OSEC

Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com
 
Subject: Re: "Device not configured" error message
From: Wietse Venema (wietseporcupine.org)
Date: Mon Jul 03 2000 - 18:42:40 CDT

Bill Sidhipong:
> Wietse Venema wrote:
> >
> > Bill Sidhipong:
> > > Jun 30 17:22:41 oahu postfix/master[18188]:
> > > warning: master_wakeup_timer_event: service public/qmgr:
> > > Device not configured
> >
> > This means that neither the Postfix master process nor the Postfix
> > queue manager process listen on the FIFO at:
> >
> > /var/spool/postfix/public/qmgr

> Okay, next time when this problem occurs, I'll know where to look,
> thanks :)
[...]
> Hmm, this is standard kernel 2.2.15 + NFS patches that shouldn't affect
> FIFO ... is there some test I can perform to help you narrow this problem?

This condition cannot happen. The only way for the master server
to close the qmgr FIFO is when it receives SIGHUP and the qmgr is
service removed from the master.cf file.

Impossible things happen when you are using flakey hardware (did
you enable memory error detection?) or flakey software: compiler,
library, kernel.

It helps to use proven solid software, that is, software that did
not need to be replaced in the past year.

> It *might* have something to do with delivering mail to an SMTP host that's
> not listening on port 25.

If that is the case, then your machine is suffering from some major
problems. Problems in the SMTP client should not even remotely
affect the operation of the Postfix master/qmgr FIFOs.

> I recall seeing this error when our ISP's SMTP
> server went offline a few days earlier. Prior to this, postfix had been
> running perfectly for *a few months*.
> ^^^^^^^^^^^^
> Then, thinking it was a problem in postfix that perhaps had been fixed, I
> upgraded postfix and tested sending out to my server at home (which would
> have been blocked by my firewall). A few days later, postfix generated
> this error message. It could be a coincidence, of course =:-)

Very, very, funny. Most likely, the problem is in the parts that
you did NOT replace.

Why do people think Postfix has a bug when it stops working? Don't
you realize there's megabytes and megabytes of code sitting underneath
Postfix in library routines and in kernel routines?

        Wietse