topic Re: Disk failure in Operating System - HP-UX

Disk failure

enrico.nic — Fri, 29 Jan 2016 20:25:49 GMT

Hi,

I have an "old" HP rp3410 with HP-UX 11.31.

In an external enclosure, I have 2 300 Gb SCSI disks, attached to an Ultra320 controller.

Today one of these disks stopped functioning; I tried to fsck the disk with the following error:

# fsck -F vxfs -o full,nolog /dev/vg03/lvol1
UX:vxfs fsck: WARNING: V-3-20836: file system had I/O error(s) on meta-data.
pass0 - checking structural files
pass1 - checking inode sanity and blocks
UX:vxfs fsck: ERROR: V-3-26117: bc_read failure devid = 0, bno = 8373899, off = 0, len = 8192

Is there anything else I can try to recover the contents of the file system in it ?

Thank you

Enrico

Re: Disk failure

Patrick Wallek — Fri, 29 Jan 2016 21:18:33 GMT

What does the disk show in the output from 'ioscan -fnC disk'? If it show's NO_HW, then you're out of luck.

If you did not have the data mirrored then you'll have to restore the data after you replace the disk.

Re: Disk failure

enrico.nic — Fri, 29 Jan 2016 23:23:42 GMT

The status is CLAIMED

# ioscan -fnC disk /dev/dsk/c4t8d0
Class I H/W Path Driver S/W State H/W Type Description
=====================================================================
disk 4 0/4/1/0.8.0 sdisk CLAIMED DEVICE HP 300 GMAW3300NC
/dev/dsk/c4t8d0 /dev/rdsk/c4t8d0

Re: Disk failure

Bill Hassell — Sat, 30 Jan 2016 00:20:23 GMT

The status CLAIMED means that the driver could talk to the electronics inside the disk. But the errors indicate that the disk has had a crash and the oxide is damaged. Therefore the data is unreadable and no HP-UX tools can restore this area. Disks are the highest failure item for any computer. Mirroring is simply mandatory for any data that is important.

If money is not o problem, there are several disk recovery companies that may be able to recover the data on the disk. It may cost more than your entire computer though.

Re: Disk failure

enrico.nic — Sat, 30 Jan 2016 10:50:45 GMT

About the error reported from fsck, I've found these informations on the veritas.com website (https://www.veritas.com/support/en_US/article.000013525):

Cause

I/O read failure at block offset 32206696 (into the file system) was detected by fsck

Solution

At this point during the fsck, no disk I/O errors are tolerated. Retry the file system check, as the I/O error might be transient. Check the console logs for any I/O errors relating to the filesystem's devices. If the problem is a disk failure, replace the disk.

I will try again to repair the disk, as they suggest, but the possibilities of a success are really limited. In the meantime, I am encountering some problems with frecover in restoring the backup on another disk, but if the problem persists this will be the subject of another post.

Thank you

Enrico

Re: Disk failure

enrico.nic — Tue, 02 Feb 2016 10:04:25 GMT

A brief follow up.

I've tried to put the damaged disk in a -20 C freezer for about 1 and a half hour.

Afterwards, I've reinserted the damaged disk in the server. At the first boot, fsck repaired successfully the disk, and it was mounted without a further manual intervention.

After the data recovery operations ("rsync" was used to copy the last updates performed on the file system to a new disk) I tried to do a last backup of the disk. Many read faults but it reached the end.

Problem solved, with a good amount of luck. Next time I will set up some type of redundancy of the data.

Thank you for your help

Enrico