[
https://issues.apache.org/jira/browse/FLINK-19703?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17219608#comment-17219608
]
Zhu Zhu commented on FLINK-19703:
---------------------------------
[~chesnay] It can happen in code path
{{JobMaster#updateTaskExecutionState(...)}} ->
{{SchedulerBase#updateTaskExecutionState(...)}} ->
{{executionGraph.updateState(...)}} -> {{Execution#markFailed(...)}} ->
{{Execution#processFail(...)}}, in the case that {{isLegacyScheduling ==
false}}.
> A result partition is not untracked after its producer task failed in
> TaskManager
> ---------------------------------------------------------------------------------
>
> Key: FLINK-19703
> URL: https://issues.apache.org/jira/browse/FLINK-19703
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Coordination
> Affects Versions: 1.10.2, 1.12.0, 1.11.2
> Reporter: Zhu Zhu
> Assignee: Zhu Zhu
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.12.0
>
>
> {{Execution#maybeReleasePartitionsAndSendCancelRpcCall(...)}} will be not
> invoked when a task is reported to be failed in TaskManager, which results in
> its partitions to still be tacked by the job manager partition tracker.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
