[jira] [Commented] (FLINK-17827) scala-shell.sh should fail early if no mode is specified, or have default logging settings

[Thomas Thornton (Jira)]

[ 
https://issues.apache.org/jira/browse/FLINK-17827?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17235100#comment-17235100
 ] 

Thomas Thornton commented on FLINK-17827:
-

Hi, [~chesnay], I want to gain consensus on the implementation. I suggest we 
add an else condition 
[here|https://github.com/apache/flink/blob/master/flink-scala-shell/start-script/start-scala-shell.sh#L93]
 that will exit with an error that the mode is not set properly and should be 
one of local, remote, or yarn. This will maintain consistency with both the 
[Scala 
requirements|https://github.com/apache/flink/blob/master/flink-scala-shell/src/main/scala/org/apache/flink/api/scala/FlinkShell.scala#L206-L209]
 and maintain consistency with the 
[documentation|https://ci.apache.org/projects/flink/flink-docs-stable/ops/scala_shell.html#full-reference].
 If this sounds reasonable to you, then feel free to assign the ticket to me.

> scala-shell.sh should fail early if no mode is specified, or have default 
> logging settings
> --
>
> Key: FLINK-17827
> URL: https://issues.apache.org/jira/browse/FLINK-17827
> Project: Flink
>  Issue Type: Improvement
>  Components: Scala Shell
>Affects Versions: 1.11.0
>Reporter: Chesnay Schepler
>Priority: Critical
>  Labels: starter
>
> The scala-shell has multiple modes it can run in: local, remote and yarn.
> It is mandatory to specify such a mode, but this is only enforced on the 
> scala side, not in the bash script.
> The problem is that the scala-shell script derives the log4j properties from 
> the mode, and if no mode is set, then the log4j properties are empty.
> This leads to a warning from slf4j that no logger was defined and all that.
> Either scala-shell.sh should fail early if no mode is specified, or it should 
> have some default logging settings (e.g., the ones for local/remote).



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

[jira] [Commented] (FLINK-27074) python_job.py should write to TEST_DATA_DIR

[Thomas Thornton (Jira)]

[ 
https://issues.apache.org/jira/browse/FLINK-27074?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17536756#comment-17536756
 ] 

Thomas Thornton commented on FLINK-27074:
-

[~chesnay] pull request is available for review 
[here|https://github.com/apache/flink/pull/19717].

> python_job.py should write to TEST_DATA_DIR
> ---
>
> Key: FLINK-27074
> URL: https://issues.apache.org/jira/browse/FLINK-27074
> Project: Flink
>  Issue Type: Technical Debt
>  Components: API / Python, Tests
>Reporter: Chesnay Schepler
>Priority: Critical
>  Labels: pull-request-available
> Fix For: 1.16.0
>
>
> Similar to other tests this job should write to TEST_DATA_DIR to make sure we 
> clean up the file after the test finished.
> Currently this job is used twice in some e2e tests, where it can then run 
> into FileAlreadyExistsExceptions, so we should additionally make sure that 
> each job execution uses a different directory.



--
This message was sent by Atlassian Jira
(v8.20.7#820007)