rp5450 || Disk Drive Issue || vxfs mount: /dev/vg01/oracle is corrupted. needs checking

[AL-MARI]

Hello Experts,

After the power failure. My oracle installation mount point (/dev/vg01/oracle) is not getting up.

Error message:

vxfs mount: /dev/vg01/oracle is corrupted. needs checking
vxfs fsck: file system had I/O error(s) on meta-data.
vxfs fsck: fsck read failure bno = 31320 , off =1, len = 8192

file system check failure, aborting ...

************************************************************************************************************************************

Ran a DD command on the drive :  c1t2d0

# dd if=/dev/rdsk/c1t2d0 of=/dev/null bs=1024k

34732+1 records in
34732+1 records out

Means there is no harddisk failure issue. Kindly advise why my mount point  (/dev/vg01/oracle) is not getting up.

what are the meaning of this error message "vxfs mount: /dev/vg01/oracle is corrupted. needs checking"

Thanks in advance.

Kind Regards,

Masaod

[AL-MARI]

Is this issue related to hardware or software?

[Bill Hassell]

Error message:
vxfs mount: /dev/vg01/oracle is corrupted. needs checking
vxfs fsck: file system had I/O error(s) on meta-data.
vxfs fsck: fsck read failure bno = 31320 , off =1, len = 8192
file system check failure, aborting ...

Not an unusual situation after a power failure with JBOD (ordinary) disks. fsck was run as the system booted up and was unable to read the disk at the location given. fsck cannot fix any I/O errors.  This is a hardware failure. While there are many problems inside the disk that could cause this condition, none of them are fixable from HP-UX. Your dd test is a good test but the error may be intermittent. You can see additional details about the error in /var/adm/syslog/syslog.log. If fsck cannot fix the disk, it cannot be used and your database will be offline because the oracle mountpoint is empty (not mounted).

The disk must be replaced. It appears that disk is not mirrored so the data must be restored on the replacement disk from your backups.

[AL-MARI]

Hello Bill Hassell, 

Thank you very much for your comment. As per HP-UX support engineer. There is no issue with hardware (Hard Drive). It is software issue.

As per him, after dd command output. Hard Drive is OK. But i am not convienced. 

# dd if=/dev/rdsk/c1t2d0 of=/dev/null bs=1024k

34732+1 records in
34732+1 records out

Regards,

Sohail

[Bill Hassell]

Did the HP engineer provide any evidence that this is a software error?
Or a patch or driver update to fix the issue?
Without a software fix, I assume that the disk still won't allow mounting.

As I mentioned, the details needed to identify the problem are probably in syslog.log.
Did you look there? Look for a disk status error at the time you run fsck. 

[Patrick Wallek]

What happens if you do:

# fsck -y -o full  /dev/vg01/oracle

I have a feeling it will fail.

The error message about failure to read a certain block number indicates that there was a read error on the disk / lun.  As mentioned before, there is no software fix for that.