[HDFS-7531] Improve the concurrent access on FsVolumeList

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.6.0
• Fix Version/s: 2.7.0, 2.6.1, 3.0.0-alpha1
• Component/s: datanode
• Labels: 2.6.1-candidate
• Target Version/s: 2.7.0
• Assignee: Lei (Eddy) Xu
• Reporter: Lei (Eddy) Xu
• Created: 16/Dec/14 04:47
• Updated: 30/Aug/16 01:40
• Resolved: 18/Dec/14 00:42

Description

FsVolumeList uses synchronized to protect the update on FsVolumeList#volumes, while various operations (e.g., checkDirs(), getAvailable()) iterate volumes without protection.

This JIRA proposes to use AtomicReference to encapture volumes to provide better concurrent access.

Issue Links

• relates to: HDFS-7496 Fix FsVolume removal race conditions on the DataNode by reference-counting the volume instances
• supercedes: HDFS-7492 If multiple threads call FsVolumeList#checkDirs at the same time, we should only do checkDirs once and give the results to all waiting threads