On 02/07/2015 10:01 AM, Emmanuel Dreyfus wrote:> Hello
>
> What is the right way to replace a dead brick with glusterfs 3.6?
>
> The old brick is listen in gluster volume status as being unavaiable.
>
> If I try to remove it I get an error, because the volume is replicated
> and I should remove bricks by pairs.
> # gluster volume remove-brick tmp cubi:/wd1 force
> Removing brick(s) can result in data loss. Do you want to Continue?
> (y/n) y
> volume remove-brick commit force: failed: Remove brick incorrect brick
> count of 1 for replica 2
>
> If I try to replace it by itself I get an error because the brick
> already
> exists.
> # gluster volume replace-brick tmp cubi:/wd1 cubi:/wd1 commit force
> volume replace-brick: failed: Brick: cubi:/wd1 not available. Brick may
> be containing or be contained by an existing brick
>
Performing a replace-brick commit force with an alternate brick and
letting self-heal move data on to the new brick is the way to go as of now.
-Vijay