prepared_statements: false. So thanks for that as well! :)
prepared_statements: falsemakes pg_stat_statements useless because virtually every query is unique.
pgbouncer -Rwill create a new process with a new pid. Perhaps you want the
%pin my PROMPT1 in psql, when I connect to a pgbouncer, I get gibberish, but not the same gibberish every time, i.e. it shows a negative number that's kinda large. Could psql be seeing things it shouldn't be seeing by looking at a location that's neither zeroed out nor populated correctly with the backend pid? I haven't yet figured out what would populate that bit in the pgbouncer code.
restart). If I run
pgbouncer -Rmanually, systemd goes crazy about the main PID changed: drops the pidfile and doesn't recognize the new one if I recreate it. Systemd service file is generated, type=forking. For this type I don't think there is a way to notify systemd when PID changes. I'm thinking of writing a wrapper script that would appear as the daemon for systemd, will restart pgbouncer when it receives certain signal and will propagate other signals pgbouncer. It looks ugly and also I have a feeling of reinventing the wheel. I don't suppose anyone knows a better or just ready solution?
statement_timeoutis a PostgreSQL setting, not something PgBouncer deals with.
no more connections allowed (max_client_conn)on 800 connections
Hello, i want to ask
Is it any chance to hide outputting db credentials when starting pg as a docker image?
docker run -ti -e DATABASES_HOST=testhost -e DATABASES_DBNAME=testdb -e DATABASES_USER=user -e DATABASES_PASSWORD=pass pgbouncer/pgbouncer
there exist QUIET option but it also hide logs from pg service