On 9/14/07, nus4u@gmx.net <nus4u@gmx.net> wrote:
> > Sounds like postgresql is being started, and then the /tmp directory
> > is being cleared of old files by your OS. What's the Sxx number on
> > the startup file in /etc/init.d/rcx.d that you're using (assuming SYSV
> > init)
>
> Thanks Scott !
>
> I change the location of the socket file and it isn t deleted anymore !
> Until now i am not quite sure which script deletes the socket files, i
> guess it is "tmpwatch" which is called from cron.hourly, but disabling
> and restarting didn t solve the problem.
> In rc3.d it has the Numer S96postgres, after this script only
> sendmail,rhnsd,haldaemon,messagebus and local are executed, but
> none of them will delete tmp files.
> Any idea which and where the script is started who deletes tmp files ?
> System is a Red Hat ES 4 !
On my laptop running fc4, there's this line in rc.sysinit:
rc.sysinit:rm -f /tmp/.X*-lock /tmp/.lock.* /tmp/.gdm_socket /tmp/.s.PGSQL.*
If your startup script runs before a line like that one above, you'd
get the kind of failure you're talking about.