[Gluster-users] Gluster volumes with different size

[Emilio Moreno]

Hi Bryan!

Thanks for response!

About Infiniband setup, you're with better configuration, with 3.3.0
version of Gluster, we can't use RDMA with Infiniband and we mount the
volumes with IPoIB.

We're waiting 3.4 version of Gluster...It will possible? I don't know...


Emilio




On 26/04/13 23:01, Bryan Whitehead wrote:
> Rebalance is a safe thing to do - but could cause a lot of heavy IO.
> 
> I'm curious on your Infiniband setup, I'm using Infniband as well but
> using IPoIB. When I've tried native RDMA I've had extremely poor
> performance. Can you show me your gluster volume and how you are
> mounting over fuse? Want to see if I'm missing something major.
> 
> On Fri, Apr 26, 2013 at 9:38 AM, Emilio Moreno <emilio.moreno at adam.es> wrote:
>> Hi!
>>
>> We have "problems" with size of volumes after replace three bricks, our
>> farm of gluster it's configured with Stripped-Distributed-Replicate
>> scheema with 8 bricks of 2TB.
>>
>> Brick 1         /gfs    Used 70%
>> Brick 2         /gfs    Used 47%
>> Brick 3         /gfs    Used 60%
>> Brick 4         /gfs    Used 52%
>> Brick 5         /gfs    Used 25%
>> Brick 6         /gfs    Used 16%
>> Brick 7         /gfs    Used 15%
>> Brick 8         /gfs    Used 52%
>>
>> The new bricks after replace process without problems are, 5, 6 and 7.
>>
>> Our question is...it's normal? Have we to do something? Rebalance volume
>> it's secure and necessary?
>>
>> Version of gluster 3.3.0 with Infiniband Network.
>>
>>
>> Thanks!
>>
>> _______________________________________________
>> Gluster-users mailing list
>> Gluster-users at gluster.org
>> http://supercolony.gluster.org/mailman/listinfo/gluster-users
> .
>