[lustre-discuss] lfsck + MDS crash

[Thomas Roth]

Hi all,

I was running an "lfsck -C -c -A" and right in the middle of things, the mgs+mdt0 crashed. oi_scrub had finished, namespace I don't know, layout was 
running.

After recovery, this MDS dutifully continues with the layout scan, as do the OSSes.

But we have two more MDS, MDT1 and MDT2, both of which show layout=completed. Also, all three show namespace=completed.

Would it (namespace) show "completed", when reality, it is "aborted" - due to the first MDT disappearing for a while?
And while the OSTs are still working on the layout, the two MDT1+2 cannot really have completed their layout scan?


The outputs of 'lctl get_param ....layout` does not show anything that looks alarming, but obviously I am not capable to understand the full output.


Regards,
Thomas
-- 
--------------------------------------------------------------------
Thomas Roth
Department: IT

GSI Helmholtzzentrum für Schwerionenforschung GmbH
Planckstraße 1, 64291 Darmstadt, Germany, www.gsi.de

Commercial Register / Handelsregister: Amtsgericht Darmstadt, HRB 1528
Managing Directors / Geschäftsführung:
Prof. Dr. Thomas Nilsson, Dr. Katharina Stummeyer, Jörg Blaurock
Chairman of the Supervisory Board / Vorsitzender des GSI-Aufsichtsrats:
State Secretary / Staatssekretär Dr. Volkmar Dietz