Re: hot(un)plugging IDE disks

Francois-Rene Rideau (fare@tunes.org)
Fri, 12 Nov 1999 02:24:20 +0100

On Fri, Nov 12, 1999 at 12:58:27AM +0000, Alan Cox wrote:
> > PS: with racks come the want of being able to have a linux disk that be
> > bootable independently from its IDE *and* BIOS numbers. I see no way
> > to nicely satisfy _both_ constraints in the way things are done currently.
>
> One of the boxes that passed briefly past me when I did some work on it
> was a rack PC with swappable IDE disk that could boot off /dev/hdc if
> hda was not present. Very nice and with a little lilo hackery it means it
> can boot either disk on a failure.
Sure. I do it the same, too (and also have a lilo bootsector on floppy with
a map on hdc, so as to be able to boot on hdc even if hda is unfriendly).

> We support scsi hot swap. If your bays do. Don't try it otherwise. You can
> blow the terminator fuses and worse
>
Hum. I never had any problem with my old (external) SCSI2 drive on an SC200,
but obviously YMMV. How do you determine that bays at hotswappable?

A crash test with an old drive on that IDE rack
indicates the drive copes well with hot(un)plugging
(using hdparm -f -Y /dev/hda before unplugging).
The remaining problem to me is about convincing the kernel
to re-read the drive's geometry and information, after re-plugging
(and no, I can't just rmmod and insmod the IDE subsystem,
since it's necessary to support the root partition on /dev/hdc).

PS: as for something completely different; I suppose this is a known bug,
but while using mutt 1.0i over NFS, I could kind of hang up the whole
2.2.{12,13,13ac1}'s VFS subsystem while mutt tried to lock the NFS'ed
~/Mail/received. During the hang up, processes wouldn't run,
but VT switching and IP pinging and forwarding would behave normally.
Killing mutt with SAK would have the machine resume normal operation.
Latest debian potato on both machines, with rpc.nfsd on 2.2.13 on the server
(oops; I'll be retrying with knfsd shortly).

[ "Faré" | VN: Уng-Vû Bân | Join the TUNES project! http://www.tunes.org/ ]
[ FR: François-René Rideau | TUNES is a Useful, Nevertheless Expedient System ]
[ Reflection&Cybernethics | Project for a Free Reflective Computing System ]
Obviously, a man's judgement cannot be better than the information on which he
has based it. Give him the truth and he may still go wrong when he has
the chance to be right, but give him no news or present him only with distorted
and incomplete data, with ignorant, sloppy or biased reporting, with propaganda
and deliberate falsehoods, and you destroy his whole reasoning processes, and
make him something less than a man.
-- Arthur Hays Sulzberger

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/