HUE 3.12 Hive Editor says success but shows no results.

[Chriss Vincelette]

I admin a HUE cluster used by about 100 users a day.

At least once a week we are encountering a situation where users queries run and show "success" but no results.

Running the same exact query in hive shell returns the expected results. (screenshot attached).

It feels like this is related to:
https://issues.apache.org/jira/browse/HIVE-14227

We do see a large number of invalid sessions mentioned in the hive logs.

We are not able to upgrade our hive (we use EMR). and we can't upgrade HUE from 3.12 since our version has some internal customizations.

Is this a known issue in HUE? Is there a fix from upstream we might be able to back-port to our version?

Right now when this happens we restart hive-server2 but then everyones cached results are lost.

[John Gilik]

I work at the same company as Chriss, and I personally receive a few reports of this issue each week.  I'd like to chime in as a "me too" / "also affected".

[Romain Rigaux]

This one does not ring a bell, what is the query status on the HS2 web ui or logs? Is the 'success' happening right away after the submission?

To be sure: to get rid of it, you need to restart HS2 or everyone stays stuck? (what happens if this user clicks on the 'close/recreate' session buttons on the top right cogs in the SQL Editor)

Sometimes, we get something a bit similar when a query submission hangs for a while or there is a very fast click, the user can submit again and we get two queries.

Also, there is no HS2 HA in your setup?

BTW: if your customizations are not too tied to your company, any upstream pull request is welcomed

--
You received this message because you are subscribed to the Google Groups "Hue-Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hue-user+unsubscribe@cloudera.org.

[Chriss Vincelette]

It can come back immediately as success with no results.

But we’ve also seen queries run for a bit then return as success with no results.

When it happens typically many users see it at around the same time. Recreating their session within the gears icon 

popout seems to clear it up for individuals.

Bouncing hiveserver2 also clears it up for a while but loses everyone’s sessions.

Prior to restarting hive. It’s logs show errors related to invalid session handles.

It seems like the hue->hive connection is still open/valid but the hive session itself is not.

We are using amazon EMR with normal (non-ha hive).

We tried using an HA hive but not being able to sticky to particular hive sessions was a problem for us. Maybe the setup instructions we followed were bad.

As far as contributing code back we would love to. The only reason we haven’t is you folks move pretty fast. :) 

With 4.x out I’m not sure how much of our stuff would be useful to merge back. Most of it is bug fixes specific to 3.12 code that was massively refactored in 4.x and ad-ons specific to our own internal use cases.

Sent from my iPhone

To unsubscribe from this group and stop receiving emails from it, send an email to hue-user+u...@cloudera.org.

[Chriss Vincelette]

Had an individual get it today and recreating his session did not help. Had to restart hive. :(

-Chriss-

To unsubscribe from this group and stop receiving emails from it, send an email to hue-user+u...@cloudera.org.