[lustre-discuss] Target index choice

Andreas Dilger adilger at whamcloud.com
Fri Apr 8 01:33:51 PDT 2022

On Apr 8, 2022, at 01:50, Hans Henrik Happe via lustre-discuss <lustre-discuss at lists.lustre.org<mailto:lustre-discuss at lists.lustre.org>> wrote:


Is or will there be a downside to choosing discontinuous index numbers? I.e. encode OSS number YY and target number XX like 0xYYXX.

I guess it could hurt if layouts are packed to save space.

Currently there is no layout packing, though this has been discussed in the past.  However, something like gzip of a large layout would also be insensitive to the OST index numbering.

In general, there are no significant problems with discontiguous OST index numbering, and there are sites in production with gaps in the ranges (e.g. OST0000-00nn are HDD and OST0100-01nn are NVMe).  However, there are still occasional bugs found because of the numbering gaps, so I wouldn't recommend using a huge sparse array like OST0000 - OST2244 with 22 OSSes and 44 OSTs as it may unnecessarily put you on the bleeding edge of finding new bugs (or at least using more memory or less efficient processing than necessary in some places).

Cheers, Andreas
Andreas Dilger
Lustre Principal Architect

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-discuss-lustre.org/attachments/20220408/078a3d8b/attachment.html>

More information about the lustre-discuss mailing list