Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!mit-eddie!aryeh
From: aryeh@eddie.mit.edu (Aryeh M. Weiss)
Newsgroups: comp.unix.xenix
Subject: Re: Bug in spooler on Xenix (fix here)
Message-ID: <1990Aug18.212631.3958@eddie.mit.edu>
Date: 18 Aug 90 21:26:31 GMT
References: <1672@ispi.COM>
Reply-To: aryeh@eddie.MIT.EDU (Aryeh M. Weiss)
Distribution: comp
Organization: MIT EE/CS Computer Facilities, Cambridge, MA
Lines: 21

In article <1672@ispi.COM> jbayer@ispi.COM (Jonathan Bayer) writes:
>
>There is a bug I have seen mentioned in the Xenix spooler recently. 
>This is that the spooler seems to stop working on a random basis.  This
>is caused by a race condition in the spooler.  SCO gave me a workaround
>to this a while ago:

Sorry, I just don't believe that this will cure the problem in the
long term.  I have a theory that the real problem is lpsched failing
to remove a certain lockfile, /usr/spool/lpd/lock.  This file gets
created when lpsched has started a specific print job, and gets
removed when that job is done.  (I can accept that a race condition
will result in this file not getting removed.)  It seems that when the
job at the top of the queue gets canceled, lpsched *sometimes* hangs.
I have had the experience of shutting down and restarting lpsched,
only to have it remain hung.  Unfortunately I didn't know about this
lockfile: I had been looking for a lockfile in /usr/spool/lp.  (The
rc.d scripts remove this file --- thus rebooting the system solves the
problem!)

--