[ http://issues.apache.org/jira/browse/HADOOP-139?page=all ]
Igor Bolotin updated HADOOP-139:
--------------------------------
Attachment: deadlock.patch
Attached is proposed patch. I removed entire lock/release method
synchronization and replaced it with critical section synchronization for sets
accesses - leaving channel lock/release calls outside of synchronized
statement.
> Deadlock in LocalFileSystem lock/release
> ----------------------------------------
>
> Key: HADOOP-139
> URL: http://issues.apache.org/jira/browse/HADOOP-139
> Project: Hadoop
> Type: Bug
> Components: fs
> Reporter: Igor Bolotin
> Attachments: deadlock.patch
>
> LocalFileSystem lock/release methods marked synchronized and inside they lock
> file channel - this produces deadlock situation. Let's see how it happens:
> 1. First thread locks the file and starts some long-running process.
> 2. Second thread tries to lock the file and it blocks inside channel lock
> method. It keeps LocalFileSystem instance "locked" as well.
> 3. First thread finished it's processing and tries to release lock - it
> blocks because LocalFileSystem instance is "locked" by second thread - both
> threads are waiting to each other.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://issues.apache.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see:
http://www.atlassian.com/software/jira
