[ 
https://issues.apache.org/jira/browse/FLINK-9280?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16511204#comment-16511204
 ] 

ASF GitHub Bot commented on FLINK-9280:
---------------------------------------

Github user zentol commented on a diff in the pull request:

    https://github.com/apache/flink/pull/6147#discussion_r195101445
  
    --- Diff: 
flink-runtime/src/main/java/org/apache/flink/runtime/rest/RestClient.java ---
    @@ -197,6 +208,88 @@ public void shutdown(Time timeout) {
                                executor);
        }
     
    +   public <M extends MessageHeaders<R, P, U>, U extends MessageParameters, 
R extends RequestBody, P extends ResponseBody> CompletableFuture<P> sendRequest(
    +                   String targetAddress,
    +                   int targetPort,
    +                   M messageHeaders,
    +                   U messageParameters,
    +                   R request,
    +                   Collection<Path> jars,
    +                   Collection<Path> userArtifacts) throws IOException {
    --- End diff --
    
    So far we managed to not expose netty stuff in the `RestClient` API, I 
would prefer if we didn't start now.
    
    Could we not just pass a single `Collection<Path>` instead? The 
content-type would then always be `application/octet-stream`.


> Extend JobSubmitHandler to accept jar files
> -------------------------------------------
>
>                 Key: FLINK-9280
>                 URL: https://issues.apache.org/jira/browse/FLINK-9280
>             Project: Flink
>          Issue Type: New Feature
>          Components: Job-Submission, REST
>    Affects Versions: 1.5.0
>            Reporter: Chesnay Schepler
>            Assignee: Chesnay Schepler
>            Priority: Critical
>             Fix For: 1.6.0, 1.5.1
>
>
> The job submission through the CLI first uploads all require jars to the blob 
> server, sets the blob keys in the jobgraph, and then uploads this graph to 
> The {{JobSubmitHandler}} which submits it to the Dispatcher.
> This process has the downside that it requires jars to be uploaded to the 
> blobserver before submitting the job graph, which does not happen via REST.
> I propose an extension to the the {{JobSubmitHandler}} to also accept an 
> optional list of jar files, that were previously uploaded through the 
> {{JarUploadHandler}}. If present, the handler would upload these jars to the 
> blobserver and set the blob keys.



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

Reply via email to