Read replica connection pool
We are anyway facing a weird problem. We have 2 instances running connected to the Session pooler, we route some queries to the primary and some to the replica.
The pool connection pool configuration on our side looks like this:
Primary maximum pool size: 30 slots
Read replica maximum pool size: 15 slots
But when we check the pooler to database connections, we see stable primary one (2 instances with max 30 connections + other tools around 70-80), but the number of connections reported by the read replica is between 130-140, which doesn't make sense because we have half of slots available for this datasource.
How can this be explained? Is the primary DB balancing to the replica without us knowing?
1 Reply
I doubt users here are going to be able to answer this.
I would try support and maybe a question in supabase/supavisor github.
But who knows, never know when a user will pop up and answer something.