[lustre-discuss] confirming behavior we're seeing

Ben Evans bevans at cray.com
Tue Dec 8 10:28:32 PST 2015

Yes, that¹s to be expected.  Lustre attempts to keep all OSTs equally
full, so when you have a severe imbalance across OSTs, new files will
preferentially be allocated on the empty OSTs.

One method for dealing with this is to have one node dedicated to
rebalancing the data.  It might take a long time, but it will get done

-Ben Evans

On 12/8/15, 12:41 PM, "lustre-discuss on behalf of John White"
<lustre-discuss-bounces at lists.lustre.org on behalf of jwhite at lbl.gov>

>A month or two ago we expanded a lustre instance (added an OSS+OTSs to a
>fairly full file system).  Since then, we¹ve seen IO patterns that
>heavily favor the new OSS/OSTs.  In the default allocation strategy, is
>this to be expected in a file system with heavily disparate free space
>among OSTs?
>We don¹t really have the luxury of rebalancing things (assuming the
>method for doing such is still ³re-write/copy files on old OSTs and let
>the allocation strategy handle it"), unfortunately, so we¹re just looking
>to confirm the behavior.
>lustre-discuss mailing list
>lustre-discuss at lists.lustre.org

More information about the lustre-discuss mailing list