[Gluster-users] Safely remove one replica

[Ravishankar N]

On 06/25/2015 03:07 AM, John Gardeniers wrote:
> No takers on this one?
>

Hi John,

If you either replace a brick of a replica or increase the replica count 
by adding another brick, you will need to perform `gluster volume heal 
<volname> full` to sync the data into the new/replaced brick.
If you are running glusterfs 3.6 or newer. there is a bug (underfix ) in 
'heal <volname> full` due to which heal won't be triggered in some 
scenarios, in which case you can trigger lookups from the mount (find 
/mount-poin |xargs stat). Lookups  would also trigger heals.

You can then remove the replaced brick if there are no more pending heals.

HTH,
Ravi

[1] https://bugzilla.redhat.com/show_bug.cgi?id=1112158


> On 22/06/15 14:37, John Gardeniers wrote:
>> Until last weekend we had a simple 1x2 replicated volume, consisting 
>> of a single brick on each peer. After a drive failure screwed the 
>> brick on one peer we decided to create a new peer and swap the 
>> bricks. Running "gluster volume replace-brick gluster-rhev 
>> dead_peer:/gluster_brick_1 new_peer:/gluster_brick_1 commit force".
>>
>> After trying for some time and not wishing to rely on a single peer 
>> we added kari as an additional replica with "gluster volume add-brick 
>> gluster-rhev replica 3 new_peer:/gluster_brick_1 force".
>>
>> Can we now *safely* remove the dead brick and revert back to replica 2?
>>
>> regards,
>> John
>>
>> _______________________________________________
>> Gluster-users mailing list
>> Gluster-users at gluster.org
>> http://www.gluster.org/mailman/listinfo/gluster-users
>>
>> ______________________________________________________________________
>> This email has been scanned by the Symantec Email Security.cloud 
>> service.
>> For more information please visit http://www.symanteccloud.com
>> ______________________________________________________________________
>
> _______________________________________________
> Gluster-users mailing list
> Gluster-users at gluster.org
> http://www.gluster.org/mailman/listinfo/gluster-users