Gluster users - Apr 2018 - Expand distributed replicated volume with new set of smaller bricks

We currently have a 3 node gluster setup each has a 100TB brick (total
300TB, usable 100TB due to replica factor 3)
We would like to expand the existing volume by adding another 3 nodes, but
each will only have a 50TB brick. I think this is possible, but will it
affect gluster performance and if so, by how much. Assuming we run a
rebalance with force option, will this distribute the existing data
proportionally? I.e., if the current 100TB volume has 60 TB, will it
distribute 20TB to the new set of servers?

Thanks
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.gluster.org/pipermail/gluster-users/attachments/20180403/a45072a5/attachment.html>

Hi,

Yes this is possible. Make sure you have cluster.weighted-rebalance enabled
for the volume and run rebalance with the start force option.
Which version of gluster are you running (we fixed a bug around this a
while ago)?

Regards,
Nithya

On 4 April 2018 at 11:36, Anh Vo <vtqanh at gmail.com> wrote:
> We currently have a 3 node gluster setup each has a 100TB brick (total
> 300TB, usable 100TB due to replica factor 3)
> We would like to expand the existing volume by adding another 3 nodes, but
> each will only have a 50TB brick. I think this is possible, but will it
> affect gluster performance and if so, by how much. Assuming we run a
> rebalance with force option, will this distribute the existing data
> proportionally? I.e., if the current 100TB volume has 60 TB, will it
> distribute 20TB to the new set of servers?
>
> Thanks
>
> _______________________________________________
> Gluster-users mailing list
> Gluster-users at gluster.org
> http://lists.gluster.org/mailman/listinfo/gluster-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.gluster.org/pipermail/gluster-users/attachments/20180404/d769ec3d/attachment.html>

We are using 3.8.15 I believe (the one comes with ubuntu 16.04). Do you
know when this was fixed?

Thanks

On Tue, Apr 3, 2018 at 11:24 PM Nithya Balachandran <nbalacha at
redhat.com>
wrote:
> Hi,
>
> Yes this is possible. Make sure you have cluster.weighted-rebalance
> enabled for the volume and run rebalance with the start force option.
> Which version of gluster are you running (we fixed a bug around this a
> while ago)?
>
> Regards,
> Nithya
>
> On 4 April 2018 at 11:36, Anh Vo <vtqanh at gmail.com> wrote:
>
>> We currently have a 3 node gluster setup each has a 100TB brick (total
>> 300TB, usable 100TB due to replica factor 3)
>> We would like to expand the existing volume by adding another 3 nodes,
>> but each will only have a 50TB brick. I think this is possible, but
will it
>> affect gluster performance and if so, by how much. Assuming we run a
>> rebalance with force option, will this distribute the existing data
>> proportionally? I.e., if the current 100TB volume has 60 TB, will it
>> distribute 20TB to the new set of servers?
>>
>> Thanks
>>
>> _______________________________________________
>> Gluster-users mailing list
>> Gluster-users at gluster.org
>> http://lists.gluster.org/mailman/listinfo/gluster-users
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.gluster.org/pipermail/gluster-users/attachments/20180404/a8b60390/attachment.html>