[jira] [Commented] (HIVE-12254) Improve logging with yarn/hdfs

["Peter Vary (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HIVE-12254?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16634066#comment-16634066 ] 

Peter Vary commented on HIVE-12254:
-----------------------------------

[~aihuaxu]: 3 quick questions:
* What happens when multiple queries are running / compiling in parallel? They might be a mix up between the context for queries, especially if async compilation is turned on {{hive.server2.async.exec.async.compile}}. Or this is handled some way by the Shims themselves?
* In the previous version of the patch [~vikram.dixit] tried to reset the context to the original value after the query is finished the run. I think it was useful to reset it to the sessionId after every query is finished. Do we need this? Or there is no hdfs/yarn commands when we do not running a query?
* In {{Hadoop23Shims.setHadoopSessionContext}} there is ":" in the name, in {{Hadoop23Shims.setHadoopQueryContext}} it is "_". This might be a mistake.
Thanks,
Peter

> Improve logging with yarn/hdfs
> ------------------------------
>
>                 Key: HIVE-12254
>                 URL: https://issues.apache.org/jira/browse/HIVE-12254
>             Project: Hive
>          Issue Type: Bug
>          Components: Shims
>    Affects Versions: 1.2.1
>            Reporter: Vikram Dixit K
>            Assignee: Aihua Xu
>            Priority: Major
>         Attachments: HIVE-12254.1.patch, HIVE-12254.2.patch, HIVE-12254.3.patch
>
>
> In extension to HIVE-12249, adding info for Yarn/HDFS as well. Both HIVE-12249 and HDFS-9184 are required (and upgraded in hive for the HDFS issue) before this can be resolved.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)