EXT3 Error causing systme to hang ...

[K.C. Chan]

Attached is the Error message from dmsg just before the system hang. I am running Redhat 7.2 kernel 2.4.18. What's I don't get is that we're using IDE drive, but why is it that the error msg is complaining about scsi drive on (8,81) which we don't have. Does anyone have any idea? Thanks.

[Vitaly Karasik_1 · Accepted solution]

I see two reasons for this:

1) physical problem with your disk - check your HD for badblocks

2) bug in ext3 in RH7.2 "stock" kernel - please update you kernel to the latest according to RH7.2 errdata


Vitaly.

[Balaji N]

hi,
are u using any scsi drives. see if there is any info in /proc/devices.

if yes, disable them. also check if ur ide disk is proper. do a dd and see if it is completely readable.

and post back the results once u have fixed them.

thanks
-balaji

[K.C. Chan]

Balaji,
attached is the result of cat /proc/device. I do not have scsi disk or any other scsi devices. As far as the dd command, do you mean "dd if="? Thanks.

[Bruce Copeland]

There are a number of devices such as CD-RW drives and various USB drives that are emulated as scsi devices under Linux through the ide-scsi architecture. Are you sure you don't have any such devices?

Bruce

[K.C. Chan]

Bruce Yes I do have CD and USB devices, but according to err it's on and sdf device which I don't have. Any idea why? Thanks.

[K.C. Chan]

Vitaly,
I did scan the disk for badblocks, it turns out there are 4 badblocks on the disk. I will dump the badblocks to a file and then mark it as bad via e2fsck. Could this be why my box keeps crashing when I started up my app which resides on the filesystem with the badblocks? Thanks.

[Vitaly Karasik_1]

>Vitaly,
>I did scan the disk for >badblocks, it turns out >there are 4 badblocks on the >disk. I will dump the >badblocks to a file and then >mark it as bad via e2fsck. >Could this be why my box >keeps crashing when I >started up my app which >resides on the filesystem >with the badblocks? Thanks.

IMHO, it may be a reason for your craches.
I suggest you to reformat this HD and run mkfs with badblock checking.

[K.C. Chan]

Vitaly, that's nice to know. But unfortunately I do not have that luxury of re-formating my drives. You see the server is at a remote location.

All,
Is there a way to run e2fsck with -c option to check for bad block when I reboot the server. I need to do this before the filesystem is mounted during the boot process. Thanks.

[Vitaly Karasik_1]

are you speaking about / filesystem or other (like /data, /home,...)?

[K.C. Chan]

Vitaly,
I've figured how to handle the filesystem check on / or any other filesystems. Thanks.