[jira] [Commented] (FLINK-9301) NotSoMiniClusterIterations job fails on travis

["Stephan Ewen (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/FLINK-9301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16466224#comment-16466224 ] 

Stephan Ewen commented on FLINK-9301:
-------------------------------------

Okay, so apparently the required memory per TaskManager is now a bit higher than it was.
Can we fix that by simply reducing the number of TaskManagers?

The test was initially a "manual test", so not executed automatically as part of a build, and it was intended to observe and debug execution of a reasonably complex job locally, to assess if deployment and ramp up happens fast.

> NotSoMiniClusterIterations job fails on travis
> ----------------------------------------------
>
>                 Key: FLINK-9301
>                 URL: https://issues.apache.org/jira/browse/FLINK-9301
>             Project: Flink
>          Issue Type: Improvement
>          Components: Tests
>    Affects Versions: 1.5.0
>            Reporter: Chesnay Schepler
>            Priority: Critical
>             Fix For: 1.5.0
>
>
> The high-parallelism-iterations-test fails on travis. After starting ~55 taskmanagers all memory is used and further memory allocations fail.
> I'm currently letting it run another time, if it fails again I will disable the test temporarily.
> https://travis-ci.org/zentol/flink-ci/builds/375189790



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)