[
https://issues.apache.org/jira/browse/HIVE-12020?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15025612#comment-15025612
]
Prasanth Jayachandran commented on HIVE-12020:
----------------------------------------------
Let me list down the changes
1) Datanucleus related logging have changed. Earlier man specific datanucleus
loggers were explicitly added. In this patch, all top level data nucleus
loggers are added (no need for specific loggers). Discussed with Sushanth about
it and he said the new change is good. If we want specific logger changes we
can do it later.
2) In log4j2.xml files, I have mistakenly added %x to pattern layout that will
be used by NDC. I don't think anything other than llap uses NDC so %x is added
only to llap properties.
3) Log4j version updated to 2.4.1 to workaround NPE with empty loggers
4) HiveEventCounter has been removed from root logger configuration. It was
added by default and I don't think it is of much significance. It publishes
count of msgs logged at different log levels to hadoop metrics. But I don't see
any configurations for hadoop-metrics in hive source. If required, this can
also be added back.
Other than these changes, it's pretty much the same one-to-one copy from
log4j2.xml.
> Revert log4j2 xml configuration to properties based configuration
> -----------------------------------------------------------------
>
> Key: HIVE-12020
> URL: https://issues.apache.org/jira/browse/HIVE-12020
> Project: Hive
> Issue Type: Sub-task
> Components: Logging
> Affects Versions: 2.0.0
> Reporter: Prasanth Jayachandran
> Assignee: Prasanth Jayachandran
> Attachments: HIVE-12020.1.patch, HIVE-12020.2.patch
>
>
> Log4j 2.4 release brought back properties based configuration. We should
> revert XML based configuration and use properties based configuration instead
> (less verbose and will be similar to old log4j properties).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
