[Gluster-users] Number of Bricks 1 x (2 + 1) = 3 and arbiter disappearing

[Ravishankar N]

On 05/11/2016 03:01 PM, Nicola Petracchi wrote:
> Hi, thx for the quick reply
>
> I have this configuration
> Description: Ubuntu 14.04.4 LTS
> Release: 14.04
> Codename: trusty
> Linux 3.13.0-86-generic #130-Ubuntu SMP
>
> ppa:gluster/glusterfs-3.7
>
> ii  glusterfs-client                         3.7.11-ubuntu1~trusty1
>            amd64        clustered file-system (client package)
> ii  glusterfs-common                         3.7.11-ubuntu1~trusty1
>            amd64        GlusterFS common libraries and translator
> modules
> ii  glusterfs-server                         3.7.11-ubuntu1~trusty1
>            amd64        clustered file-system (server package)

Okay so you are running 3.7.11.
1. Was your volume created when you were running an older version?
2.What is the glusterd op-version on all nodes? (`cat 
/var/lib/glusterd/glusterd.info|grep operating-version`)
3.Does /var/lib/glusterd/vols/gvo10/info have 'arbiter_count=1' ?

-Ravi
>
> N
>
> 2016-05-11 10:57 GMT+02:00 Ravishankar N <ravishankar at redhat.com>:
>> On 05/11/2016 02:17 PM, Nicola Petracchi wrote:
>>> Hello, I have deployed a gluster configuration of this type:
>>> Number of Bricks: 1 x (2 + 1) = 3
>>> 2 data + 1 arbiter
>>>
>>> After deploy, all machines were reporting the correct volume info
>>> informations:
>>> Volume Name: gvol0
>>> Type: Replicate
>>> Status: Started
>>> Number of Bricks: 1 x (2 + 1) = 3
>>> Transport-type: tcp
>>> Bricks:
>>> Brick1: pc01:/var/lib/gvol0/brick1
>>> Brick2: pc02:/var/lib/gvol0/brick2
>>> Brick3: pcgw:/var/lib/gvol0/brickgw (arbiter)
>>>
>>> Then, I had to restart gluster service on brick2, after that, only on
>>> that machine the volume info was:
>>> Volume Name: gvol0
>>> Type: Replicate
>>> Status: Started
>>> Number of Bricks: 1 x 3 = 3
>>> Transport-type: tcp
>>> Bricks:
>>> Brick1: pc01:/var/lib/gvol0/brick1
>>> Brick2: pc02:/var/lib/gvol0/brick2
>>> Brick3: pcgw:/var/lib/gvol0/brickgw
>>> others nodes were still reporting the original volume info informations.
>>>
>>> This thing appears strange to me, since I suppose all the members of a
>>> cluster should be consistent on output of informations of a "shared"
>>> resource.
>>>
>>> Btw, later on I had to reboot the arbiter server and again, on that
>>> server:
>>> Volume Name: gvol0
>>> Type: Replicate
>>> Status: Started
>>> Number of Bricks: 1 x 3 = 3
>>> Transport-type: tcp
>>> Bricks:
>>> Brick1: pc01:/var/lib/gvol0/brick1
>>> Brick2: pc02:/var/lib/gvol0/brick2
>>> Brick3: pcgw:/var/lib/gvol0/brickgw
>>>
>>> Only one node, the one never rebooted is stating that brick3 is an
>>> arbiter, aside this information the arbiter node is working as an
>>> arbiter since the data folder contains empty files and is updated with
>>> the others.
>>>
>>> Is this a situation to solve? an issue?
>> What version of gluster are you running? This was bug that got fixed
>> (http://review.gluster.org/12479) in 3.7.7. I would advise you to use the
>> latest release 3.7.11 if you are trying out arbiter.
>> Hope this helps,
>> Ravi
>>> Considerations are appreciated.
>>> Regards
>>> N
>>> _______________________________________________
>>> Gluster-users mailing list
>>> Gluster-users at gluster.org
>>> http://www.gluster.org/mailman/listinfo/gluster-users
>>
>>
>
>