[Gluster-users] Move brick to new Node ( Distributed - Replicated Mode )

Anuradha Talur atalur at redhat.com
Fri Aug 12 08:18:52 UTC 2016



----- Original Message -----
> From: "tecforte jason" <tecforte.jason at gmail.com>
> To: "Anuradha Talur" <atalur at redhat.com>
> Cc: gluster-users at gluster.org
> Sent: Friday, August 12, 2016 1:30:21 PM
> Subject: Re: [Gluster-users] Move brick to new Node ( Distributed - Replicated Mode )
> 
> Hi Anuradha,
> 
> Thanks for the advice, the version i use is 3.8.
> 
> You are correct with my requirement with original 2 nodes ( Distributed -
> Replicated Mode ) expanded to 3 nodes  ( Distributed - Replicated Mode ).May
> i know the additional work for heal to be triggered based on the glusterfs
> version 3.8 ?
> 
If you are in version 3.8 then there is no additional work to be done.

> My another question is : IF i only have 2 bricks in each nodes,
> any possibility add the 4th node ( also only have 2 bricks ) or 5th node
> etc.. to the cluster ? With the condition cannot add more brick to existing
> node.
> 
> 
I'm not sure I get what you are trying to ask.
You can add 4th and 5th node the same way you would be adding the third one.
The main thing that you would need to plan is how you want to pair the replica
bricks.
If you are going to add 4th and 5th node together, you can peer probe them into
the cluster first.

Assume the bricks on them as follows:
node4:/exp4/brick1 and node4:/exp4/brick2
node5:/exp5/brick1 and node5:/exp5/brick2

Then run the following command:
gluster v add-brick <volname> node4:/exp4/brick1 node5:/exp5/brick1 node4:/exp4/brick2
node5:/exp5/brick2

This should add all new four bricks into the volume.

> Thank you for the guide.
> 
> Jason
> 
> 
> On Fri, Aug 12, 2016 at 3:19 PM, Anuradha Talur <atalur at redhat.com> wrote:
> 
> >
> >
> > ----- Original Message -----
> > > From: "tecforte jason" <tecforte.jason at gmail.com>
> > > To: gluster-users at gluster.org
> > > Sent: Friday, August 12, 2016 5:41:07 AM
> > > Subject: [Gluster-users] Move brick to new Node ( Distributed -
> > Replicated    Mode )
> > >
> > > Hi,
> > >
> > > If i have existing 2 nodes with Distributed - Replicated setup like
> > below :
> > >
> > > gluster volume create test-volume replica 2
> > > node1:/exp1/brick1 node2:/exp2/brick2
> > > node1:/exp1/brick3 node2:/exp2/brick4
> > > node1:/exp1/brick5 node2:/exp2/brick6
> > >
> > > And now i want to add another new node to the gluster and
> > > node2:/exp2/brick2 replicated with the new brick node3"/exp3/brick7
> > >
> > > with the condition cannot add brick to existing node 1 and node 2.
> >
> > What version of glusterfs are you using?
> >
> > If I understand correctly, the current volume configuration is as follows:
> >
> > node1:/exp1/brick1 node2:/exp2/brick2 (replica of each other)
> > node1:/exp1/brick3 node2:/exp2/brick4
> > node1:/exp1/brick5 node2:/exp2/brick6
> >
> > And you want to change it to:
> >
> > node2:/exp2/brick2 node3:/exp3/brick7 (replica of each other)
> > node1:/exp1/brick3 node2:/exp2/brick4
> > node1:/exp1/brick5 node2:/exp2/brick6
> >
> > You can do the following:
> > gluster peer probe node 3 (run this from one of the nodes in your cluster)
> > gluster v replace-brick <volname> node1:/exp1/brick1 node3:/exp3/brick7
> > commit force
> >
> > This should now make node2:/exp2/brick2 and node3:/exp3/brick7 as replica
> > of each other.
> >
> > You might have to do some additional work for heal to be triggered based
> > on the glusterfs version
> > you are using. So do mention the gluster version being used.
> >
> > Let me know if my understanding of the requirement given was incorrect.
> >
> > Hope this helps.
> > >
> > > May i know how to do this ?
> > >
> > > Appreciate for the advice.
> > >
> > > Thanks
> > > Jason
> > >
> > > _______________________________________________
> > > Gluster-users mailing list
> > > Gluster-users at gluster.org
> > > http://www.gluster.org/mailman/listinfo/gluster-users
> >
> > --
> > Thanks,
> > Anuradha.
> >
> 

-- 
Thanks,
Anuradha.


More information about the Gluster-users mailing list