[HDFS-9661] Deadlock in DN.FsDatasetImpl between moveBlockAcrossStorage and createRbw

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.7.0, 2.8.0, 2.7.1, 2.7.2
• Fix Version/s: 2.8.0, 2.7.3, 3.0.0-alpha1
• Component/s: datanode
• Target Version/s: 2.7.3
• Hadoop Flags: Reviewed
• Assignee: chunde
• Reporter: chunde
• Created: 19/Jan/16 07:37
• Updated: 06/Jan/17 07:42
• Resolved: 20/Jan/16 15:48

Description

We found a deadlock in dn.FsDatasetImpl between moveBlockAcrossStorage and createRbw from rpc call: replaceBlock/writeBlock. The dn's jstack result is

Issue Links

• duplicates: HDFS-9783 DataNode deadlocks after running hdfs mover
• is broken by: HDFS-7489 Incorrect locking in FsVolumeList#checkDirs can hang datanodes
• relates to: HDFS-7310 Mover can give first priority to local DN if it has target storage type available in local DN