[
https://issues.apache.org/jira/browse/FLINK-1099?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ufuk Celebi closed FLINK-1099.
------------------------------
Resolution: Fixed
I think these are covered in the Local/NetworkBufferPool unit tests.
> Check that the GlobalBufferPool has all buffers available after a task
> crashed with an exception
> ------------------------------------------------------------------------------------------------
>
> Key: FLINK-1099
> URL: https://issues.apache.org/jira/browse/FLINK-1099
> Project: Flink
> Issue Type: Test
> Reporter: Till Rohrmann
> Assignee: Ufuk Celebi
> Priority: Minor
>
> [~StephanEwen] noticed that not all buffers are returned to the
> GlobalBufferPool in case of an exception in a job. He already fixed the
> problem and also introduced assertions in the JobManagerITCase to check this.
> Due to the incorporation of the actor model this information is no longer
> available in the context of an JobManager integration test. Furthermore, I
> think that the JobManagerITCase is not the right place to check for these low
> level functionalities. Therefore, I propose to write an explicit test at the
> level of the RuntimeEnvironment to check the correct behaviour of releasing
> allocated resources.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
