airflow-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Xinbin Huang (Jira)" <j...@apache.org>
Subject [jira] [Commented] (AIRFLOW-6225) Better Logging for the K8sPodOperator
Date Tue, 17 Dec 2019 17:26:00 GMT

    [ https://issues.apache.org/jira/browse/AIRFLOW-6225?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16998405#comment-16998405
] 

Xinbin Huang commented on AIRFLOW-6225:
---------------------------------------

Hi [~dimberman],

Sorry for the late reply. I would definitely like to try on this ticket, but I have limited
experience with Kubernetes, (i.e. understanding the big picture and concepts about K8S but
have not worked with it that much). Ii would be really helpful if you can point me to some
directions

> Better Logging for the K8sPodOperator
> -------------------------------------
>
>                 Key: AIRFLOW-6225
>                 URL: https://issues.apache.org/jira/browse/AIRFLOW-6225
>             Project: Apache Airflow
>          Issue Type: Improvement
>          Components: operators
>    Affects Versions: 1.10.6
>            Reporter: Daniel Imberman
>            Assignee: Daniel Imberman
>            Priority: Minor
>             Fix For: 1.10.7
>
>
> If a user uses the k8sPodOperator and a pod dies, there's valuable info in the {{kubectl
describe pod}} that is NOT being reported in either airflow or ES. We should determine if
there is a better way to track that information in airflow to bubble up to users who do not
have direct k8s access.
>  
> Possible additions:
>  * getting all events for the pod 
> kubectl get events --field-selector involvedObject.name=\{pod_name}]
>  * having a delete mode such as "only_on_success"
>  * Adding a prestop hook to propagate exception information in cases of failures



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Mime
View raw message