Pgpool-II, reviews?

Question

PostgreSQL

Hello!
Who uses pgpool-II in production, how does it feel? what configurations are you using? what are the rake?

Answer 1

of those "jambs" that I found:
- in the balancing mode, when the slave falls, pgpool can start ignoring the master))) as a result, the master is alive, but pgpool does not send requests to it; whether this is an oddity in health_ and sr_ checks, or something else ...
- inefficient connection pooling, you can’t push more clients into it than max_connections in postgresql.conf (you can do this in pgbouncer); those. it’s sort of like pulling, but inside one worker =)
- I didn’t understand how to hook pgpool to postgres via a unix socket, it doesn’t see it and that’s it ... dog , everything seems to be trivially solved with one line of the config, but no zhe ... all backends are down
- all the time he tries to make me an auto-filer when the slave fails, ... well, I'm not a supporter of automation when raking up an accident with databases ...
so far I made it work as I need, but at what cost ... keepalived + haproxy + pgbouncer and some kind of makara on application side would be better