[ https://issues.apache.org/jira/browse/FLINK-2628?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14733626#comment-14733626 ]
Stephan Ewen commented on FLINK-2628: ------------------------------------- This is a similar issue as we had in one other test. We cannot rely on the fact that a checkpoint occurred before the first failure on Travis. This is due to very different delays in the between the scheduling of different source tasks. The tests works well when executed locally. I am changing the {{fail()}} to a warning, when the error came before the first checkpoint. > Failing Test: StreamFaultToleranceTestBase.runCheckpointedProgram > ----------------------------------------------------------------- > > Key: FLINK-2628 > URL: https://issues.apache.org/jira/browse/FLINK-2628 > Project: Flink > Issue Type: Bug > Components: Tests > Reporter: Martin Liesenberg > Assignee: Stephan Ewen > > In pullrequest #1097 the test > StreamFaultToleranceTestBase.runCheckpointedProgram > The changes introduced in the pull request are most likely unrelated. I can > not reproduce it locally. -- This message was sent by Atlassian JIRA (v6.3.4#6332)