[
https://issues.apache.org/jira/browse/BEAM-7670?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16877354#comment-16877354
]
Kyle Weaver edited comment on BEAM-7670 at 7/2/19 11:11 PM:
------------------------------------------------------------
Looks like there are at least two different causes for this same behavior.
First (more common) has one subtask hanging because it's never getting a
"worker ready" response [1]. Second has two subtasks hanging trying to access
the environment cache [2].
[1] [https://gist.github.com/ibzib/fd67c1fad1fc4d1475ae63a2560a86ab]
[2] [https://gist.github.com/ibzib/15f2d3ae9bdd476457d22bf16d54defc]
was (Author: ibzib):
Looks like there are at least two different causes for this same behavior.
First (more common) has one subtask hanging because it's never getting a
"worker ready" response [1]. Second has two subtasks hanging trying to access
the environment cache [2].
[[1]
https://gist.github.com/ibzib/fd67c1fad1fc4d1475ae63a2560a86ab|https://gist.github.com/ibzib/fd67c1fad1fc4d1475ae63a2560a86ab]
[2] [https://gist.github.com/ibzib/15f2d3ae9bdd476457d22bf16d54defc]
> Flink portable worker gets stuck if one of the task does not get any data
> -------------------------------------------------------------------------
>
> Key: BEAM-7670
> URL: https://issues.apache.org/jira/browse/BEAM-7670
> Project: Beam
> Issue Type: Bug
> Components: runner-flink
> Reporter: Ankur Goenka
> Priority: Major
>
> When using parallelism > 1 with flink portable runner, the job gets stuck if
> the data is partitioned in such a way that one of the task does not get any
> data.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
