[ 
https://issues.apache.org/jira/browse/AIRFLOW-3413?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Trent Telfer updated AIRFLOW-3413:
----------------------------------
    Description: Currently running airflow in a docker container on an Ubuntu 
VM. When using the KubernetesPodOperator I get a condition where a pod will 
have completed on the kubernetes cluster but airflow still thinks it is 
running. The only way to get the pod start following its execution pattern 
again is to manually mark the pod as completed in airflow. Other solution that 
didn't work was attempting to use the dagrun_timeout which had no affect.  
(was: Currently while running airflow in a docker container on an Ubuntu VM. 
When using the KubernetesPodOperator I get a condition where a pod will have 
completed on the kubernetes cluster but airflow still thinks it is running. The 
only way to get the pod start following its execution pattern again is to 
manually mark the pod as completed in airflow. Other solution that didn't work 
was attempting to use the dagrun_timeout which had no affect.)

> Airflow sometimes doesn't know when pod has completed when using 
> KubernetesPodOperator.
> ---------------------------------------------------------------------------------------
>
>                 Key: AIRFLOW-3413
>                 URL: https://issues.apache.org/jira/browse/AIRFLOW-3413
>             Project: Apache Airflow
>          Issue Type: Bug
>          Components: kubernetes
>    Affects Versions: 1.10.1
>         Environment: Azure VM - Ubuntu 18.04.1 LTS - Docker CE 18.09.0
>            Reporter: Trent Telfer
>            Priority: Major
>
> Currently running airflow in a docker container on an Ubuntu VM. When using 
> the KubernetesPodOperator I get a condition where a pod will have completed 
> on the kubernetes cluster but airflow still thinks it is running. The only 
> way to get the pod start following its execution pattern again is to manually 
> mark the pod as completed in airflow. Other solution that didn't work was 
> attempting to use the dagrun_timeout which had no affect.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to