[
https://issues.apache.org/jira/browse/FLINK-33534?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17790429#comment-17790429
]
Yunfeng Zhou commented on FLINK-33534:
--------------------------------------
Hi [~gyfora] [~mxm] [~fanrui] , I'm interested in this ticket and am willing to
fix this problem. Could you please assign it to me?
I would also like to learn if there are existing programs that could be used to
reproduce this problem, which could help me know your concerns and needs in
detail.
Besides, the Flink CLI's document claims that we can specify configurations via
`-D`
arguments([link|https://nightlies.apache.org/flink/flink-docs-stable/docs/deployment/cli/#submitting-a-job]).
For example,
{code:java}
$ ./bin/flink run \
--detached \
--Dpipeline.max-parallelism=120 \
./examples/streaming/StateMachineExample.jar{code}
Could this method be a workaround to this problem?
> PipelineOptions.PARALLELISM_OVERRIDES is not picked up from jar submission
> request
> ----------------------------------------------------------------------------------
>
> Key: FLINK-33534
> URL: https://issues.apache.org/jira/browse/FLINK-33534
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Configuration, Runtime / REST
> Affects Versions: 1.18.0, 1.17.1
> Reporter: Gyula Fora
> Priority: Major
>
> PARALLELISM_OVERRIDES are currently only applied when they are part of the
> JobManager / Cluster configuration.
> When this config is provided as part of the JarRunRequestBody it is
> completely ignored and does not take effect.
> The main reason is that the dispatcher reads this value from it's own
> configuration object and does not include the extra configs passed through
> the rest request.
> This is a blocker for supporting the autoscaler properly for FlinkSessionJobs
> in the autoscaler
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
