file system corrupt

[Ing Meloni]

I had a problem of corruption in a hfs filesystem. I solved it with a boot and fsck.
After a few days the problem reappeared in the same point. I now suspect it is a hardware problem.
This is the error message from dmesg
/interni: bad dir ino 1648129 at offset 2048: mangled entry
The inode corresponds to a directory without files newer than a month.
This logical volume consists of two physical disks. How do I know which is the faulty one?
I have found no other logs that can help me.

[Krishna Prasad]

Do you see any stale extents when you do a lvdisplay -v /dev/vg##/lvlol# ? If you see stale extents I would replace the drive.

also do a /etc/diskinfo /dev/rdsk/c#t#d# on the disk(s) the filesystem is using.

[Sanjay_6]

Hi,

Do a lvdisplay for the lv having this problem.

lvdisplay -v /dev/vg_name/lv_name

It will give the list of extents on each disk. Look for stale extents.

Hope this helps.

regds

[Sanjay_6]

Hi,

If the solution suggested before doesn't work, do this,

dd if=/dev/rdsk/cxtydz of=/dev/null bs=1024K

Hope this helps.

regds

[pap]

Hi,

Look for the Stale extents in suspected LV by diplaying logical volume info using lvdisplay -v.

Replace the bad dirve after that.

Thanks,
-Piyush.

[K.Vijayaragavan.]

See the command output of following commands,

#vgdisplay -v /dev/vg00
#vgdisplay -v /dev/vg01
#diskinfo /dev/rdsk/c#t#d#
#ioscan -kfnC disk
#cstm

K.Vijay

[Ing Meloni]

did the lvdisplay: no stale sectors.
did some checks with stm: no errors.

I am trying dd. It will take some time.

[S.K. Chan]

Since this is a hfs FS, it is bound to have "mangled inode" problem (sometimes not due to bad disk). What I would do is backed up the directory to a different location, blow away the FS and recreate it. Copy the data back and "wait-and-see". Only do this if you can't find any hardware error.

[Haresh_2]

Hi,
check if you have lost+found directory. If you do not have that then create with mklost+found(1M)
Hope this help.
Good luck.
Haresh