Gluster users - Jul 2016 - Expand distributed replicated volume

2016-07-12 12:34 GMT+02:00 Pranith Kumar Karampuri <pkarampu at
redhat.com>:
> If you add 6 new disks to the cluster by bringing in a new node with
replica
> count 3 you are essentially are adding 2 replica sets. Since you can't
have
> replica sets with bricks from same node, we need to get empty bricks from
> other nodes which require replace-brick operation.
If I replace existing bricks, i'm not adding free space but only move
data around.

On Tue, Jul 12, 2016 at 4:06 PM, Gandalf Corvotempesta <
gandalf.corvotempesta at gmail.com> wrote:
> 2016-07-12 12:34 GMT+02:00 Pranith Kumar Karampuri <pkarampu at
redhat.com>:
> > If you add 6 new disks to the cluster by bringing in a new node with
> replica
> > count 3 you are essentially are adding 2 replica sets. Since you
can't
> have
> > replica sets with bricks from same node, we need to get empty bricks
from
> > other nodes which require replace-brick operation.
>
> If I replace existing bricks, i'm not adding free space but only move
> data around.
>
True. But at the end of 4 replace-bricks, you have 4 bricks from earlier
configuration which are empty now, which can be re-used. So essentially you
have 6 empty bricks which can be put back into use, but these can be used
to form replica sets as they are not from same machine. So you add-brick 2
replica sets with these 6 bricks and do rebalance. Makes sense?


-- 
Pranith
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://www.gluster.org/pipermail/gluster-users/attachments/20160712/8c0297a5/attachment.html>