GuidesAPI ReferenceDiscussions

Generate Airflow DAGs

You can use this mode when you don't want to manage your DAGs via DAG python files, but via configuration or external service like Databand.

DAGs provider from YAML file

dbnd-airflow can load DAGs for you from YAML file. DBND function get_dags_from_file will generate DAGs based on your YAML definition.

  1. You need to define the location of the file

[scheduler] Configuration Section Parameter Reference

  • config_file - Set a path to the file defining scheduled jobs to execute.
  • never_file_sync - Disable syncing the scheduler config_file to the database.
  • always_file_sync - Enable forcibly syncing the scheduler config_file to the database.
  • no_ui_cli_edit - Disable creating, editing, and deleting scheduled jobs from the CLI and UI. Scheduled job definitions will only be taken from the scheduler config file.
  • refresh_interval - Set the interval to refresh the scheduled job list from the db and/or a config file
  • active_by_default - Determine whether new scheduled jobs will be activated by default.
  • default_retries - Set the number of times to retry a failed run, unless set to a different value on the scheduled job
  • shell_cmd - If shell_cmd is True, the specified command will be executed through the shell. This can be useful if you are using Python primarily for the enhanced control flow it offers over most system shells and still want convenient access to other shell features such as shell pipes, filename wildcards, environment variable expansion, and expansion of ~ to a user's home directory.
  1. You need to add the following code to your DAGs. Folder
from dbnd_airflow.scheduler.dags_provider_from_file import get_dags_from_file

# airflow will only scan files containing the text DAG or airflow. This comment performs this function
dags = get_dags_from_file()
if dags:

DAGs definition

Every element in your yaml file will define a DAG with only one operator which will execute the command line of your choice. You need to define the following fields that will be mapped into DAG and Task definition.

DAG level params

  • name will be used as dag_id.
  • schedule_interval, catchup will be used at DAG definition.
  • start_date, end_date, depends_on_past, and owner will be used as DAG default parameters.

Task level params

  • cmd will be executed via BashOperator.
  • retries will be used as retries for the operator.

For example:

- name: dbnd_sanity_check check!
  cmd: dbnd run dbnd_sanity_check --set task_target_date={{tomorrow_ds}} --task-version now
  schedule_interval: "1 3-19 * * *"
  start_date: 2021-02-15T00:00:00
  catchup: false
  active: true

DAGs provider via Databand Service API

Every job defined at Databand Service will be transformed to DAG in a similar way as it's done at get_dags_from_file. You can manage Jobs via CLI.

Make sure you are connected to Databand Service Connecting DBND to Databand (Access Tokens)

  1. Use this code to define Apache Airflow DAGs:
from dbnd_airflow.scheduler.dags_provider_from_databand import get_dags_from_databand

# airflow will only scan files containing the text DAG or airflow. This comment performs this function
dags = get_dags_from_databand()
if dags:
  1. Manage your jobs via the dbnd schedule command.

job Create or edit scheduled job
list List all scheduled jobs defined at Databand Service
pause Pause existing scheduled job
enable Enable scheduled job
undelete Un-Delete deleted scheduled job
delete Delete existing scheduled job

You can find parameters for every job with the --help flag. For example dbnd schedule job --help

If you have a lot of scheduled jobs, it is possible to split DAGs into multiple files, to speed up Airflow execution.

Use the following command:

python -m dbnd_airflow.scheduler.utils.generate_partitioned_dags --partitions=3 --dag-folder=dags --base-name=dbnd_dags_from_databand

--partitions - number of files to split scheduled jobs into
--dag-folder - $AIRFLOW_HOME/dags folder
--base-name - file name prefix (,,
--template (optional) - custom file template to use instead of default
--dag-folder and --template can be a relative or absolute path