[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
dump kills system - OpenBSD 2.9
- To: misc@openbsd.org
- Subject: dump kills system - OpenBSD 2.9
- From: John Eisenschmidt <jweisen@eisenschmidt.org>
- Date: Sat, 30 Mar 2002 03:54:45 +0000
- Content-Disposition: inline
- Mail-Followup-To: misc@openbsd.org
- User-Agent: Mutt/1.2.5i
I posted this earlier today from work, but I suspect it doesn't look like it
made it out to the list. Apologies if you've already received this.
I'm running OpenBSD 2.9 on an AMD k6-2 450, 3Ware 6200 IDE RAID
controller with (2) IBM 40GB Deskstars in a RAID-1 config, and a
Seagate DDS-3 Tape Drive. I've had the tape drive about six months,
and when I first installed it I tried to get dump working and I killed
the system. I thought it was something I did, so I've been doing
backups using tar (which is a huge PITA in my opinion).
In the last couple weeks I've gotten pretty familiar with ufsdump on
Solaris, so I thought I'd switch to using dump at home. Now, stupid me
is trying this from work while my box is at home. I do one of these:
/sbin/dump -0ua /dev/nrst0 /usr
And I get one of these (this is obviously an approxomation):
DUMP: read error from [the raw device under /usr] INPUT/OUTPUT ERROR
[block number]
I do a control-c to stop dump, it asks "Errors were found - do you
want to abort the dump? (Y/N)" and I say 'Y'. Except by the time I hit
'Y' it's already dead and dropped my connection. This is the same
thing that happened the first time I tried this.
When I use tar, it's usually something like this:
mt rewing
tar -cfv /dev/nrst0 /usr/foo
tar -cfv /dev/nrst0 /usr/bar
tar -cfv /dev/nrst0 /usr/baz
...
mt rewind
eject /dev/rst0
These filesystems fsck fine (except when I kill the machine) and
softupdates are enabled. My *only* thought is the possibility that
dump doesn't like/behave with the 3Ware driver? But would that jive
considering tar does?
Any light someone can shed on this would really be appreciated.
UPDATE: I got home and I brought the system back up. I didn't realize I needed
to do a 'boot crash' to get the system to write a core dump. Interestingly
enough, there was no message when I booted about savecore *not* finding a
core, but there is nothing in /var/crash
root@kaitain$ dmesg | grep core
root@kaitain$
I'd be willing to recreate the problem and generate a core for someone to look
at. I'm not above killing my system.
BTW: A million 'thanks' for softupdates. It's so nice when you whack you
system not to have applesauce for filesystems.
--
John W. Eisenschmidt <jweisen@eisenschmidt.org>
Homepage URL | http://www.eisenschmidt.org/jweisen
PGP Public Key | http://www.eisenschmidt.org/jweisen/misc/jeisenschmidt.asc
PGP Fingerprint | 5F9B F916 5AD1 3295 CF99 BC1E 1F97 E6A3 37E3 BEF2
[demime 0.98d removed an attachment of type application/pgp-signature]