This diff consolidates shared env vars in one place to avoid drift between dagit, celery, etc.
Also, by converting the run master configs to environment variables, it is easier to keep configuration in sync across the run master and step execution pods. Pipelines can now be executed with the config:
execution: celery-k8s: config: backend: env: DAGSTER_K8S_CELERY_BACKEND broker: env: DAGSTER_K8S_CELERY_BACKEND job_image: env: DAGSTER_K8S_PIPELINE_RUN_IMAGE job_namespace: env: DAGSTER_K8S_PIPELINE_RUN_NAMESPACE instance_config_map: env: DAGSTER_K8S_INSTANCE_CONFIG_MAP image_pull_policy: "Always" postgres_password_secret: env: DAGSTER_K8S_PG_PASSWORD_SECRET env_config_maps: - env: DAGSTER_K8S_PIPELINE_RUN_ENV_CONFIGMAP
And this will pick up the appropriate step execution configs from the container environment.