[jira] [Updated] (FLINK-11533) Retrieve job class name from JAR manifest in ClassPathJobGraphRetriever

["Ufuk Celebi (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/FLINK-11533?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Ufuk Celebi updated FLINK-11533:
--------------------------------
    Fix Version/s: 1.9.0

> Retrieve job class name from JAR manifest in ClassPathJobGraphRetriever
> -----------------------------------------------------------------------
>
>                 Key: FLINK-11533
>                 URL: https://issues.apache.org/jira/browse/FLINK-11533
>             Project: Flink
>          Issue Type: New Feature
>          Components: Runtime / Coordination
>    Affects Versions: 1.7.0
>            Reporter: Ufuk Celebi
>            Assignee: Ufuk Celebi
>            Priority: Minor
>              Labels: pull-request-available
>             Fix For: 1.9.0
>
>          Time Spent: 20m
>  Remaining Estimate: 0h
>
> Users running job clusters distribute their user code as part of the shared classpath of all cluster components. We currently require users running {{StandaloneClusterEntryPoint}} to manually specify the job class name. JAR manifest entries that specify the main class of a JAR are ignored since they are simply part of the classpath.
> I propose to add another optional command line argument to the {{StandaloneClusterEntryPoint}} that specifies the location of a JAR file (such as {{lib/usercode.jar}}) and whose Manifest is respected.
> Arguments:
> {code}
> --job-jar <path-to-jar>
> --job-classname name
> {code}
> Each argument is optional, but at least one of the two is required. The job-classname has precedence over job-jar.
> Implementation wise we should be able to simply create the PackagedProgram from the jar file path in ClassPathJobGraphRetriever.
> If there is agreement to have this feature, I would provide the implementation.



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