[
https://issues.apache.org/jira/browse/AIRFLOW-1641?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16212960#comment-16212960
]
Nick Bauer commented on AIRFLOW-1641:
-------------------------------------
Not sure you to ping the author of this, but one thing that made our situation
WAY better is to increase the dag bag loading timeout. We were experiencing
timeouts when the task would start, and the first thing it does is load the dag
it belongs to. If that process timed out (ours was taking 6 minutes),
especially when the server was under heavy load
> Task gets stuck in queued state
> -------------------------------
>
> Key: AIRFLOW-1641
> URL: https://issues.apache.org/jira/browse/AIRFLOW-1641
> Project: Apache Airflow
> Issue Type: Bug
> Components: scheduler
> Affects Versions: 1.8.0
> Environment: Linux
> Reporter: Mas
> Labels: queued, scheduler, stuck, task
>
> Hello,
> I have one dag with ~20 tasks.
> The dags runs daily and some tasks can sometime last for hours, depending on
> the processed data behind.
> There are some interactions with AWS and a remote DB.
> I only use LocalExecutor.
> What this issue is about, is the fact that sometime (randomly, and without
> any clear reason) one of the tasks (here also, it is random) gets stuck in
> "queued" state and never starts running.
> The manual workaround is to restart the task manually by clearing it.
> Does anyone have ideas about the issue behind, and how to avoid it for the
> future?
> Thanks in advance for your help.
> PS: other people are facing the same behaviour:
> [link|https://stackoverflow.com/questions/45853013/airflow-tasks-get-stuck-at-queued-status-and-never-gets-running]
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
