[HDFS-5211] Race condition between DistributedFileSystem#close and FileSystem#close can cause return of a closed DistributedFileSystem instance from the FileSystem cache.

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.2.1
• Fix Version/s: 1-win, 1.3.0
• Component/s: hdfs-client
• Target Version/s: 1.3.0
• Hadoop Flags: Reviewed
• Assignee: Chris Nauroth
• Reporter: Chris Nauroth
• Created: 15/Sep/13 16:07
• Updated: 17/Sep/13 19:46
• Resolved: 17/Sep/13 19:46

Description

DistributedFileSystem#close closes the underlying DFSClient before the superclass call to FileSystem#close, which removes the instance from the cache. This leaves a small window during which another thread can call FileSystem#get and receive the closed instance from the cache.

Issue Links

• relates to: MAPREDUCE-5508 JobTracker memory leak caused by unreleased FileSystem objects in JobInProgress#cleanupJob