[
https://issues.apache.org/jira/browse/HDFS-796?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12783923#action_12783923
]
Todd Lipcon commented on HDFS-796:
----------------------------------
just realized I was actually running an earlier checkout that didn't have
HDFS-723 included. These stacks aren't exactly the same, so I'll leave it open,
but feel free to resolve as dup if it's the same root cause.
> 4-thread Deadlock in DFS Client
> -------------------------------
>
> Key: HDFS-796
> URL: https://issues.apache.org/jira/browse/HDFS-796
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: hdfs client
> Reporter: Todd Lipcon
> Attachments: deadlock.txt
>
>
> Managed to get this deadlock running TestHDFSCLI this afternoon. It seems to
> be related to a signal, though I don't recall hitting ^C. The one thing I did
> that might help trigger this was reduce the Thread.sleep(1000) in
> LeaseChecker down to a shorter interval.
> "SIGTERM handler":
> waiting to lock monitor 0x0000000041f3bd18 (object 0x00007fa43c132328, a
> java.lang.Class),
> which is held by "SIGINT handler"
> "SIGINT handler":
> waiting to lock monitor 0x000000004230b410 (object 0x00007fa441603030, a
> org.apache.hadoop.fs.FileSystem$Cache$ClientFinalizer),
> which is held by "Thread-3"
> "Thread-3":
> waiting to lock monitor 0x00007fa4341fe3f0 (object 0x00007fa441d3b150, a
> org.apache.hadoop.hdfs.DFSClient$DFSOutputStream),
> which is held by "main"
> "main":
> waiting to lock monitor 0x0000000042b1ec80 (object 0x00007fa44167eff8, a
> org.apache.hadoop.hdfs.DFSClient$LeaseChecker),
> which is held by "Thread-3"
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
