Created attachment 1020467 [details] NFS process statedump Description of problem: ======================= client mount and IO hung after the bricks came back up with gluster v start force. This is seen with the pathch for ec heal sent by pranith Version-Release number of selected component (if applicable): ============================================================ [root@ninja ~]# gluster --version glusterfs 3.8dev built on Apr 30 2015 11:14:47 Repository revision: git://git.gluster.com/glusterfs.git Copyright (c) 2006-2011 Gluster Inc. <http://www.gluster.com> GlusterFS comes with ABSOLUTELY NO WARRANTY. You may redistribute copies of GlusterFS under the terms of the GNU General Public License. How reproducible: ================= 100% Steps to Reproduce: 1. NFS mount a distributed disperse volume on a client 2. Run IO, creating directories and files 3. Bring down 2 of the bricks and start the volume with force options after some time and check the client mount and IO Actual results: Expected results: Additional info: ================ Attaching the nfs process statedump.
Last I checked soumya and Niels are working on a similar bug. Assigning it to Niels for now.
As of glusterfs-6.x have not heard of this issue. Considering we haven't run a test to validate this, marking it as DEFERRED for now, and revisiting this after couple of releases. If you find the issue is happening still with glusterfs-6.x releases, feel free to re-open it.