[Gluster-users] Replicated with bricks of different sizes

[John Gardeniers]

Hi Ravi,

I understand that it's not recommended and also understand why.

For the benefit of anyone wondering about the background of this, it's 
worth mentioning that this was going to be a very temporary solution for 
a volume which is currently 5TB and hold less than 3TB of data. The data 
are VM images and we don't expect to be creating any new ones for a 
while, so the volume of data would be very static. A new server is being 
ordered, so at most this solution would be in place for only a few weeks.

regards,
John


On 15/06/15 15:18, Ravishankar N wrote:
>
>
> On 06/15/2015 10:41 AM, John Gardeniers wrote:
>> Hi Victor,
>>
>> Thanks for the response. Since writing my question we have been able 
>> to take over another server, which does have enough drive space for 
>> this job, so it's now merely an academic question.
>>
>
> While there is no check for equal brick sizes during volume create and 
> things will work fine until you hit ENOSPC on the replica brick of 
> smaller size, Viktor is right in saying that it is not recommended.
>
> Regards,
> Ravi
>
>> regards,
>> John
>>
>>
>> On 15/06/15 12:51, Viktor Villafuerte wrote:
>>> Hi John,
>>>
>>> On Mon 15 Jun 2015 09:15:53, John Gardeniers wrote:
>>>> We are/were running a simple 1x2 replica volume and have had a 
>>>> major failure
>>>> on one of our gluster servers. We want to create a new replica, at 
>>>> least for
>>>> the time being, but don't have another available server with 
>>>> sufficient
>>>> drive space. Working on the assumption that we will not increase 
>>>> the disk
>>>> usage, and will therefore not get anywhere near the capacity of the 
>>>> smaller
>>>> server's drives, is it safe to create a replica using mismatched brick
>>>> sizes? Will gluster even allow that?
>>> I did bit of testing of this some time ago. Not with the lastest 
>>> Gluster
>>> mind you... You can have different size bricks in a volume but you
>>> should always have the same size bricks in a replica.
>>>
>>> You say that there's enough space and no space will be consumed after
>>> etc and in theory this could work. But I wouldn't certainly recommend
>>> this kind of setup.
>>>
>>>
>>> v
>>>
>>>
>>>> regards,
>>>> John
>>>>
>>>> _______________________________________________
>>>> Gluster-users mailing list
>>>> Gluster-users at gluster.org
>>>> http://www.gluster.org/mailman/listinfo/gluster-users
>>
>> _______________________________________________
>> 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
> ______________________________________________________________________