flink-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Till Rohrmann (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (FLINK-12214) Add JobListener to capture job submission process
Date Wed, 17 Apr 2019 13:50:00 GMT

    [ https://issues.apache.org/jira/browse/FLINK-12214?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16820102#comment-16820102
] 

Till Rohrmann commented on FLINK-12214:
---------------------------------------

I understand the problem you are trying to solve. But I'm not sure whether we should add a
{{JobStatusListener}} to every {{ExecutionEnvironment}}. {{ExecutionEnvironment}} are public
APIs and one should be careful with what one adds to it. Before doing such a change we need
consensus in the community that this is a good thing to do.

Instead of exposing such a feature directly via the {{ExecutionEnvironment}}, I could rather
see it work by making the {{RestClusterClient}} pluggable so that you can have an extended
version which offers the respective callbacks.

> Add JobListener to capture job submission process
> -------------------------------------------------
>
>                 Key: FLINK-12214
>                 URL: https://issues.apache.org/jira/browse/FLINK-12214
>             Project: Flink
>          Issue Type: New Feature
>          Components: Runtime / Coordination
>            Reporter: Jeff Zhang
>            Assignee: Jeff Zhang
>            Priority: Major
>              Labels: pull-request-available
>         Attachments: image-2019-04-17-00-11-00-020.png, image-2019-04-17-00-13-15-999.png,
image-2019-04-17-00-15-16-750.png
>
>          Time Spent: 20m
>  Remaining Estimate: 0h
>
> Third party library would like to add some hook when submitting jobs. So it would be
nice to have JobListener for flink.
> More details about the motivation:
> The background is that I am working on integrating flink with apache zeppelin notebook.
In apache zeppelin notebook, we allow user to run flink scala code or sql. For each piece
fo code, I'd like to associate it with its corresponding flink job. So that user can link
to flink job ui in zeppelin notebook, also he can cancel this job. Here's one screenshot of
my POC, in this screenshot you can see the FLINK JOB link which will link to the flink job
ui. That's why I'd like to add hook in flink's job submission, so that I can link each piece
of flink code to paragraph (the following screenshot) which is a concept of zeppelin and is
a container that host this piece of code. 
> Here's the sample code for reference about how I use JobListener in zeppelin.
> [https://github.com/zjffdu/zeppelin/blob/blink_poc/flink/src/main/scala/org/apache/zeppelin/flink/FlinkScalaInterpreter.scala#L251]
> !image-2019-04-17-00-15-16-750.png!
>  



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

Mime
View raw message