[
https://issues.apache.org/jira/browse/AIRFLOW-4515?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16903934#comment-16903934
]
Steven Locke commented on AIRFLOW-4515:
---------------------------------------
Also, in what world would this not be a bug if every case of using a sudag
operator defaults to an incorrect implementation that just happens to not fail
even though its incorrect?
> KubernetesExecutor hangs in the subdag context
> ----------------------------------------------
>
> Key: AIRFLOW-4515
> URL: https://issues.apache.org/jira/browse/AIRFLOW-4515
> Project: Apache Airflow
> Issue Type: Bug
> Components: executors
> Affects Versions: 1.10.3
> Reporter: Steven Locke
> Assignee: Daniel Imberman
> Priority: Major
> Labels: kubernetes
>
> The bug is that when using the kubernetes executor in a k8s cluster and
> running a subdag with `executor=KubernetesExecutor()` the subdag never
> returns. Just runs infinitely. Here is an example repo
> https://github.com/slocke716/test_subdags
> This is occurring on 1.10.3 1.10.3post1 and 1.10.3b2 at least
> Not passing `executor=KubernetesExecutor()` into the subdag operator is not
> an option since it defaults to the SequentialExecutor and changing to the
> LocalExecutor causes thread errors because we require quite a bit of
> concurrency
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
