Red Hat Bugzilla – Bug 763658
peer detach from one cluster and adding it to another cluster
Last modified: 2011-10-03 00:56:40 EDT
after adding bricks to afr -it created untar kernel to fail on dht and afr clusters.
1.created a dht cluster on brick1 and brick2.
2.created a afr gluster on brick3 and brick4.
and start them.
Now,by mistake,peer probe for new brick 5,brick 6 started from dht cluster instead of afr cluster.
then did peer detach for brick5 and brick6 from dht cluster.
and did peer probe afr gluster -for these new bricks.
now volume info shows dht clusters too on afr bricks and showmount says
afr-brick#showmount -e localhost
Export list for localhost:
#gluster volume info
Volume Name: afr46
Number of Bricks: 2 x 2 = 4
Volume Name: dht46
Number of Bricks: 3
Please update the status of this bug as its been more than 6months since its filed (bug id < 2000)
Please resolve it with proper resolution if its not valid anymore. If its still valid and not critical, move it to 'enhancement' severity.
Lets see what needs to be done here. Check the behavior again, and we will come up with a solution for this. Work with Lakshmipati on this.
The problem here is that volume files of the cluster are leftover in the peer when peer detach happens. So when the peer is added to a new cluster the volumes from the old cluster are added to the new cluster even if those volumes do not belong to the new cluster.
A solution would be to purge the detached peer of any volumes belonging to the cluster from which it was detached. This could be at the time of detach, or as a new command which could clean unneeded volume configs.
Any other ideas?
CHANGE: http://review.gluster.com/431 (Performs cleanup on the detached peer and in the cluster after a) merged in master by Vijay Bellur (email@example.com)
*** Bug 1785 has been marked as a duplicate of this bug. ***