[lustre-discuss] an OST is dead

[Noskov, Dr. Sergey]

Hi everyone,

we have a problem and I would be very grateful for help or advice.

We are using ZFS for metadata and object storage target in our Lustre. One of the pools is a draid3:12d:42c:2s-0 is became no longer importable after one of the disks failed.

on startup zpool import:

pool: l1fs-OST010b
     id: 6548008278833985886
  state: FAULTED
status: One or more devices contains corrupted data.
action: The pool cannot be imported due to damaged devices or data.
   see: https://openzfs.github.io/openzfs-docs/msg/ZFS-8000-5E
   ...

the commands

zpool import -f l1fs-OST010b
zpool import -F l1fs-OST010b
zpool import -F -X l1fs-OST010b

yield:
cannot import 'l1fs-OST010b': I/O error
                Destroy and re-create the pool from
                a backup source.

zdb -e ends very quickly with the I/O error
Some tool said there is missing metadata in the zpool.

Does anyone have any productive advice for us other than what is on the website
https://openzfs.github.io/openzfs-docs/msg/ZFS-8000-5E
?

Now the question: how do we proceed if we have to reformat the OST?
At the moment we see the plan as:
Disable the broken OST (or is it possible to use the same name for OST?)
create a new pool and integrate it into the Lustre

What do we do with the files on the broken OST, can we just delete them now? Can the difference in metadata and object targets be repaired with Lustre-FSCK or somehow?

Thanks in advance

With best regards

Sergey Noskov


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-discuss-lustre.org/attachments/20250512/329bf628/attachment.htm>