[lustre-discuss] [EXTERNAL] ran out of MDT inodes

Mohr, Rick mohrrf at ornl.gov
Fri Sep 16 10:01:26 PDT 2022


Liam,

As far as I know, all the data contained in ZFS would be transferred (Lustre or otherwise).  We were able to move an mdt to new storage on a different server with no known issues.  There was also a talk at LUG several years ago given by a site that would use ZFS snapshots as a safeguard during Lustre upgrades so that they could easily roll back if needed.  Perhaps that would be of use to you.  I can't remember which year it was, but it may have been the one hosted at IU.  Anyway, all the old LUG talks/slides should be online so if you dig around you should be able to find it. 

-Rick

On 9/16/22, 12:06 PM, "Liam Forbes" <loforbes at alaska.edu> wrote:

    Morning Rick!

    At first we were planning to do that. However, we weren't sure if that would capture all the internal Lustre data. Specifically the note in 18.3.1 about index backups made us think it wouldn't for version 2.10.X. Also the fact that ldiskfs backups and LVM snapshots are detailed in the manual but ZFS snapshots are not caused us to think they wouldn't work, at least with our older lustre version. Maybe we were taking things too literally. Would a ZFS snapshot contain all the Lustre data?



    On Fri, Sep 16, 2022 at 7:45 AM Mohr, Rick <mohrrf at ornl.gov> wrote:


    Liam,

     If you have another zpool configured somewhere, you could always take a snapshot of your mdt and then used send/received to copy that snapshot to another zpool.  I helped someone do this one time in order to move the mdt to new hardware.

    -Rick




    -- 
    Regards,
    -liam

    -There are uncountably more irrational fears than rational ones. -P. Dolan
    Liam Forbes           loforbes at alaska.edu                       ph: 907.450.8618
    UAF GI Research Computing Systems Manager
    hxxps://calendly.com/ualoforbes/30min



More information about the lustre-discuss mailing list