Recovering a VxFs file system

[Diego Balgera]

Hi,
I have a VxFs file system that looks to be seriously corrupted ans I'm trying to "fsck" it.
Running fsck I get the error: "full file system check required, exiting ...". I tried with the "-o full" (and several others) option but without any result, the filesystem is not checked. I tried using the raw and block special file as well.

Can you help me in discovering how to proceed to recover this file system?
By the way, it is the /dev/vg00/lvol8 hosting the /var mount.

#fsck -F vxfs -o full /dev/vg00/lvol8
log replay in progress
pass0 - checking structural files
pass1 - checking inode sanity and blocks
fileset 999 primary inode 2 root inode is not a directory clear? (ynq)y
root inode for fileset 999 is invalid
full file system check required, exiting ...

[Pete Randall · Accepted solution]

Diego,

Try adding the -y option to your fsck command:

#fsck -F vxfs -y -o full /dev/vg00/lvol8


Pete

[Ivan Ferreira]

Is the filesytem umounted? I think yes, but I must ask.

Considering that is the /var filesystem, it will be really easy restore it from a backup. Re-create the filesystem and restore your last backup.

[Diego Balgera]

Hi,

thank you!

Pete, I tried already. With the "-y" option the fsck exists with the same notification (full file system check required, exiting ...) as it happens when I type the "y" manually (like in the enclosed example). I attached the result below.

Ivan, yep, the file system is unmounted.

It looks really strange. If we won't succeed in recovering we are going to restore from a backup, but we would like to leave this the very last option.

Thank you again!
Diego.



#fsck -F vxfs -y -o full /dev/vg00/lvol8
log replay in progress
pass0 - checking structural files
pass1 - checking inode sanity and blocks
fileset 999 primary inode 2 root inode is not a directory clear? (ynq)y
root inode for fileset 999 is invalid
full file system check required, exiting ...

[Robert-Jan Goossens]

Diego,

Could you post your HPUX version please.

regards,
Robert-Jan.

ps 10.20 link with almost the same error.

http://www4.itrc.hp.com/service/cki/docDisplay.do?docLocale=en_US&docId=200000062912401

[Diego Balgera]

Hi Robert,

yes, you are right. This problem under 10.20 looks very similar to my one.
My Hp-Ux version is B.11.00, installed on a A180C 9000 series A server.

Again, thank you!
Diego.

[V. Nyga]

Hi Diego,

you can try to mount this volume with the option -F for force - so you can mount it also system want to fsck first.
(Well, this worked once for me)

HTH
Volkmar

[Diego Balgera]

Hi,

unfortunately I couldn't manage to recover the file system, so I recreated it and restored a backup.
Thank you to everyone for your valuable suggestions (10 points to everyone).

Diego.

[Pawel Orzechowski]

Hi Diego,

I know you have restored filesystem from backup, but just for your information:
fsck should be run over character device so fsck command should look:
#fsck -F vxfs -o full /dev/vg00/rlvol8
^
|
(not lvol8!!!)
This could be a problem.

Best regards

Pawel

[Diego Balgera]

Hi Pawel,
yes, you are right. When I tried to recover the file system, I tried with rlvol8 too but it didn't succeed. I hope do not have damaged the file system attempting to execute the fsck with lvol8 ...
Anyway, file system recreated and backup restored, so the issue is now closed.
Thank you again to everyone!
Diego.