[jira] [Commented] (YARN-8356) yarn.log-aggregation why can auto clean

["Jason Lowe (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/YARN-8356?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16488922#comment-16488922 ] 

Jason Lowe commented on YARN-8356:
----------------------------------

Couple of questions to clarify the setup:

Is yarn.nodemanager.remote-app-log-dir configured to /app-logs/hive/logs-ifile?  That's not the default location for aggregated logs, so I want to verify this property was overridden to this path and we really are talking about YARN's aggregated logs vs. some other type of logging.

Are you running the MapReduce job history server?  That server hosts the aggregated log deletion service which is responsible for deleting old logs.  If that server is running, you can check its logs for evidence of the aggregated log deletion service running.  Grep for "aggregated log deletion started" and "Deleting aggregated logs".


> yarn.log-aggregation   why can auto clean 
> ------------------------------------------
>
>                 Key: YARN-8356
>                 URL: https://issues.apache.org/jira/browse/YARN-8356
>             Project: Hadoop YARN
>          Issue Type: Wish
>          Components: log-aggregation
>    Affects Versions: 2.7.5
>            Reporter: ma_lie
>            Priority: Major
>         Attachments: image-2018-05-24-17-56-04-555.png
>
>
> when I enable yarn.log aggregation  and define 
> yarn.log-aggregation.retain-seconds=3600
>  
> but after 3600s /app-logs/hive/logs-ifile  can't decrease and many many file exists for a long time than 3600s before now;
> so how can I do so that the yarn.log can auto clean itself



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

---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org