Re: Postgres server output log - Mailing list pgsql-admin

From Tom Lane
Subject Re: Postgres server output log
Date
Msg-id 20466.1043948825@sss.pgh.pa.us
Whole thread Raw
In response to Re: Postgres server output log  (Steve Crawford <scrawford@pinpointresearch.com>)
Responses Re: Postgres server output log  (Bruno Wolff III <bruno@wolff.to>)
List pgsql-admin
Steve Crawford <scrawford@pinpointresearch.com> writes:
> I think configuring PostgreSQL to write to syslog is the correct way to go.

syslog is a good alternative, but not perfect.  I know of two strikes
against it:

1. It can't catch everything that could go to a stderr log.  While I
think we've cleaned up all the backend code (there used to be lots of
random fprintf's), there are messages that are not under our control.
In particular, on most platforms any dynamic-linker error messages go
to stderr --- when you're trying to debug why the backend can't load
a .so file, this info is invaluable, but syslog can't capture it.

2. On some platforms, syslog is reputed to drop or truncate messages
under heavy load.

You have to figure out whether the many advantages of syslog outweigh
these problems for your use.

            regards, tom lane

pgsql-admin by date:

Previous
From: Oliver Elphick
Date:
Subject: Re: Postgres server output log
Next
From: "guillermo schulman"
Date:
Subject: PostgreSQL Replication/Clustering solution