Xuan Gong commented on YARN-7891:

Thanks for the review.

Previous code does not work. It does not handle the har file correctly. So, I 
have to make the change to make it work.

In general, the HAR file should not exist with the ifiles(tfiles). Because, 
when we call HAR CLI, it would check whether the application finishes and what 
is the log aggregation status. But we do have the corner case. The only corner 
case that I can think is that the application finishes while the log 
aggregation is still processing, but at this time, the RM failover/restart 
happens. In this case, we would lose the log aggregation status, and some of 
the logs have been aggregated. At this time, when we call HAR command, it is 
possible that we can have har file as well as ifiles(tfiles).


But, this issue can be solved by 


bq.  In {{readAggregatedLogs}}/{{readAggregatedLogsMeta}}, following logics are 
mostly identical, could you merge common logics if possible?


We can. But the temporary variable we are created here will be used later. In 
order to re-use them, we have to make lots of re-factory (maybe, create a 
common object includes all the variables) which i think, is too much work for 
this jira. I would like to solve it later.

> LogAggregationIndexedFileController should support HAR file
> -----------------------------------------------------------
>                 Key: YARN-7891
>                 URL: https://issues.apache.org/jira/browse/YARN-7891
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>            Reporter: Xuan Gong
>            Assignee: Xuan Gong
>            Priority: Major
>         Attachments: YARN-7891.1.patch

This message was sent by Atlassian JIRA

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

Reply via email to