[
https://issues.apache.org/jira/browse/FLINK-12921?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16873370#comment-16873370
]
Konstantin Knauf commented on FLINK-12921:
------------------------------------------
[~vim] Thanks for opening this ticket. The documentation is not correct here.
The behavior changed after FLIP-6.
I have linked a related ticket to enable the old behavior again.
> Flink Job Scheduling
> --------------------
>
> Key: FLINK-12921
> URL: https://issues.apache.org/jira/browse/FLINK-12921
> Project: Flink
> Issue Type: Improvement
> Components: API / DataSet, API / DataStream
> Affects Versions: 1.8.0
> Environment: flink 1.8.0
> jdk8
> Reporter: vim
> Priority: Major
>
> This is an example from flink-docs:
> Consider a program with a data source, a _MapFunction_, and a
> _ReduceFunction_. The source and MapFunction are executed with a parallelism
> of 4, while the ReduceFunction is executed with a parallelism of 3. A
> pipeline consists of the sequence Source - Map - Reduce. On a cluster with 2
> TaskManagers with 3 slots each, the program will be executed as described
> below.
> !https://ci.apache.org/projects/flink/flink-docs-release-1.8/fig/slots.svg!
> But after I tried, I found that it was not like this. My result is
> TaskManager 1 used 3 slot, but TaskManager 2 just used 1 slot. Who else has
> tested this example?
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
