Hi Marian, We have a volume quota translator which can be used to provide user, group and directory level quota also ( e.g., create a volume per user need - home, bin, public etc and mount as such ). So in that sense, user, group and directory level quota can be provided as a solution, even though its not present as a feature per se. It works on 2.x. We are putting some of the fixes on 3.x now so it will work there as well. Till a while back, we did not get much requests for quota support so that side of the FS was less aggressively pursued and hence it was not officially supported. We still, however aggressively fix bugs reported by users. There are now a few volunteers ( including you :-) ) and we are also thinking of a best possible implementation of this. I think this is an extremely good opportunity to do some real "community development" line item for GlusterFS. Let me discuss with our lead on how best to organize this so we can leverage all volunteers. Regards, Tejas Bhise. ----- Original Message ----- From: "Marian Marinov" <mm@xxxxxxxx> To: gluster-devel@xxxxxxxxxx Sent: Monday, March 1, 2010 9:06:39 PM GMT +05:30 Chennai, Kolkata, Mumbai, New Delhi Subject: gluster 'quota' implementation Hello, today we had a conversation with one of the users at #gluster and I mentioned a way of implementing quota without actually implementing quota :) The idea is simple, we can use the quota support of the underlying FS and replicate it to the whole cluster using a predefined dir in the glusterfs cluster. What I'm proposing is: 1. create a dir within glusterfs, for example .quota 2. every machine will export its results from repquota or at least the code behind repquota and save the output to a file which will be named after the hostname of the machine. 3. Wait 1 min after step 2 and read all files from .quota 4. Save or update all the information from step 3 in the memory 5. Use the stored into memory information to enforce quota My question is, are you interested in a translator that does this? If you are, I'm interested in writing it :) -- Best regards, Marian Marinov _______________________________________________ Gluster-devel mailing list Gluster-devel@xxxxxxxxxx http://lists.nongnu.org/mailman/listinfo/gluster-devel
