Bug#396944: exim4-daemon-light: upgrading causes panic

[Bruce Sass]

On Sat November 4 2006 05:09, Marc Haber wrote:
> On Sat, Nov 04, 2006 at 04:40:01AM -0700, Bruce Sass wrote:
> > On Sat November 4 2006 02:29, Marc Haber wrote:
> > > On Fri, Nov 03, 2006 at 01:06:01PM -0700, Bruce Sass wrote:
> > > > ...would seem to indicate that it is a transitory situation
> > > > caused by the upgrade process.
> > >
> > > Probably. Did killing the existing daemon manually help with
> > > starting the new daemon?
> >
> > I did not need to restart exim4. I saw the message in the INBOX
> > associated with /var/mail/spool, then ran the ps|grep command you
> > saw the output of.
>
> Anyway, you had an extra daemon running which interfered with your
> update.

I don't think that was the case, and know it wasn't when I had the two 
failures out of nine reinstalls mentioned in the last message.

> > The mainlog just prior to the paniclog entry I included shows
> > periodic Start and End queue run messages, 10 socket bind()
> > failures at 30s intervals, then more Start/End queue runs.
>
> Yes, that's exim4's default behavior if it cannot bind to any
> configured smtp listening port.

Ya. I only mentioned it so you would know I've looked at the logs and 
didn't see anything unusual.

> > > > The result is daily...
> > > >
> > > > -e Subject: exim paniclog on onegee has non-zero size
> > > > To: root
> > > >
> > > > exim paniclog /var/log/exim4/paniclog on onegee has non-zero
> > > > size, mail system might be broken
> > > >
> > > > ...messages when there is no problem.
> > >
> > > Yes, you'll need to rotate the paniclog away manually. See
> > > /usr/share/doc/exim4-base/README.Debian.gz chapter 2.5.1
> > >
> > > If the message is continuously being written again and again over
> > > and over, you have probably still an old daemon running.
> >
> > I understand that, but it doesn't seem right to have to do it after
> > an upgrade if all is well... hence the report.
>
> Yes. I am desparately trying to find out what went wrong on your
> system as the issue is not reproducible here and no other users are
> reporting this.

The only possibility I can think of is the daemon not actually stopping 
during the upgrade. Feel free to ask me to muck about with it if you 
have ideas.

> > Unfortunately I can not reliably reproduce this bug. Nine attempts
> > at: #apt-get --reinstall install exim4-daemon-light
> > got me two failures initially then seven successes. I have another
> > box running the light daemon and one running the heavy daemon,
> > neither of them have had a panic.  :-/
>
> I am afraid that then there is nothing to do except tagging this bug
> unreproducible and closing it by the end of November.

Sounds reasonable.


- Bruce