search for: range_lock

...;txg_wait_open+" 360 bash-3.00# cat nfsd.txt | grep "txg_wait_open+" | grep -c "337d1e" 360 So all those txg_wait_open are for the same transaction group - txg_wait_open+0x58(6000ad0c910, 337d1e, All the other threads were in: bash-3.00# cat nfsd.txt | grep -c "zfs_range_lock_reader+" 289 Several different dnodes (however several threads to the same dnode). Also using ''zpool iostat 1'' while nfsd was shuting down I could see mostly reads and some writes every 40-60s. Why nfsd was closing for almost 3 minutes? Because we were sitting in txg_wait_...