Description of problem:
gluster peer probe localhost.localdomain
peer probe: success. Probe on localhost not needed
gluster --mode=script --wignore volume create r3 replica 3 localhost.localdomain:/home/gfs/r3_0 localhost.localdomain:/home/gfs/r3_1 localhost.localdomain:/home/gfs/r3_2
volume create: r3: success: please start the volume to access data
gluster --mode=script volume start r3
volume start: r3: success
mkdir: cannot create directory ‘/mnt/r3’: File exists
mount -t glusterfs localhost.localdomain:/r3 /mnt/r3
# cd /mnt/r3
# touch abc
Attach the mount process in gdb and put a break point on function afr_lock()
From second terminal:
# exec 200>abc
# echo abc >&200
# When the break point is hit, on third terminal execute "gluster volume stop r3"
# quit gdb
# execute "gluster volume start r3 force"
# On the first terminal execute "exec abc >&200" again and this command hangs.
Version-Release number of selected component (if applicable):
REVIEW: https://review.gluster.org/22515 (cluster/afr: Remove local from owners_list on failure of lock-acquisition) posted (#1) for review on master by Pranith Kumar Karampuri
REVIEW: https://review.gluster.org/22515 (cluster/afr: Remove local from owners_list on failure of lock-acquisition) merged (#6) on master by Pranith Kumar Karampuri