[7371] in testers
Re: linux shutdown hang
daemon@ATHENA.MIT.EDU (Bill Cattey)
Fri Sep 16 15:54:23 2005
From: Bill Cattey <wdc@MIT.EDU>
To: Jacob Morzinski <jmorzins@mit.edu>
Cc: testers@mit.edu
In-Reply-To: <w6mk6hgaltc.fsf@kamp-krusty.mit.edu>
Content-Type: text/plain
Content-Transfer-Encoding: 7bit
Date: Fri, 16 Sep 2005 15:54:10 -0400
Message-Id: <1126900450.20751.2.camel@tokata.mit.edu>
Mime-Version: 1.0
Good catch.
Lou Isgur happened to be visiting when I read this message.
He says that he doesn't see this as a problem. People who see a powered
off system after a power failure will probably just turn it back on.
In the W92 Test Cluster, I played with the HP system.
It turns out that in the BIOS setup there's an advanced option
that controls behavior after a power fail. I toggled this bit
and sure enuf, the system would power itself back on.
Bottom line:
We can change this if it becomes a problem.
We should pay attention to see if it becomes a problem.
At the moment it does not seem to be a problem.
-wdc
On Fri, 2005-09-16 at 14:50 -0400, Jacob Morzinski wrote:
> (Jonathon Weiss) writes:
> > Just FTR this seems to be a moderately common problem. I power-cycled
> > 4 or 5 machine with this problem in the w20 cluster today, and saw it
> > on SIPBs GX280 (cj) earlier this week.
>
> FWIW, I also power-cycled about a half-dozen w20 linux machine
> with that problem, last weekend (Sunday Sep 9).
>
>
>
> While power-cycling, I noticed another issue, that will affect us
> if we ever have a large power outage: the HP (linux) machines
> don't power back up, if they lose power.
>
> I tried pulling machine's power cords and plugging cords back in
> to reboot the machines. They didn't power up when the cord was
> plugged back in. From this, I conclude that someone will have to
> walk through clusters turning machines back on, if we ever have a
> power outage.
>
>
> -Jacob
