You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@airflow.apache.org by "Grant McKenzie (Jira)" <ji...@apache.org> on 2020/04/04 05:18:00 UTC
[jira] [Resolved] (AIRFLOW-6893) Per dag worker image selection
[ https://issues.apache.org/jira/browse/AIRFLOW-6893?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Grant McKenzie resolved AIRFLOW-6893.
-------------------------------------
Resolution: Invalid
This functionality is already available by passing in executor_config to each Dag task as described here:
[https://marclamberti.com/blog/airflow-kubernetes-executor/]
> Per dag worker image selection
> ------------------------------
>
> Key: AIRFLOW-6893
> URL: https://issues.apache.org/jira/browse/AIRFLOW-6893
> Project: Apache Airflow
> Issue Type: New Feature
> Components: executor-kubernetes
> Affects Versions: 1.10.9
> Reporter: Grant McKenzie
> Assignee: Daniel Imberman
> Priority: Minor
>
> Hi,
> one of the challenges of a multi-tenant deployment of Airflow is that different teams might require different versions of common python libraries - pandas, numpy for example.
> Configuration of the worker image in the KubernetesExecutor is static via airflow.cfg.
> Has any consideration been given to allowing specification of a per-Dag worker image that would allow different teams on a shared Airflow instance to customize the set of dependencies available to them?
> Thanks.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)