On 21.09.05 22:09:41, Martin Pitt wrote:
> Hi Andreas!
>
> Andreas Pakulat [2005-08-23 23:59 +0200]:
> > Tue Aug 23 22:18:12 2005: Starting PostgreSQL 8.0 database server: mainThe
> > PostgreSQL
> > server failed to start. Please check the log output:
> > Tue Aug 23 22:18:18 2005: ^[[31mfailed!^[[39;49m
>
> Hmm, very odd... I cannot reproduce it here, but I just committed an
> improvement which could help
>
> * pg_ctlcluster:
> - Increase the timeout for the started postmaster to 20 seconds to cope
> with slow startup.
> - Immediately fail if the postmaster produced log output and does not run
> any more. This avoids unnecessary timeouts on configuration errors and
> the like.
>
> For testing this, could you please edit /usr/bin/pg_ctlcluster, search the
> line
>
> for (my $attempt = 0; $attempt < 10; $attempt++) {
>
> and replace the "10" by a "40"? This will increase the timeout to 20 seconds.
As I said, the "2nd" start of PG now doesn't fail, but the one during
boot-time does.
If this really is a timeout-thing, I think I know why it might happen
here:
I changed kdm to start earlier as that takes some time, so kdm is
started as 14th in runlevel 2, while PG is 20. So it could be that the
startup of KDE (which is quite hdd-intensive) makes PG take quite some
time to come up. (Did I tell you that this is a notebook hdd with only
4500 RPM?)
If this is indeed the problem an option to increase this timeout in
pg_ctlcluster would be really nice - for example in
/etc/default/postgresql-server-8.0 or so.
Andreas
--
You will meet an important person who will help you advance professionally.
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
