[Gluster-users] Convert replica 2 to replica 3 arbiter 1

[Russell Purinton]

That is what I understood as well, the last brick is the arbiter brick, but I had no idea how to expect the add-brick function to behave with regard to how it would figure this out.  One way I’m sure should work would be just creating new volumes, rsyncing the data.  Also renaming the volumes after.


> On Mar 29, 2016, at 1:36 PM, Atin Mukherjee <atin.mukherjee83 at gmail.com> wrote:
> 
> 
> -Atin
> Sent from one plus one
> On 29-Mar-2016 10:43 pm, "Russell Purinton" <russell.purinton at gmail.com <mailto:russell.purinton at gmail.com>> wrote:
> >
> > I have used the add-brick command in the past but I’m not sure how to go about using it to do this…
> >
> > Could anyone provide an example?
> >
> > As shown below, the current setup, xs141 is replica with xs138.  xs139 is replica with xs140.  I would like to have 139 serve as arbiter for 141 and 138.   And I’d like 141 to serve as arbiter for 139 and 140.
> 
> I don't think that's possible. IIRC, the last brick in the replica pair is considered to be an arbiter brick and I am not aware of any changes in the  behaviour post that. So in short, if you execute add-brick then all the new bricks will be considered as arbiter IMO.
> 
> Ravi can correct me otherwise.
> >
> >> On Mar 29, 2016, at 1:07 PM, Atin Mukherjee <atin.mukherjee83 at gmail.com <mailto:atin.mukherjee83 at gmail.com>> wrote:
> >>
> >> -Atin
> >> Sent from one plus one
> >> On 29-Mar-2016 10:21 pm, "Russell Purinton" <russell.purinton at gmail.com <mailto:russell.purinton at gmail.com>> wrote:
> >> >
> >> > I have several existing volumes configured in the following way:
> >> >
> >> > Volume Name: sonic001
> >> > Type: Distributed-Replicate
> >> > Volume ID: 362eb3a8-c211-4f35-bcd3-87785c51f15c
> >> > Status: Started
> >> > Number of Bricks: 4 x 2 = 8
> >> > Transport-type: tcp
> >> > Bricks:
> >> > Brick1: xs141:/brick1/sonic001p0r0
> >> > Brick2: xs138:/brick1/sonic001p0r1
> >> > Brick3: xs139:/brick1/sonic001p1r0
> >> > Brick4: xs140:/brick1/sonic001p1r1
> >> > Brick5: xs141:/brick0/sonic001p2r0
> >> > Brick6: xs138:/brick0/sonic001p2r1
> >> > Brick7: xs139:/brick0/sonic001p4r0
> >> > Brick8: xs140:/brick0/sonic001p3r1
> >> >
> >> > I recently learned that replica 2 can’t be considered HA because of the split brain scenario that would occur if a network partition occurred.   I also recently learned that using replica 3 arbiter 1 was a way to solve this problem without consuming additional space.
> >> >
> >> > Is there a way I can change this existing volume to be replica 3 arbiter 1, or do I need to create new volumes and rsync the data?
> >>
> >> I think you should be able to convert the volume topology using add-brick command.
> >>
> >> >
> >> > Thanks,
> >> >
> >> > Russ
> >> >
> >> > _______________________________________________
> >> > Gluster-users mailing list
> >> > Gluster-users at gluster.org <mailto:Gluster-users at gluster.org>
> >> > http://www.gluster.org/mailman/listinfo/gluster-users <http://www.gluster.org/mailman/listinfo/gluster-users>
> >>
> >> -Atin
> >> Sent from one plus one
> >
> >

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20160329/5d49e169/attachment.html>