Re: postmaster startup time - Mailing list pgsql-general

From Tom Lane
Subject Re: postmaster startup time
Date
Msg-id 7477.1138846514@sss.pgh.pa.us
Whole thread Raw
In response to postmaster startup time  ("Steve Oualline" <soualline@stbernard.com>)
List pgsql-general
"Steve Oualline" <soualline@stbernard.com> writes:
> What's the longest time you'd expect between the execution of the=20
> postmaster command and being able to connec?

Normal startup is a second or two on any modern hardware.  If you have
to recover from WAL, though, it could be very long.  A rule of thumb is
that it shouldn't take longer than your inter-checkpoint time
(checkpoint_timeout) ... but that's only assuming that the most recent
checkpoint attempt before the crash succeeded.  We've seen cases where
checkpoints were failing repeatedly (eg, because of out-of-space
condition on the main data partition) but the system kept merrily
spooling transactions into WAL.  In a scenario like that, there could be
a whole lot of WAL to replay :-(

            regards, tom lane

pgsql-general by date:

Previous
From: Bob Pawley
Date:
Subject: Re: Stack Depth
Next
From: Benjamin Arai
Date:
Subject: Creating views (Multiple Databases)