[lustre-discuss] GID only mapping in 2.8.60?

Dilger, Andreas andreas.dilger at intel.com
Fri Nov 4 16:48:04 PDT 2016

On Nov 4, 2016, at 16:53, Stephane Thiell <sthiell at stanford.edu> wrote:
> Hi all,
> I am currently evaluating Lustre 2.8.60 with UID/GID mapping for a Lustre storage system that will be connected to multiple clusters. These clusters have matching UIDs but same groups have different GIDs. Actually GID mapping is working fine but I can’t find an easy way to avoid explicit UID mapping (all undefined UIDs are squashed). I just want GID mapping so I am looking for a way to bypass UID mapping, a bit like the ‘trusted’ idmap attribute in order to use canonical UIDs but still map GIDs according to gid idmaps.
> Also as UID ranges are not supported, I created a bunch of single UID idmaps with identity mapping (uid:uid) for all our reserved UID ranges at Stanford but those are very large (999 to 999999) and I am limited by nodemap_add_idmap rate (~170 idmap / sec, so it takes about 1h40).

I agree that allowing ranges of UID/GID to be mapped would be very useful in different sites, instead of creating so many separate IDs.

In your case, it might be simplest to add a flag to the nodemap like "identity_uid" that avoids the need for UID maps, and then only install GID maps and a select number of UID maps as needed?  It might also be worthwhile to add an "identity_gid" flag at the same time, even if you aren't using it at this time.

Cheers, Andreas

More information about the lustre-discuss mailing list