To run your code on Astro, you need to deploy it to a Deployment. You can deploy part or all of an Astro project to an Astro Deployment.
There are several options for deploying code to a Deployment:
astro deploy to build every non-Dag file in your Astro project as a Docker image and deploy the image to all Airflow components in a Deployment. This includes your Dockerfile, plugins, and all Python and OS-level packages. Dags are deployed separately to each Airflow component through a sidecar container. See Deploy a project image.astro deploy --dags to deploy only your Dag files to Astro. If you only need to deploy Dag changes, running this command is faster than running astro deploy since it doesn’t require installing dependencies. See Deploy Dags.astro deploy --image to build and deploy your Astro project configurations as a Docker image without deploying your Dags. This is useful if you have a multi-repo CI/CD strategy, and you want to deploy your Dags and project configurations from different repositories or storage buckets. See Image-only deploys.astro deploy --dbt to deploy only a dbt project to Astro. See Deploy dbt projects to Astro. You must use this command to deploy dbt projects to Astro.astro deploy to build and deploy the Runtime (orchestration plane) image, and use astro remote deploy to build and push client images for your Remote Execution Agents. See Initialize and deploy Remote Execution projects for full instructions.For each deploy option, you can either trigger the deploy manually or through CI/CD. CI/CD pipelines can include both image deploys and Dag-only deploys, and they can deploy to multiple different Deployments based on different branches in your git repository. See CI/CD overview.
If multiple deploys are triggered simultaneously or additional deploys are triggered while a deploy is still processing, whether manually or through CI/CD, the first deploy is processed and then the subsequent deploys are completed. This behavior is different from how Astro processes simultaneous code deploys with the GitHub integration, which cancels the first deploy, and applies the most recent.