[Gluster-users] Automation of single server addition to replica

[Joe Julian]

On 11/09/2016 11:22 AM, Gandalf Corvotempesta wrote:
> 2016-11-09 19:32 GMT+01:00 Joe Julian <joe at julianfamily.org>:
>> Yes, and ceph has a metadata server to manage this
> And that's why I really prefere gluster, without any metadata or similiar.
> But metadata server aren't mandatory to archive automatic rebalance.
> Gluster is already able to rebalance and move data around the cluster,
> and already has the tool to add a single server even in a replica 3.
>
> What i'm asking is to automate this feature.
> Gluster could be able to move bricks around without user intervention.
>
> The proper way to add a single server in a replica 3 is the following:
>
> # gluster volume replace-brick gv0 server1:/data/brick2
> server3:/data/brick2 commit
> # gluster volume add-brick gv0 server1:/data/brick2 server3:/data/brick1
> # gluster volume rebalance gv0 start
>
> These step could be automated by gluster, by adding a new command,
> something like:
> # gluster volume add-server gv0 server3
> # gluster volume rebalance gv0 start
>
> Could also be scripted with bash, but integrating this in gluster
> would be better.
>
> Dump question: I don't have a big cluster here with me, can someone post me a
> "gluster volume info" for a replica 3 with multiple servers and
> multiple bricks per server ?
Which is only true if you have /data/brick{1,2}. If you have only two 
servers with one brick each and you want to add one more server?

It's complicated. I've started trying to do some of this in Salt and 
there's a lot of logic involved to dynamically grow, shrink, and 
otherwise change gluster volumes.