[
https://issues.apache.org/jira/browse/HIVE-25184?focusedWorklogId=607005&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-607005
]
ASF GitHub Bot logged work on HIVE-25184:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 04/Jun/21 11:03
Start Date: 04/Jun/21 11:03
Worklog Time Spent: 10m
Work Description: kgyrtkirk commented on pull request #2337:
URL: https://github.com/apache/hive/pull/2337#issuecomment-854618988
actually; its quite nice that almost all the reoptimization tests have
fallen because of this change :)
Testing / split-14 / PostProcess /
org.apache.hadoop.hive.cli.split19.TestMiniLlapLocalCliDriver.testCliDriver[retry_failure_oom]
Testing / split-14 / PostProcess /
org.apache.hadoop.hive.cli.split19.TestMiniLlapLocalCliDriver.testCliDriver[retry_failure_reorder]
Testing / split-14 / PostProcess /
org.apache.hadoop.hive.cli.split19.TestMiniLlapLocalCliDriver.testCliDriver[retry_failure_stat_changes]
Testing / split-14 / PostProcess /
org.apache.hadoop.hive.ql.plan.mapping.TestReOptimization.testReOptimizationCanSendBackStatsToCBO
Testing / split-14 / PostProcess /
org.apache.hadoop.hive.ql.plan.mapping.TestReOptimization.testReExecutedIfMapJoinError
Testing / split-14 / PostProcess /
org.apache.hadoop.hive.ql.plan.mapping.TestReOptimization.testStatsAreSetInReopt
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
Issue Time Tracking
-------------------
Worklog Id: (was: 607005)
Time Spent: 40m (was: 0.5h)
> ReExecDriver Only Run afterExecute If No Exceptions
> ---------------------------------------------------
>
> Key: HIVE-25184
> URL: https://issues.apache.org/jira/browse/HIVE-25184
> Project: Hive
> Issue Type: Improvement
> Reporter: David Mollitor
> Assignee: David Mollitor
> Priority: Minor
> Labels: pull-request-available
> Time Spent: 40m
> Remaining Estimate: 0h
>
> Was recently troubleshooting an issue and noticed a NPE in the logs. I
> tracked it down to {{ReExecDriver}} code. The "afterExecute" code gets
> called if the Driver call succeed or fails. However, if there is a failure,
> the Driver is instructed to "clean up" by some internal try-catch and so
> there the afterExecute code fails with a NPE when it tried to read state out
> of the Driver class.
>
> [https://github.com/apache/hive/blob/1cc87d09cf0514f3fb962a816babb7eea859163c/ql/src/java/org/apache/hadoop/hive/ql/reexec/ReExecDriver.java#L163-L170]
>
> Move this afterExecute code into the try-catch block so it's only executed on
> success (and there is valid state within the Driver). I looked at the code a
> bit and it seems like the only listener that handles this afterExecute code
> assumes the state is always valid, so there is currently no way to pass it
> 'null' on a failure or 'state' on a success.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
