[7590] in Release_7.7_team

home help back first fref pref prev next nref lref last post

Re: auto-update issue on w20-575-[2..5]

daemon@ATHENA.MIT.EDU (Jonathan Reed)
Thu Aug 4 10:03:30 2011

Date: Thu, 4 Aug 2011 10:03:23 -0400 (EDT)
From: Jonathan Reed <jdreed@MIT.EDU>
To: Geoffrey Thomas <geofft@mit.edu>
cc: release-team@mit.edu
In-Reply-To: <alpine.DEB.2.00.1108040058090.11792@tyger.mit.edu>
Message-ID: <alpine.DEB.2.02.1108040955430.24575@INFINITE-LOOP.MIT.EDU>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed

OK, I just kicked w20-575-2 (by rebooting from the shutdown menu on the 
login screen).  It came back up, finished the update, and is now happy. 
According to the log, the machines have been in this state since they
began the borked update on _Tuesday_ (18:23 for this machine, 16:05 for 
-3, 14:44 for -4, and 16:25 for -5).  It is as yet unclear to me what 
happened.  I'll note that w20-575-1 took those same set of updares at 
20:10 on Aug 2 with success, and -6 did so at 20:28 and -7 at 20:31.

I suppose it's worth checking if Ubuntu did something stupid like release 
another borked Update, pull it from the mirrors, and replace it with a 
working one sometime between 18:23 and 20:10 on Aug 2?

Anyway, -2 is back, but it's /var/log is available at 
/mit/joeuser/Public/var-log.tgz if anybody cares.

I'm..... not convinced we shouldn't just reboot the damn machines every 
day at 5am if nobody is logged in.  Thoughts?

-Jon


  On Thu, 4 Aug 2011, Geoffrey Thomas wrote:

> w20-575-[2..5], all natty, have all been stuck in the middle of taking 
> updates since this afternoon/early evening. w20-575-2 at least is running 
> dpkg-preconfigure, probably spinning in dpkg-preconfigure, according to the 
> kernel "show blocked tasks" magic sysrq.
>
> Nothing in any of the config (or preinst) maintainer scripts of any of the 
> packages being installed looks suspicious, or looks like it's even capable of 
> looping.
>
> I can't think of any clever way to get more debugging information. Maybe I'll 
> come up with one in the morning.
>
> -- 
> Geoffrey Thomas
> geofft@mit.edu
>

home help back first fref pref prev next nref lref last post