Deadlock when cluster.close() called

Description

This happens intermittently on systems which repeatedly enter and leave the hive shell. For example several dozen hive shell jobs in serial. For some of these jobs hive will hang and the process will remain frozen.

The thread dump is

cluster.close() is blocked

Environment

None

Pull Requests

None

Activity

Show:
Russell Spitzer
August 14, 2014, 4:41 PM

No reproduction with patched Jar after 45 minutes. Previously I could repo this in about 5 minutes. I will continue to run the test.

Russell Spitzer
August 14, 2014, 5:17 PM

Still no repo at 1 hour 20 minutes

Russell Spitzer
August 14, 2014, 8:58 PM
Edited

Still working well at 5 hours in
I'm +1 on this fix, but i'll let this run overnight just in case

Russell Spitzer
August 15, 2014, 5:41 PM

Ran for about 24 hours no problems!

Alex Liu
August 16, 2014, 12:15 AM

@Russell @Olivier Michallat can you attach the path to this ticket, I will include it in my custom build.

Fixed

Assignee

Olivier Michallat

Reporter

Alex Liu

Labels

None

PM Priority

None

Reproduced in

None

Affects versions

Fix versions

Pull Request

None

Doc Impact

None

Size

None

External issue ID

None

External issue ID

None

Time remaining

0m

Components

Priority

Major
Configure