Thread: BUG #1926: /etc/init.d/postgresql - script version 7.4.5 fails on startup

The following bug has been logged online:

Bug reference:      1926
Logged by:          Kevin Waldron
Email address:      kevin@zazzybob.com
PostgreSQL version: 8.0.3
Operating system:   CentOS 4.1
Description:        /etc/init.d/postgresql - script version 7.4.5 fails on
startup
Details:

I'm using Version 7.4.5 of the postgresql start script (acquired in the
postgresql-8.0.3 rpms for rhel-es-4).

On my HP DL320(G3) server, running CentOS 4.1 (further, detailed specs can,
of course, be provided if required) I have had to modify the sleep at Line
192 from 1 second TO 2 seconds.

Otherwise the test was failing as the postmaster.pid file wasn't written in
time.

Hi,

On Fri, 30 Sep 2005, Kevin Waldron wrote:

>
> The following bug has been logged online:
>
> Bug reference:      1926
> Logged by:          Kevin Waldron
> Email address:      kevin@zazzybob.com
> PostgreSQL version: 8.0.3
> Operating system:   CentOS 4.1
> Description:        /etc/init.d/postgresql - script version 7.4.5 fails on
> startup
> Details:
>
> I'm using Version 7.4.5 of the postgresql start script (acquired in the
> postgresql-8.0.3 rpms for rhel-es-4).
>
> On my HP DL320(G3) server, running CentOS 4.1 (further, detailed specs can,
> of course, be provided if required) I have had to modify the sleep at Line
> 192 from 1 second TO 2 seconds.
>
> Otherwise the test was failing as the postmaster.pid file wasn't written in
> time.

See : https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=152931

As Tom wrote, increasing the sleep time will work.

I've increased the sleep time from 1 to 2 for PGDG 8.1 beta RPMs. Let's
see what will happen in next beta.

Regards,
--
Devrim GUNDUZ
Kivi Biliþim Teknolojileri - http://www.kivi.com.tr
devrim~gunduz.org, devrim~PostgreSQL.org, devrim.gunduz~linux.org.tr
                       http://www.gunduz.org

Devrim GUNDUZ <devrim@gunduz.org> writes:
> See : https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=152931

> I've increased the sleep time from 1 to 2 for PGDG 8.1 beta RPMs. Let's
> see what will happen in next beta.

I've been using 2 in Red Hat's RPMs for some time.  The whole thing is
pretty mystifying though, because there's no way it should take any
significant amount of time for the postmaster to get to the point of
writing its pid-file.  It doesn't really do anything except parse its
command-line arguments and the postgresql.conf file before it does that.
I've certainly never been able to reproduce the problem.

It'd be useful to try running the "su" command that launches the
postmaster under "strace -f -tt" to try to gather some information
about why it's so slow.  It might be the su rather than the postmaster
per se that's at fault?

            regards, tom lane