[4812] in testers
9.0.6->9.0.7 update failure on byte-me
daemon@ATHENA.MIT.EDU (Mitchell E Berger)
Fri Jun 15 15:48:48 2001
Message-Id: <200106151948.PAA05015@byte-me.mit.edu>
To: testers@MIT.EDU
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Date: Fri, 15 Jun 2001 15:48:43 -0400
From: Mitchell E Berger <mitchb@MIT.EDU>
Bob and I witnessed a strange failure mode on byte-me, an Ultra1 taking
autoupdates. It took the 9.0.7 update sometime between 2am and 3am Thursday
morning, and did not come back to the point where I could log in from home.
It ended up with a blank white console (a cursor appeared if you hit enter,
but didn't do much). Finally hitting Ctrl-C resulted in a slew of messages
going by stating that they were recursively removing part of the packs I had
copied local to work on the Sawfish focus issues under
/var/server/usr/athena/share-local. In the middle of this, X respawned, the
Athena login screen came up, and the update completed in the little console
window. It even rebooted when it was done. We're fairly sure the failure
occurred during mkserv (byte-me is just mkserv remote), and unfortunately,
according to /var/athena/update.log, nothing went wrong - it looks like a
perfectly normal log.
We tried having it take the update again by editing /etc/athena/version, and
the problem didn't repeat. I tried copying the same piece of the packs to the
same place and editing /etc/athena/version, and the problem also didn't repeat.
Separately, right after coming up, attempting to log in results in a "Waiting
for workstation to finish activating" message for quite a while. I think Bob
is working on fixing this, and that it relates to reactivate.pid not getting
cleaned up on boot.
Also separately, my SGI monitor is powersaving while xscreensaver runs. I do
have a .xscreensaver, but Bob and I didn't notice anything but one change to
the default (such that I can't be logged out after 15 minutes). I'll try it
without the .xscreensaver today and tell you if it still does this.
Mitch
