Thank you Andreas for answering. However for the time being upgrading
our lustre installation is not an option. The emergency is to be able to
retrieve our data.
So far I tried to issue this command:
# e2fsck --mdsdb mds_database_file /dev/sdb
But then realized it was pointless as I can't run lfsck since I can't
mount the volume with -t lustre. The system still crashes with the error
Guillaume posted and then reboot when doing mount.lustre /dev/sdb /mdt.
In some other thread someone encountered an almost identical issue but
with lfsck:
Message from syslogd@localhost at Nov 28 10:52:49 ...
kernel:LustreError: 8355:0:(osd_handler.c:1017:osd_trans_start())
ASSERTION( get_current()->journal_info == ((void *)0) ) failed:
Message from syslogd@localhost at Nov 28 10:52:49 ...
kernel:LustreError: 8355:0:(osd_handler.c:1017:osd_trans_start()) LBUG
Sep 13 15:10:28 n00a kernel: [ 8414.619833] Pid: 11696, comm: lfsck
but then managed to find a workaround by mounting the volume as ldiskfs
and then delete some namespace file :
# mount -t ldiskfs -o rw /dev/mdt /mnt/mdt.ldiskfs
# rm -f /mnt/mdt.ldiskfs/lfsck_namespace
# umount /mnt/mdt.ldiskfs
Could this be the solution in our case ?
Cheers,
J.
Le 29/11/2018 15:40, Guillaume Postic a écrit :
--- message retransmis ---
--
Julien REY
Plate-forme RPBS
Molécules Thérapeutiques In Silico (MTi)
Université Paris Diderot - Paris VII
tel : 01 57 27 83 95
_______________________________________________
lustre-discuss mailing list
lustre-discuss@lists.lustre.org
http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org