[
https://issues.apache.org/jira/browse/SPARK-14750?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15253237#comment-15253237
]
SuYan commented on SPARK-14750:
-------------------------------
<property>
<name>yarn.log-aggregation-enable</name>
<value>true</value>
</property> // if this =true, means application log in container will recycle
in aggregated hdfs folder.
<property>
<name>yarn.log-aggregation.retain-seconds</name>
<value>259200</value>
</property> // this means how long the application will retained in
aggregated hdfs folder.
<property>
<name>yarn.nodemanager.delete.debug-delay-sec</name>
<value>600</value>
</property> // this means, for some debug requirement, how long application
log retained in nodemanager folder even if the application logs had recycled in
aggregated folder.
<property>
<name>yarn.nodemanager.log.retain-seconds</name>
<value>86400</value>
</property> // this property is for log-aggregate = false. how long will the
log retained in the nodemanager folder.
eh...I found some improvements point for that PR, I need consider aggregate =
false, while the yarn.nodemanager.log.retain-seconds = some large value.
> Make historyServer refer application log in hdfs
> ------------------------------------------------
>
> Key: SPARK-14750
> URL: https://issues.apache.org/jira/browse/SPARK-14750
> Project: Spark
> Issue Type: Improvement
> Components: Web UI
> Affects Versions: 1.6.1
> Reporter: SuYan
>
> Make history server refer application log, just like MR history server
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
