We had a failed brick, bad disk. We replaced the drive and are following the documentation (See "Replacing brick in Replicate/Distributed Replicate volumes" in http://gluster.readthedocs.org/en/latest/Administrator%20Guide/Managing%20Volumes/#replace-brick). However this seems odd. If I have a six node cluster, each with 1 brick, where I have three-way replication, I'll end up data distributed across two bricks which in turn are replicated three ways. In this scenario, if a node goes down for a time and comes back, its brick will get self-healed to match the other replicas and life goes on. Why can't I do the same with that one brick? Just take it out of service, replace the brick, remount and allow it to self-heal?
@Peter, the changes proposed for this bug, https://bugzilla.redhat.com/show_bug.cgi?id=1266876 seem to line up to your requests. Could you take a look at the commit message for the change here [1], and add your comments on weather this satisfies your request? [1] http://review.gluster.org/#/c/12250/
@Shyamsundar, I posted a comment on what I would like to see for a command syntax in that commit. Thanks!
*** This bug has been marked as a duplicate of bug 1266876 ***
Closed as duplicate as same patch being worked on for bug 1266876 will fix this too. If there is any problem with the functionality then please raise bug. Thanks, Anuradha