[Gluster-users] gluster-fuse and rebalance (Re: (3.1.X) add-brick during re-balancing)

[Tomoaki Sato]

Shishir,

Thank you for getting back to me.
I have confirmed gNFS servers work fine with only two packages in my environment.

# rpm -ivh --nodeps glusterfs-3.3.0-1.x86_64.rpm glusterfs-server-3.3.0-1.x86_64.rpm

Note: rpm -ivh fails without --nodeps option.

Thanks,

Tomo

(2012/06/28 14:22), Shishir Gowda wrote:
> Hi Tomo,
>
> That is correct. These are harmless, and if you still want to migrate these issues, then use the force option.
>
> With regards,
> Shishir
>
> ----- Original Message -----
> From: "Tomoaki Sato" <tsato at valinux.co.jp>
> To: "Shishir Gowda" <sgowda at redhat.com>
> Cc: gluster-users at gluster.org, "Amar Tumballi" <amar at gluster.com>
> Sent: Thursday, June 28, 2012 10:23:59 AM
> Subject: Re: [Gluster-users] gluster-fuse and rebalance (Re: (3.1.X) add-brick during re-balancing)
>
> Shishir,
>
> Thank you for providing the requested information.
>
> I saw non-zero 'failures' count in output of 'gluster volume rebalance <volume> status' and saw following log massages.
> Let me confirm that these are harmless.
>
> /var/log/glusterfs/vol1-rebalance.log:[2012-06-28 12:49:06.381196] W [dht-rebalance.c:353:__dht_check_free_space] 0-vol1-dht: data movement attempted from node (vol1-client-0) with higher disk space to a node (vol1-client-2) with lesser disk space (/dir1/file-1)
>
> Regards,
>
> Tomo
>
> (2012/06/28 13:29), Shishir Gowda wrote:
>> Hi Tomo,
>>
>> gluster volume rebalance will re-distribute data only when the destination server/brick has higher available space (also taking into account the file to be migrated) than the source server/brick. This is done to maintain a balance across the bricks/servers.
>>
>> A force option would bypass this check, and would migrate the file as long as the destination has enough space to accommodate the new file.
>>
>> With regards,
>> shishir
>>
>> ----- Original Message -----
>> From: "Tomoaki Sato" <tsato at valinux.co.jp>
>> To: "Shishir Gowda" <sgowda at redhat.com>
>> Cc: gluster-users at gluster.org, "Amar Tumballi" <amar at gluster.com>
>> Sent: Thursday, June 28, 2012 9:33:12 AM
>> Subject: Re: [Gluster-users] gluster-fuse and rebalance (Re: (3.1.X) add-brick during re-balancing)
>>
>> Shishir,
>>
>> Thank you for your prompt reply.
>> When should I specify 'force' option for 'gluster volume reparance <volume> start' ?
>>
>> -- excerpt from manual page --
>> 8.5.2. Rebalancing Volume to Fix Layout and Migrate Data
>>
>> After expanding or shrinking a volume (using the add-brick and remove-brick commands respectively),
>> you need to rebalance the data among the servers.
>>
>> To rebalance a volume to fix layout and migrate the existing data
>>
>> • Start the rebalance operation on any one of the server using the following command:
>>     # gluster volume rebalance VOLNAME start
>>     For example:
>>      # gluster volume rebalance test-volume start
>>      Starting rebalancing on volume test-volume has been successful
>>
>> • Start the migration operation forcefully on any one of the server using the following command:
>>     # gluster volume rebalance VOLNAME start force
>>     For example:
>>      # gluster volume rebalance test-volume start force
>>      Starting rebalancing on volume test-volume has been successful
>> -- --
>>
>> Regards,
>>
>> Tomo
>>
>> (2012/06/27 17:32), Shishir Gowda wrote:
>>> Hi Tomo,
>>>
>>> That is correct. The gluster volume rebalance is no more dependent on gluster-fuse package.
>>>
>>> With regards,
>>> Shishir
>>>
>>> ----- Original Message -----
>>> From: "Tomoaki Sato" <tsato at valinux.co.jp>
>>> To: "Amar Tumballi" <amar at gluster.com>
>>> Cc: gluster-users at gluster.org
>>> Sent: Wednesday, June 27, 2012 1:36:44 PM
>>> Subject: [Gluster-users] gluster-fuse and rebalance (Re: (3.1.X) add-brick during re-balancing)
>>>
>>> Amar,
>>>
>>> Let me confirm that gluster-fuse package is not required in NFS-server-use of GlusterFS-3.3.0 as 'glaster volume rebalance' dose not use FUSE.
>>>
>>> Regards,
>>>
>>> Tomo
>>>
>>> (2011/08/29 15:09), Amar Tumballi wrote:
>>>>
>>>>        1. fuse module is required to 'glaster volume rebalance'.
>>>>
>>>>        2. 'gluster volume rebalance start' and 'gluster volume rebalance stop' must be issued in exactly same node.
>>>>
>>>>        By the way, why multiple re-balancing on a volume can be exist in a cluster ?
>>>>
>>>>
>>>> Enhancement to make sure 'rebalance' operation co-operate with other peers is in progress, and will be available in 3.3.0 release. For now, yes, there is a possibility that single volume can have multiple rebalance operations going on (on different machines). Even though it should not break any thing, it will degrade the performance. For now, preventing this should be taken care by admin.
>>>>
>>>> Regards,
>>>> Amar
>>>
>>>
>>> _______________________________________________
>>> Gluster-users mailing list
>>> Gluster-users at gluster.org
>>> http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
>>>
>>
>>
>
>