[Gluster-users] disperse configuration example

[Xavier Hernandez]

Hi Pedro,

On 11/12/2014 10:56 AM, Pedro Serotto wrote:
> Hi Xavi,
> what about if I need to add more bricks than 3 ?

Currently the number of bricks in a disperse set is fixed, as it is in a 
replica set.

Basically you decide the level of reliability of the volume when you 
create it (proportion between number of bricks and redundancy). 
Redundancy determines how many bricks can fail without losing access to 
data (for a fixed redundancy, the probability of losing access to data 
increases as the number of bricks increases).

Once defined you can increase the total volume space by adding more 
disperse sets using add-brick command:

gluster volume add-brick <volname> <bricks...>

The number of bricks you add must be equal to the number of bricks 
defined in the creation command, so if you created a 'disperse 3 
redundancy 1' volume, you can grow it by adding 3 more bricks. For a 
'disperse 5 redundancy 1', you should add 5 bricks.

This is called a Distributed-Dispersed volume, similar to the 
Distributed-Replicated volume.

Xavi

>
> BR
>
> Pedro
>
>
> El Miércoles 12 de noviembre de 2014 10:09, Xavier Hernandez
> <xhernandez at datalab.es> escribió:
>
>
> Hi Pedro,
>
> you can create a dispersed volume with 3 bricks and 1 of redundancy (or
> parity, like in RAID5) using the following command:
>
> gluster volume create <volname> disperse 3 redundancy 1 <brick1>
> <brick2> <brick3>
>
> A RAID6-like configuration would use 2 as a redundancy value, and at
> least 5 bricks are needed.
>
> The number of bricks must be always greater than 2 times the redundancy.
>
> Xavi
>
> On 11/11/2014 04:56 PM, Pedro Serotto wrote:
>  > Dear all,
>  > can someone please show me a configuration with disperse for 3 nodes, in
>  > order to obtain a redundancy similar to raid 5?
>  >
>  > BR
>
>  >
>  >
>  >
>  > _______________________________________________
>  > Gluster-users mailing list
>  > Gluster-users at gluster.org <mailto:Gluster-users at gluster.org>
>  > http://supercolony.gluster.org/mailman/listinfo/gluster-users
>
>  >
>
>