[Gluster-users] mount with one alive node in replica set

[Игорь Бирюлин]

Thank you very much for your detailed description.
I've understood how use glusterfs in this situation.

2015-06-19 4:41 GMT+03:00 Ravishankar N <ravishankar at redhat.com>:

>
>
> On 06/19/2015 01:06 AM, Игорь Бирюлин wrote:
>
>  Is it a bug?
>  How can I understand that volume stopped if in "gluster volume info" I
> see "Status: Started"?
>
>
> Not a bug. If `volume info ` says it is started and  `gluster volume
> status` shows gluster processes as offline, it means something is wrong and
> needs investigation. In your case you need to see why the other node is
> down.  Having identified the second node is down and cannot be brought back
> anytime soon, the `volume start force`  is just a workaround to restart the
> gluster processes.
>
>
>
> 2015-06-18 22:07 GMT+03:00 Atin Mukherjee <atin.mukherjee83 at gmail.com>:
>
>> Sent from one plus one
>> On Jun 18, 2015 8:51 PM, "Игорь Бирюлин" <biryulini at gmail.com> wrote:
>> >
>> > Sorry, I didn't check this, because after reboot my second node I had
>> looked "gluster volume info" only and I have found "Status: Started".
>> > Now I've checked your recomendation and you are right!
>> > "gluster volume start <volname> force" didn't changed output of
>> "gluster volume info" but I have mounted my share!
>> > Thank you very much for your advice!
>> >
>> > But why does "gluster volume info" show that my volname "started"
>> before "gluster volume start <volname> force"?
>>  In this case glusterd stops the brick processes but doesn't mark its
>> status as stopped.
>>
>> Ravi, correct me if I am wrong.
>>
>> >
>> >
>> >
>> > 2015-06-18 14:18 GMT+03:00 Ravishankar N <ravishankar at redhat.com>:
>> >>
>> >>
>> >>
>> >> On 06/18/2015 04:25 PM, Игорь Бирюлин wrote:
>> >>>
>> >>> Thank you for you answer!
>> >>>
>> >>> I check recomendation:
>> >>> 1. On first node closed all connection from second node by iptables.
>> Check that on both nodes "gluster peer status" return  "Disconnected".
>> Check that on both nodes share was mounted and work well like local file
>> system.
>> >>> 2. Rebooted second node (remind first node closed by iptables).
>> Second node booted without problem and proccesses of glusterfs started:
>> >>> # ps aux | grep [g]luster
>> >>> root      4145  0.0  0.0 375692 16076 ?        Ssl  13:35   0:00
>> /usr/sbin/glusterd -p /var/run/glusterd.pid
>> >>>
>> >>> "gluster peer status" return "Disconnected" and volume started on
>> localhost:
>> >>> # gluster volume info
>> >>>  Volume Name: files
>> >>> Type: Replicate
>> >>> Volume ID: 41067184-d57a-4132-a997-dbd47c974b40
>> >>> Status: Started
>> >>> Number of Bricks: 1 x 2 = 2
>> >>> Transport-type: tcp
>> >>> Bricks:
>> >>> Brick1: xxx1:/storage/gluster_brick_repofiles
>> >>> Brick2: xxx2:/storage/gluster_brick_repofiles
>> >>>
>> >>> Вut I cann't mount this volume:
>> >>> # cat /etc/fstab |grep gluster
>> >>> 127.0.0.1:/files                            /repo
>> glusterfs       rw,_netdev              0 0
>> >>> # mount /repo
>> >>> Mount failed. Please check the log file for more details.
>> >>>
>> >>> Part of log I have sent in first message.
>> >>>
>> >>> If I will open first node by iptables I could mount without problem,
>> but what must I do, when I lost one node and I have probability reboot
>> another node?
>> >>>
>> >>>
>> >> `gluster volume start <volname> force` doesn't work?
>> >>
>> >>>
>> >>> 2015-06-17 18:46 GMT+03:00 Ravishankar N <ravishankar at redhat.com>:
>> >>>>
>> >>>>
>> >>>>
>> >>>> On 06/17/2015 07:04 PM, Игорь Бирюлин wrote:
>> >>>>>
>> >>>>> If we turn off one server, another will be work and mounted volume
>> will be use without problem.
>> >>>>> But if we rebooted our another server, when first was turned off
>> (or gluster was stopped on this server), our volume cann't mount (glusterd
>> started).
>> >>>>
>> >>>> If both nodes are down and you bring up only one node, glusterd will
>> not start the volume (i.e. the brick, nfs and glustershd processes)
>> automatically. It waits for the other node's glusterd also to be up so that
>> they are in sync. You can override this behavior by doing a `gluster volume
>> start <volname> force` to bring up the gluster process only on this node
>> and then mount the volume.
>> >>>>
>> >>>> -Ravi
>> >>>
>> >>>
>> >>
>> >
>> >
>>  > _______________________________________________
>> > Gluster-users mailing list
>> > Gluster-users at gluster.org
>> > http://www.gluster.org/mailman/listinfo/gluster-users
>>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20150619/a542c918/attachment.html>