[HIVE-9120] Hive Query log does not work when hive.exec.parallel is true

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.1.0
• Component/s: HiveServer2, Logging
• Assignee: Dong Chen
• Reporter: Dong Chen
• Created: 16/Dec/14 08:52
• Updated: 29/Dec/18 09:49
• Resolved: 23/Dec/14 15:43

Description

When hive.exec.parallel is true, the query log is not saved and Beeline can not retrieve it.
When parallel, Driver.launchTask() may run the task in a new thread if other conditions are also on. TaskRunner.start() is invoked instead of TaskRunner.runSequential(). This cause the threadlocal variable OperationLog to be null and query logs are not logged.
The OperationLog object should be set in the new thread in this case.

Issue Links

• relates to: HIVE-9121 Enable beeline query progress information for Spark job[Spark Branch]