Description of problem: 2 replica volume. Each brick running on different peers. Now bring the network of one of the peers down (say peer1). Create some files on the mount point. Bring peer1's network up. Now peer1's glustershd thinks that peer2's brick came up and hence does not initiate a crawl since its not local brick. In the sameway from peer2's point of view peer1's brick came up which his remote. Hence it also does not initiate the crawl, thus self-heal not being triggered even though connection has been established. For self-heal to happen one has to wait till the glustershd starts crawling for every 10 minutes. Version-Release number of selected component (if applicable): How reproducible: always Steps to Reproduce: 1. create a 2 replica volume, start it and then mount it. 2. Bring the network of one of the peers down 3. Create some files on the mount point 4. bring the network up Actual results: self-heal does not happen when there is connection establishment with the brick after a down network is brought up. Expected results: glustershd should crawl whenever there is a connection establishment with a brick which is part of the replica set it also belongs to. Additional info:
I was sure I remembered a commit for this, and I found it. http://review.gluster.org/3972 Moving to MODIFIED accordingly.