Specifying operator dependencies

Now that we have set up the scheduling using a cron expression, we know when the first of our tasks in a DAG should run, but we haven’t mentioned anything about how tasks are related. What we usually run is a number of tasks, and these tasks are modular pieces that fit together in order to complete a full DAG. We need to introduce some logic between our tasks, which we call dependencies, to tell Airflow when each task should start. That is done with the methods .set_upstream() and .set_downstream, which are defined on any Airflow operator.

In this exercise, we have already defined a number of operators for you. Their variable identifiers are the same as the task_ids shown in the figure below. Let’s try to recreate the graph.

Airflow DAG illustrating the sequence of tasks for preparing a pizza

Set prepare_crust to precede apply_tomato_sauce using the appropriate method.
Set apply_tomato_sauceto precede each of tasks in tasks_with_tomato_sauce_parent using the appropriate method.
Set the tasks_with_tomato_sauce_parent list to precede bake_pizza using either the bitshift operator >> or <<.
Set bake_pizza to succeed prepare_oven using the appropriate method.

Exercise

Specifying operator dependencies

Instructions