[HIVE-15435] HiveCLI: two jobs are submitted consecutively when tez is used

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 2.1.1
• Component/s: CLI, Hive, Tez

Environment

Hadoop 2.7.3, Hive 2.1.1, Tez 0.8.4

• Assignee: Unassigned
• Reporter: Alexandre Linte
• Created: 15/Dec/16 13:40
• Updated: 20/Dec/16 07:48

Description

Since Hive 2.1.1 (Hive 2.1.0 was OK), the HiveCLI has a strange behavior when submitting a job with Tez as the exection engine.

Doing a query like:

hive (shfs3453)> set hive.execution.engine=tez;
hive (shfs3453)> set tez.queue.name=HQ_OLPS;
hive (shfs3453)> SELECT COUNT(year) FROM chicagocrimes GROUP BY year;

The query will submit two jobs: one submitted in the queue defined, one submitted in the queue default. Both finish successfully. But only the second job processes the data.
You can find attached a screenshot.

This behavior is not visible when using beeline.