[Gluster-users] Problem with free space limits on servers.

[Keith Freedman]

At 08:56 AM 12/22/2008, Amar S. Tumballi wrote:
>Hi Lior,
>
>2008-12-18 19:20:09 W [alu.c:892:alu_scheduler] alu: No node is 
>eligible to schedule
>
>
>However it doesn't enforce the limit, I can still copy files, untill 
>all the servers have their partitions fully filled.

as I understand it, quotas are managed by the underlying filesystem.  so:
depending on your scheduler, if you have 5 servers unified, for 
example, and a quota on each machine of 1GB per user, you may end up 
with one user having 5GB if their data gets distributed evenly across 
the unified bricks.

>Is this a normal behavior ?
>Is there a way to enforce the limit, i.e. get <no space left on 
>device> type of error to the process that tries to write.
>
>
>Unify's schedulers themself  won't implement any Quotas (ie, 
>limits). These are limits, after crossing which, you will get 
>warnings, critical messages in log file, so you are aware that all 
>nodes are full, and this is time for storage upgrade. If extra nodes 
>are not added, it will continue to schedule file creation within 
>available nodes.
>
>You may want to try 'quota' translator in that case (which is in 
>1.4.0rc+ releases), which just implements export directory limits. 
>(And this has to be added just over the posix translator, in server part).
>
>Regards,
>Amar
>
>_______________________________________________
>Gluster-users mailing list
>Gluster-users at gluster.org
>http://zresearch.com/cgi-bin/mailman/listinfo/gluster-users