From: | Stefan Kaltenbrunner <stefan(at)kaltenbrunner(dot)cc> |
---|---|
To: | Phoenix Kiula <phoenix(dot)kiula(at)gmail(dot)com> |
Cc: | pgsql-general <pgsql-general(at)postgresql(dot)org> |
Subject: | Re: Postgresql performance in production environment |
Date: | 2007-08-19 09:35:37 |
Message-ID: | 46C80EE9.3020805@kaltenbrunner.cc |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
Phoenix Kiula wrote:
> On 19/08/07, Phoenix Kiula <phoenix(dot)kiula(at)gmail(dot)com> wrote:
[... ]
> Well based on some past posts, I looked into my pg_log stuff and found
> a number of these lines:
>
>
> [----------------
> LOG: could not fork new process for connection: Resource temporarily
> unavailable
> LOG: could not fork new process for connection: Resource temporarily
> unavailable
> LOG: could not fork new process for connection: Resource temporarily
> unavailable
> LOG: could not fork new process for connection: Resource temporarily
> unavailable
> LOG: could not fork new process for connection: Resource temporarily
> unavailable
> LOG: could not fork new process for connection: Resource temporarily
> unavailable
> ----------------]
>
>
> Which suggests that our guess of running out of connections is the right one.
>
> So, we have three options (to begin with) --
>
> 1. Increase the number of max_connections. This seems to be a voodoo
> art and a complex calculation of database size (which in our case is
> difficult to predict; it grows very fast), hardware, and such. I
> cannot risk other apps running on this same machine.sql
this error is a sign that the OS(!) is running out of resources(or at
least won't allow pg to fork another process) - either you hit an ulimit
for the user postgresql runs under or you need to flip some kernel
setting to increase the number of processes. increasing max_connections
wil NOT help because you are not even hitting the current one yet ...
>
> 2. Use connection pooling. I've found pgpool2 and pgbouncer from the
> Skype group. Does anyone have experience using either? The latter
> looks good, although we're usually skeptical about connection pooling
> in general (or is that just the mysqli_pconnect() hangover?)
pgbouncer works quite fine here.
Stefan
From | Date | Subject | |
---|---|---|---|
Next Message | Magnus Hagander | 2007-08-19 09:38:46 | Re: server closed the connection unexpectedly |
Previous Message | Magnus Hagander | 2007-08-19 09:35:14 | Re: Postgresql performance in production environment |