[Gluster-users] FailOver

[Vijay Bellur]

On 10/23/2014 01:35 PM, Punit Dambiwal wrote:
>     On Mon, Oct 13, 2014 at 11:54 AM, Punit Dambiwal <hypunit at gmail.com
>     <mailto:hypunit at gmail.com>> wrote:
>
>         Hi,
>
>         I have one question regarding the gluster failover...let me
>         explain my current architecture,i am using Ovirt with gluster...
>
>         1. One Ovirt Engine (Ovirt 3.4)
>         2. 4 * Ovirt Node as well as Gluster storage node...with 12
>         brick in one node...(Gluster Version 3.5)
>         3. All 4 node in distributed replicated structure with replica
>         level =2...
>         4. I have one spare node with 12 brick for Failover purpose..
>
>         Now there is two questions :-
>
>         1. If any of brick failed...how i can failover this brick...how
>         to remove the failed brick and replace with another brick....??
>         Do i need to replace the whole node or i can replace the single
>         brick ??
>

Failure of a single brick can be addressed by performing "replace-brick 
commit force" to replace the failed brick with a new brick and then 
trigger self-heal to rebuild data on the replace brick.

>         2. If one of the whole node with 12 brick down and can not come
>         up...how i can replace it with the new one....do i need to add
>         two node to main the replication level... ??
>

You can add a replacement node to the cluster and use "replace-brick 
commit force" to adjust the volume topology. Self-healing will rebuild 
data on the new node. You might want to replace one/few bricks at a time 
to ensure that your servers do not get bogged down by self-healing.

-Vijay