We were able to resolve our issue by freeing up inodes using the process
outlined in https://wiki.lustre.org/ZFS_MDT_ENOSPC_Recovery. This resolved
the "rm: cannot remove XXX: No space left on device" problem. Then we grew
the MDT zpool by adding more disks using a standard `zpool add` command. We
were a little concerned about that operation in particular since it was not
clear if our older Lustre version supports dynamically increasing the size
of the MDT. We found a few vague mentions of folks doing so, but no
specific commands or versions were listed. It worked fine and we now have
twice as many inodes as we had before.

Upgrading the Lustre version on our servers now has a higher priority so we
can get to a version that lets us do file level backups of the MDT.

-- 
Regards,
-liam

-There are uncountably more irrational fears than rational ones. -P. Dolan
Liam Forbes           [email protected]                       ph:
907.450.8618
UAF GI Research Computing Systems Manager
https://calendly.com/ualoforbes/30min
_______________________________________________
lustre-discuss mailing list
[email protected]
http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org

Reply via email to