SCSI-controller reports wrong disk size

Gardener · 08-13-2002, 12:29 PM

I've just upgraded my server to an AMD Athlon Thunderbird 1200 and also a new scsi card (since my old m/b had one built in). Everything worked great at first but quite soon everything went straight to h*ll.

For some reason my Adaptec SCSI card seems to think that my first 4.3 GB disk is 9105MB and usually doesn't find the second one at all. Because of this the partition sizes in Linux gets totally screwed up.

When the second disk is found I get error reports indicating that the disk is corrupted. But since this happened at the same time as the upgrade and the first disk is recognized with the wrong size I'm guessing (or hoping...) that there is some configuration error with the scsi card.

If anyone can help me with this I would be very happy. I've been googling the net for a couple of days and tried just about everything I can think of and now I'm completely stumped.

I can give you exact error reports, info from messages etc upon request.

Tinkster · 08-13-2002, 05:52 PM

Quote:

Originally posted by Gardener
When the second disk is found I get error reports indicating that the disk is corrupted. But since this happened at the same time as the upgrade and the first disk is recognized with the wrong size I'm guessing (or hoping...) that there is some configuration error with the scsi card.

Gidday,

usually that kind of error comes up when there's problems with either
the termination, term power or SCSI parity checks.

This is a very nice place to look when you have
SCSI-hardware and don't know what's stopping it
from working properly...

http://www.hardwarecentral.com/hardw...utorials/36/1/

As for your new & old Controller ... was the built in one a SymBios
or Adaptec?
Adaptec and Symbios use different low-level formats for SCSI HD's,
and if you switched from Sym to Adaptec chances are that you'll
have to low-level the drives... but first of all make sure termination
is set-up right (e.g. that you didn't change the order of your HD's
when rebuilding the system without swapping the termination)

HIH...

Cheers,
Tink

Gardener · 08-13-2002, 08:23 PM

Quote:

Originally posted by Tinkster

usually that kind of error comes up when there's problems with either the termination, term power or SCSI parity checks.

I don't have direct access to the server myself at the moment, but the guy who has been checking on it says the termination should be correct. The disk is first on the chain and a cd-rom is after it, which is terminated. It doesn't say on the disk (a Quantum Lightning I think it is) how to terminate it, that's why it is set up like this. The CD-ROM is not needed actually, in fact it is possible that it is broken, it hasn't worked for quite some time but I haven't been bothered to check on it since it is only 4x.

Quote:

This is a very nice place to look when you have
SCSI-hardware and don't know what's stopping it
from working properly...

http://www.hardwarecentral.com/hardw...utorials/36/1/

Thanks, I'll read that tutorial, I surely need it. =]

Quote:

As for your new & old Controller ... was the built in one a SymBios
or Adaptec?
Adaptec and Symbios use different low-level formats for SCSI HD's, and if you switched from Sym to Adaptec chances are that you'll have to low-level the drives... but first of all make sure termination is set-up right (e.g. that you didn't change the order of your HD's when rebuilding the system without swapping the termination)

Luckily, the old one was also Adaptec, so this doesn't apply to me.

I have been able to start the server with the disk working and reporting no errors and all partition sizes being OK. I don't know if it has something to do with the fact that it's been started with kernel 2.4.2-2, but I don't think so since the scsi card doesn't always find the disk but it did this time.

Since I got the disk mounted I've backed up all data so the the loss won't be so big anymore. But it would be nice if I didn't have to buy a new drive.

I tried to unmount the /home partition (the only one on the problem disk) but it didn't work, and afterwards the size of the last partition on the first disk is screwed up again. The error I got was this:

Code:

bash-2.05# umount /home
umount: /home: device is busy
bash-2.05# umount /home -f
umount2: Device or resource busy
umount: /dev/sdb1: not mounted
umount: /home: Illegal seek

I wanted to mount as read-only and try to run reiserfsck on it. But I guess I could erase it and make an ext3 fs on it instead. But I will still have to find out what the actual problem is, so I'll keep trying a bit more.

Thanks for taking your time to help me, it's much appreciated.