Re: GMT FATAL: remaining connection slots are reserved for non-replication superuser connections, but I'm using pgBouncer for connection pooling

Hi Steve, thanks for your help.
Your comment made me realise that maybe the problem is my pgBouncer
configuration, specifically default_pool_size. It took me a while to
understand pgbouncer, and I still had some doubts when I configured it. Now
I undesrtand better.

I connect to all databases with the same user. However, I had set
default_pool_size=10. So with more than 150 databases, it was very probable
that postgresql reach max_connections=250 limit.

I didn't have set reserve_pool_timeout or max_db_connections, but docs say
their default values are reserve_pool_timeout=5 seconds,
max_db_connections=unlimited.

I've reviewed pgBouncer configuration and did some search. I've found this
thread where the first person that responds gives a great explanation abount
how pgbouncer do its maths:
https://github.com/pgbouncer/pgbouncer/issues/174

So, what I did for now was to set this in pgbouncer configuration:
default_pool_size=1
min_pool_size = 0
server_idle_timeout = 30
reserve_pool_size = 10
reserve_pool_timeout = 5
max_db_connections = 5

I'll keep monitoring with this setup, but I can already tell you that the
total number of connections in postgres has significantly reduced (from ~100
to ~60). So I guess the problem was a bad setup of pgbouncer.

Thank you all for your help!

--
View this message in context: http://www.postgresql-archive.org/GMT-FATAL-remaining-connection-slots-are-reserved-for-non-replication-superuser-connections-but-I-m-g-tp5946245p5946827.html
Sent from the PostgreSQL - general mailing list archive at Nabble.com.