Make the data that's available in a state dump available in heal output. Currently you can only determine the progress of a heal by dumping state on the source brick: [xlator.features.locks.gv-cinder-locks.inode] path=/vol-ad3357ba-c264-4e58-a1ac-1d90cb0f960d mandatory=0 inodelk-count=2 lock-dump.domain.domain=gv-cinder-replicate-1:self-heal inodelk.inodelk[0](ACTIVE)=type=WRITE, whence=0, start=0, len=0, pid = 18446744073709551610, owner=a4fe7f42607f0000, client=0x7f1bdea90e20, connection-id=storage07-29419-2016/01/29-00:00:46:182426-gv-cinder-client-3-0-0, granted at 2016-01-29 00:00:51 lock-dump.domain.domain=gv-cinder-replicate-1 inodelk.inodelk[0](ACTIVE)=type=WRITE, whence=0, start=4801027506176, len=131072, pid = 18446744073709551615, owner=ec425469a27f0000, client=0x7f1bdea8a200, connection-id=compute012-3825-2016/01/18-23:43:04:714824-gv-cinder-client-3-0, granted at 2016-01-29 17:41:31 lock-dump.domain.domain=gv-cinder-replicate-1:metadata We need to know which node is doing the heal (compute012), and how far along it has progressed (4801027506176) in order to answer that inevitable question, "How much longer?" This data should be available to us from the CLI.
Migrated to github: https://github.com/gluster/glusterfs/issues/598 Please follow the github issue for further updates on this bug.