Explore queries running mapreduce jobs fail if Hive optimization is turned on

Description

When running an explore query that kicks off a mapreduce and the mapreduce fails, the UI shows finished with a count of 0. This happens in the new and original UIs.

Release Notes

None

Attachments

1

Activity

Show:

Matt Wuenschel February 17, 2017 at 11:10 PM

This issue doesn't exist in HDP 2.4.x. With the property hive.compute.query.using.stats set to true, the query executed successfully

Amabri 2.4.2
HDP 2.4.3
Hive 1.2.1.2.4

Matt Wuenschel February 17, 2017 at 10:11 PM

In HDP 2.5.x when the setting hive.compute.query.using.stats is set to true, the query runs on stats stored in the metastore instead of triggering a mapreduce. CDAP returns 0 each time this type of query is run.

Amabri 2.4.2
HDP 2.5.3
Hive 1.2.1000

Sreevatsan Raman February 16, 2017 at 1:57 AM

Is Hive configured correctly? Can you provide more details when this happens, so we can figure out the cause and fix appropriately.

Ajai Narayanan February 16, 2017 at 1:49 AM

Checked the REST api. The stauts seems to be returned as FINISHED for the UI. Assigning it to to assign it to the person working on explore.

Pinned fields
Click on the next to a field label to start pinning.

Details

Assignee

Reporter

Affects versions

Components

Sprint

Priority

Created February 16, 2017 at 12:30 AM
Updated August 3, 2020 at 7:25 PM