[5088] in testers
Re: ultra 5 panics after taking update
daemon@ATHENA.MIT.EDU (Garry Zacheiss)
Thu Mar 21 07:59:51 2002
Date: Thu, 21 Mar 2002 07:56:32 -0500 (EST)
Message-Id: <200203211256.HAA11664@never-never-land.mit.edu>
From: Garry Zacheiss <zacheiss@MIT.EDU>
To: Oliver Thomas <othomas@mit.edu>
CC: Angie Kelic <sly@mit.edu>, Garry Zacheiss <zacheiss@mit.edu>,
testers@mit.edu
In-reply-to: "[5083] in testers"
I took a look at ferd this morning. It was a victim of the
volume containing the Solaris patches not being released when Greg put
out the release. What happened was:
- ferd tries to take the update, and succeeds, but the patches
aren't in /install/patches/patches.link and they don't get
applied. The machine reboots after the update.
- During the initial reboot, the machine runs oscheck because
it's PUBLIC=true and it's version matches what's on the
packs. oscheck brings over some parts of the new kernel
that's in 9.0.25, but not all of it. The machine's software
is now in a corrupt state, but nothing notices because the
kernel is already loaded.
- A couple of days later, the machine gets rebooted for some
unrelated reason (it wasn't trying to take 9.0.26) and this
time fails to boot, because the kernel modules on disk are
inconsistant.
A PUBLIC=true machine that updated after all the volumes in the dev were
release (bagheera) didn't have this problem, so I think ferd was just in
the wrong place at the wrong time, and this isn't a systematic issue
with 9.0.25/9.0.26.
The machine is reinstalling now.
Garry
