[
https://issues.apache.org/jira/browse/FLINK-19123?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Aljoscha Krettek updated FLINK-19123:
-------------------------------------
Fix Version/s: 1.12.0
> TestStreamEnvironment does not use shared MiniCluster for executeAsync()
> ------------------------------------------------------------------------
>
> Key: FLINK-19123
> URL: https://issues.apache.org/jira/browse/FLINK-19123
> Project: Flink
> Issue Type: Bug
> Components: API / DataStream, Runtime / Coordination, Tests
> Reporter: Aljoscha Krettek
> Priority: Major
> Fix For: 1.12.0
>
>
> TestStreamEnvironment does override {{execute()}} but not {{executeAsync()}}
> . Now, {{execute()}} goes against the {{MiniCluster}} session that was
> started by a {{MiniClusterWithClientResource}} or some other method that uses
> {{TestStreamEnvironment}}. However, {{executeAsync()}} will go through the
> normal {{StreamExecutionEnvironment}} logic and tries to find an executor,
> does not know that it is a testing environment.
> Up until recently, you would have gotten an exception that tells you that no
> executor is configured, then we would have found out that we need to override
> {{executeAsync()}} in {{TestStreamEnvironment}}. However, we currently
> configure a local executor in the constructor:
> [https://github.com/apache/flink/blob/2160c3294ef87143ab9a4e8138cb618651499792/flink-test-utils-parent/flink-test-utils/src/main/java/org/apache/flink/streaming/util/TestStreamEnvironment.java#L59].
> With this, you basically get the “local environment” behaviour when you call
> {{executeAsync()}}, which starts a cluster for the job and shuts it down when
> the job finishes. This basically makes the {{TestStreamEnvironment}} cluster
> sharing useless.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
