[jira] [Commented] (FLINK-15345) CurrentExecution and priorExecutions are inconsistent in the test case

["Chesnay Schepler (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/FLINK-15345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17047433#comment-17047433 ] 

Chesnay Schepler commented on FLINK-15345:
------------------------------------------

You haven't answered the question wheter there is any code that relies on the prior execution still existing in the ExecutionGraph. I don't think that is the case, and think this is fine as it allows us to purge older executions to free up some space.

When we think long-term a long-running job with some failures, re-scaling and such can easily have hundreds of execution attempts, and there should be no requirement to keep all of them around.

> CurrentExecution and priorExecutions are inconsistent in the test case
> ----------------------------------------------------------------------
>
>                 Key: FLINK-15345
>                 URL: https://issues.apache.org/jira/browse/FLINK-15345
>             Project: Flink
>          Issue Type: Improvement
>          Components: Runtime / REST, Tests
>            Reporter: lining
>            Priority: Minor
>
> When the length of priorExecutions does not exceed the value of the configuration item jobmanager.execution.attempts-history-size, the attempt should be equal to the length of priorExecutions. So [code in JobExceptionsHandlerTest|https://github.com/apache/flink/blob/master/flink-runtime/src/test/java/org/apache/flink/runtime/rest/handler/job/JobExceptionsHandlerTest.java#L104-L121] and [code in SubtaskCurrentAttemptDetailsHandlerTest|https://github.com/apache/flink/blob/master/flink-runtime/src/test/java/org/apache/flink/runtime/rest/handler/job/SubtaskCurrentAttemptDetailsHandlerTest.java#L94-L111] are wrong, we need to update the attempt to 0.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)