fsck error

[tareq_2]

hi
i have two VG vg01 and vg02 on vg02 i have two LV lv12 mounted on /u02 and lv13 mounted on /u01 for oracle i tried to mount /u01 but there error say you need to run fsck but i have face this error:
# fsck -F vxfs -o full -n /dev/vg01/rLvol12
pass0 - checking structural files
pass1 - checking inode sanity and blocks
pass2 - checking directory linkage
vxfs fsck: fsck read failure bno = 194512, off = 0, len = 8192
how i can fix this error?
thnx for all

[Duncan Edmonstone]

did you try it without the -n option?

fsck -F vxfs -o full n /dev/vg01/rLvol12

HTH

Duncan

[tareq_2]

yes i tried without n but the same

[Duncan Edmonstone]

can you read the whole LV without error?

so can you do this:

dd if=/dev/vg01/rLvol12 of=/dev/null bs=8k

and see it complete without error?

what does lvdisplay show for the lvol...

possibly you have a failing disk

HTH

Duncan

[Ganesan R]

Hi,

It looks you have some media error on the disk/LV where the filesystem is residing.

Just check if you have stale extends on the lv.
#lvdisplay -v /dev/vg01/lvol12 |grep stale

Sometime lvdisplay won't display media errors as stale. So try with dd to read the entire logical volume or even entire disk.

dd if=/dev/vg01/rLvol12 of=/dev/null bs=1024

dd if=/dev/rdsk/cxtxdx of=/dev/null bs=1024k

[tareq_2]

hi there
i have tried the
dd if=/dev/vg01/rLvol12 of=/dev/null bs=1024 command but the result is
IO error. i have faced the same problem but in another server at that time i have delete the LV and make it again and its work.
but in this case i need the data in the LV and i can't delete it

[Duncan Edmonstone]

looks a lot like a bad disk to me... anything in syslog.log indicating a problem?

you *do* have a backup don't you?

Duncan

[Patrick Wallek]

Basically your problem is that you can't read a block on the disk.

What kind of disk makes up this VG? Is it a LUN on a SAN or is it a regular disk drive?


If this is a regular disk drive, then the drive needs to be replaced and if there is no mirroring, then the data will have to be restored.

If this disk is part of a SAN, then you probably still need to replace a disk, but it will be more difficult determining which disk. You will need to use SAN utilities to determine this.

You may still need to restore data though. Hopefully you have a relatively recent backup of this data.

[Steven E. Protter]

Shalom,

Based on a careful read of the thread, you have a bad disk block. A physical sector of the disk can not be read.

There appears to be an oracle table sitting the bad spot.

Therefore your alternative is to fix the disk and restore the database.

http://docs.hp.com/en/T1859-90048/ch04s07.html

What course of action you take depends on whether or not your disk is under local control or SAN.

SEP

[tareq_2]

thanks for all