If the plan is hourly, can the server be shut down when not needed? For example, for the project I am thinking about, the server can be down at night (from 10 PM to 5 AM).
Right now, this is not supported.
It’s important to note that just the worker specs are billed hourly as of right now - there is still a base cost for keeping the scheduler, webserver, db, etc. up.
We’re also trying to encourage your further adoption of Airflow, are there some hourly DAGs that you could put out there, perhaps to power dashboards for your customers or management?
Quick update on this one since we transitioned to metered billing for Astronomer Cloud. Scheduled downtime is still not supported, but a bit more detail here:
Since the deployment exists on our infrastructure for that entire period, we charge per hour that the Airflow deployment exists. In an upcoming release, we’ll be incorporating the Kubernetes executor, which will allow for “scale-to-zero” functionality. We’re waiting for that feature to be fully stable in Airflow before we rope it into our platform, so for now we charge based on the exact resources allocated to your deployments.
The resource configurations for your deployment are now fully customizable, so feel free to jump into the Astronomer UI and head over to
Configure tab to adjust your executor (Celery or Local) and worker count to fit your needs.
For more details on pricing and our default set ups, check out our pricing page.