[jira] [Comment Edited] (HAWQ-762) Hive aggregation queries through PXF sometimes hang

["Michael Andre Pearce (IG) (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HAWQ-762?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15325778#comment-15325778 ] 

Michael Andre Pearce (IG) edited comment on HAWQ-762 at 6/11/16 8:43 AM:
-------------------------------------------------------------------------

The environment that we experienced it in whilst Goden was on webex and the logs we sent was one of our development area ones where we are trying to setup hawq and test it:

Cluster topology:
3 master nodes
5 data node

each data node is 256GB ram.
Hawq yarn queue is allocated 25% resource with a max of 50%.
heaps were all defaults as per install doc here http://hdb.docs.pivotal.io/20/requirements/system-requirements.html#topic_uzf_flz_g5 so allocated 1G, we used Ambari to install and configure.

The table we used during the demo of how we were looking to use tableau with hawq during the demo was some airports data as we couldn't use our private data sets, its relatively small but good for feature testing (not performance).

you can get it here http://openflights.org/data.html

If we get the issue again any other logs or details you'd like us to try capture and send over privately?





was (Author: michael.andre.pearce):
The environment that we experienced it in whilst Goden was on webex and the logs we sent was one of our development area ones where we are trying to setup hawq and test it:

Cluster topology:
3 master nodes
5 data node

each data node is 256GB ram.
Hawq yarn queue is allocated 25% resource with a max of 50%.
heaps were all defaults as per install doc here http://hdb.docs.pivotal.io/20/requirements/system-requirements.html#topic_uzf_flz_g5 so allocated 1GB

The table we used during the demo of how we were looking to use tableau with hawq during the demo was some airports data as we couldn't use our private data sets, its relatively small but good for feature testing (not performance).

you can get it here http://openflights.org/data.html

If we get the issue again any other logs or details you'd like us to try capture and send over privately?




> Hive aggregation queries through PXF sometimes hang
> ---------------------------------------------------
>
>                 Key: HAWQ-762
>                 URL: https://issues.apache.org/jira/browse/HAWQ-762
>             Project: Apache HAWQ
>          Issue Type: Bug
>          Components: Hcatalog, PXF
>            Reporter: Oleksandr Diachenko
>            Assignee: Goden Yao
>              Labels: performance
>
> Reproduce Steps:
> {code}
> select count(*) from hcatalog.default.hivetable;
> {code}
> sometimes, this query will hang and we see from pxf logs that Hive thrift server cannot be connected from PXF agent. 
> While users can still visit hive metastore (through HUE) and execute the same query.
> After a restart of PXF agent, this query goes through without issues.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)