance Window

Problem description

I am experiencing unexpected restarts of the built-in PgBouncer layer on my Azure Database for PostgreSQL flexible server. These restarts occur outside the scheduled maintenance window and are causing significant service disruptions, including disconnection of clients and a sharp drop in connection counts.

Environment

Azure Database for PostgreSQL flexible server with built-in PgBouncer enabled, located in an unspecified region.

What I've already tried

I have enabled and reviewed PgBouncer diagnostic logs through Azure Monitor, counted and timestamped process-up events, and compared these with PostgreSQL logs. I also checked server metrics around restart times and confirmed that no server-level operations such as restart, update, or failover occurred during these events.

Current status

I am seeking guidance on whether such PgBouncer restarts are expected outside planned maintenance, the root cause of these restarts, and recommended steps to prevent or mitigate their impact.