topic Re: SCSI errors in Operating System - Linux

SCSI errors

admin1979 — Mon, 30 Nov 2009 07:44:23 GMT

Hello,

We have SLES 10 System with SCSI disks.
We have noticed attached errors occurred for lets say 5 minutes. Wonder if the disk is going to fail?
If yes what are the best possible ways to minimise the imapct? I have already taken the backup of the partitions like,
/
/boot
/opt
/var
/var/home
/srv/fpt etc..

Please suggest.

Thanx,
Admin

Re: SCSI errors

Ivan Ferreira — Mon, 30 Nov 2009 12:52:37 GMT

SCSI disks are able to self handle bad blocks. but yes, there is the possibility that the disk is about to fail.

Probably would be better if you replace the disk before disaster strikes.

Take a good backup and restore to the new disk. From your output, I cannot say what is the file system associated with your disk.

The following information is also needed:

fdisk -l
cat /etc/fstab
vgdisplay -v

Re: SCSI errors

admin1979 — Mon, 30 Nov 2009 13:05:00 GMT

Here is the required data.
sda is the boot disk :(

Thanx,
admin

Re: SCSI errors

Vitaly Karasik_1 — Tue, 01 Dec 2009 08:23:24 GMT

I'll suggest to run something like

dd if=/dev/sda of=/dev/null

for read whole disk and see how many errors you'll see.

Re: SCSI errors

admin1979 — Tue, 01 Dec 2009 08:26:03 GMT

Ok that can be one check. But now that we already know that there is a problem in this disk. More important thing is to make sure that there is a minimal damage if the disk is going to go down.

We have taken the backup of the root partitions. But still we also would
like to prepare a disk which in case of the failure can be inserted and bring up the system as it was before.
Is there any such efficient way to do it?

We have a new disk of similar size.
So any suggestions?

Thanx,
admin

Re: SCSI errors

Vitaly Karasik_1 — Tue, 01 Dec 2009 11:09:50 GMT

What is sda disk - it it physical disk or hardware mirror of 2 physical disks?

In second case you should break mirror, replace disk and re-build RAID - all using REIAd manager.

In the first case, if you have the same size or bigger disk and free SCSI slot in your server, I'll just insert new disk and clone old -> new disk using "dd".

Re: SCSI errors

admin1979 — Tue, 01 Dec 2009 11:13:56 GMT

Hi,

Its a single disk without RAID. So lets suppose I insert a new disk in the slot and run

dd if=/dev/sda of=/dev/sdd bs=1M

Will work?

Will this disk boot as the original as well?

Thanks,
admin

Re: SCSI errors

Vitaly Karasik_1 — Tue, 01 Dec 2009 12:51:48 GMT

Yes, this should work.

Re: SCSI errors

Steven E. Protter — Tue, 01 Dec 2009 20:26:03 GMT

Shalom,

You could also use software raid tools mdadm

http://tldp.org/HOWTO/Software-RAID-HOWTO-5.html#ss5.6

SEP

Re: SCSI errors

admin1979 — Tue, 01 Dec 2009 21:56:17 GMT

mdadm look great. But since never tried before , trying out on a production system directly , is bit of hesitation.
Cold you please clarify the steps involved?

Thanx,
admin

Re: SCSI errors

admin1979 — Thu, 02 Sep 2010 06:27:34 GMT

Not much help.