Description of problem: ganesha.sh has a operation called teardown, this operation helps bring down the nfs-ganesha related HA based cluster. As per the operation it is suppose to delete the cluter.conf from /etc/cluster/ dir. Presently ganesha.sh --teardown just deletes the cluster.conf directory from the node where the script is executed not from all the nodes of the cluster in consideration. Version-Release number of selected component (if applicable): glusterfs-3.7dev-0.910.git17827de.el6.x86_64 nfs-ganesha-2.2-0.rc7.el6.x86_64 How reproducible: always Steps to Reproduce: 1. bring the nfs ganesha ha using "gluster features.ganesha enable" 2. once the cluster is up, try bring it down using the command "ganesha.sh --teardown"
This causes some old cluster info to be present on those nodes. And a subsequent setup doesn't work the way it should. I am not sure as to who should clean this, pacemaker or the script itself. Assigning it to Kaleb, who'd have more information.
Posted the changes for review, http://review.gluster.org/#/c/10283/
Kaleb has posted changes as part of the following patch, Master, http://review.gluster.org/#/c/10234/ Release-3.7 http://review.gluster.org/#/c/10318/ and merged on both of them
fixed long ago, not sure how this didn't closed