You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is there a way of allowing the webserver to resolve the workers through their names?
Currently, this stack puts workers into StatefulSets. I think the barrier to using a Deployment is that before log files are put in GCS, they fail to resolve:
*** Fetching from: http://airflow-worker-6dd9f5bc74-h97vf:8793/log/[***]
*** Failed to fetch log file from worker. HTTPConnectionPool(host='airflow-worker-6dd9f5bc74-h97vf', port=8793): Max retries exceeded with url: /log/[***] (Caused by NewConnectionError('<urllib3.connection.HTTPConnection object at 0x7f142598b210>: Failed to establish a new connection: [Errno -2] Name or service not known',))
If we could surmount that, we could use a Deployment, which means we can run on preemptive machines, and generally allow moving pods around
The text was updated successfully, but these errors were encountered:
Is there a way of allowing the webserver to resolve the workers through their names?
Currently, this stack puts workers into StatefulSets. I think the barrier to using a Deployment is that before log files are put in GCS, they fail to resolve:
If we could surmount that, we could use a Deployment, which means we can run on preemptive machines, and generally allow moving pods around
The text was updated successfully, but these errors were encountered: