[
https://issues.apache.org/jira/browse/YARN-10547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17510442#comment-17510442
]
András Győri commented on YARN-10547:
-------------------------------------
Only had a few comments on this one, I will commit it if you fix them:
* SLSRunner#L488: Checked exception is not thrown
* SLSRunner#L464: Raw use of Map
* AMDefinitionSLS#L41: Unchecked conversion, this could easily fail. In case
of conversion exception, I would default it to an empty list, because it is
already checked in the next line properly
> Decouple job parsing logic from SLSRunner
> -----------------------------------------
>
> Key: YARN-10547
> URL: https://issues.apache.org/jira/browse/YARN-10547
> Project: Hadoop YARN
> Issue Type: Improvement
> Reporter: Szilard Nemeth
> Assignee: Szilard Nemeth
> Priority: Minor
> Attachments: YARN-10547.001.patch, YARN-10547.002.patch,
> YARN-10547.003.patch, YARN-10547.004.patch, YARN-10547.005.patch
>
>
> SLSRunner has too many responsibilities.
> One of them is to parse the job details from the SLS input formats and launch
> the AMs and task containers.
> As a first step, the job parser logic could be decoupled from this class.
> There are 3 types of inputs:
> - SLS trace
> - Synth
> - Rumen
> Their job parsing method are:
> - SLS trace:
> https://github.com/apache/hadoop/blob/005b854f6bad66defafae0abf95dabc6c36ca8b1/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/SLSRunner.java#L479-L526
> - Synth:
> https://github.com/apache/hadoop/blob/005b854f6bad66defafae0abf95dabc6c36ca8b1/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/SLSRunner.java#L722-L790
> - Rumen:
> https://github.com/apache/hadoop/blob/005b854f6bad66defafae0abf95dabc6c36ca8b1/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/SLSRunner.java#L651-L716
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]