Review Request 19161: Heartbeats are not being sent when DbLockMgr is used and an operation holds locks

[Alan Gates <ga...@hortonworks.com>]

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/19161/
-----------------------------------------------------------

Review request for hive and Ashutosh Chauhan.


Bugs: HIVE-6635
    https://issues.apache.org/jira/browse/HIVE-6635


Repository: hive-git


Description
-------

Added a thread to Driver to send heartbeats. This thread only runs during the main loop in Driver.execute. I added this in a separate thread because otherwise I would have needed to add threads in every task to see if heartbeats needed to be sent. This would be very invasive, and also it's not clear it would be possible to cover all cases as there are actions that may simply take a long time (like certain metastore operations). The downside is that a query will keep running even after it's found out it's locks were aborted and only be terminated at the end.


Diffs
-----

  metastore/src/java/org/apache/hadoop/hive/metastore/txn/TxnHandler.java 4441c2f 
  ql/src/java/org/apache/hadoop/hive/ql/Driver.java 7dbb8be 
  ql/src/java/org/apache/hadoop/hive/ql/lockmgr/DbLockManager.java 535912f 
  ql/src/java/org/apache/hadoop/hive/ql/lockmgr/DbTxnManager.java 7773f66 

Diff: https://reviews.apache.org/r/19161/diff/


Testing
-------

Ran unit tests specific to transaction operations, as well as manual system testing.


Thanks,

Alan Gates