Lustre discuss - Mar 2009 - finding files belonging to dead/offline ost

Hello,

We recently had an OST become badly corrupted (850G or so in
lost+found).  I deactivated this OST on the MDT server & the clients.
The OST is not mounted on the OSS (not possible).

I would like to determine which files were on that OST.

On one of the clients, I tried the following:

"lfs find -O fs01-OST00005-osc-fff810800335400 /mnt/lustre/fs01" does
not return anything.
"lfs getstripe -O fs01-OST0005-osc-ffff810800335400" returns
"error:
setup_obd_uuid: unknown obduuid: fs01-OST0005-osc-ffff810800335400"

Lctl dl returns ost5 as: 

8 UP osc fs01-OST0005-osc-ffff810800335400
6df4e008-e988-d8d0-d50e-cad4d448183c 5

Any suggestions?

Did I miss something obvious?  This data has to be somewhere in the mdt,
correct?

Robert Hendelman Jr
Magnetar Capital LLC
Rob.Hendelman at magnetar.com
1-847-905-4557



The information contained in this message and its attachments 
is intended only for the private and confidential use of the 
intended recipient(s).  If you are not the intended recipient 
(or have received this e-mail in error) please notify the 
sender immediately and destroy this e-mail. Any unauthorized 
copying, disclosure or distribution of the material in this e-
mail is strictly prohibited.

To answer my own question, the following seems to  be working (from the
client):

Lfs getstripe -O fs01-OST0005_UUID -r /mnt/lustre/fs01

Robert Hendelman Jr
Magnetar Capital LLC
Rob.Hendelman at magnetar.com
1-847-905-4557



The information contained in this message and its attachments 
is intended only for the private and confidential use of the 
intended recipient(s).  If you are not the intended recipient 
(or have received this e-mail in error) please notify the 
sender immediately and destroy this e-mail. Any unauthorized 
copying, disclosure or distribution of the material in this e-
mail is strictly prohibited.

On Mar 23, 2009  13:19 -0500, Hendelman, Rob wrote:
> We recently had an OST become badly corrupted (850G or so in
> lost+found).  I deactivated this OST on the MDT server & the clients.
> The OST is not mounted on the OSS (not possible).
You can likely recover many of these files by running the
ll_recover_lost_and_found program that comes with newer Lustre
installations.
> On one of the clients, I tried the following:
> 
> "lfs find -O fs01-OST00005-osc-fff810800335400 /mnt/lustre/fs01"
does
> not return anything.
Note that you have one too many 0s in your UUID above (only OST0005).
> "lfs getstripe -O fs01-OST0005-osc-ffff810800335400" returns
"error:
> setup_obd_uuid: unknown obduuid: fs01-OST0005-osc-ffff810800335400"
Use the UUIDs as listed at the start of "lfs getstripe" or "lfs
find".

This is a pet peeve of mine that the UUID specification should be more
forgiving.  I have a patch I''ve been meaning to polish up for this so
that just specifying "fs01-OST0005" would be enough, and filing a bug
on the user tools would get this underway.

Cheers, Andreas
--
Andreas Dilger
Sr. Staff Engineer, Lustre Group
Sun Microsystems of Canada, Inc.