Task groups for visual organization in Apache Airflow - Mini Project: Build & Apply

Task groups for visual organization

📖 Scenario: You are managing a data pipeline in Apache Airflow. Your pipeline has multiple related tasks that you want to group visually to keep the workflow clear and organized.

🎯 Goal: Build an Airflow DAG that uses TaskGroup to visually group related tasks together.

📋 What You'll Learn

Create a DAG named example_task_group_dag with default arguments

Create a TaskGroup named processing_tasks

Inside the processing_tasks group, create three DummyOperator tasks named extract, transform, and load

Set the task dependencies so that extract runs before transform, and transform runs before load

Print the DAG structure at the end

💡 Why This Matters

🌍 Real World

TaskGroups help organize complex workflows visually in Airflow UI, making pipelines easier to understand and maintain.

💼 Career

Understanding TaskGroups is important for data engineers and DevOps professionals managing scalable and readable data pipelines.

1

Create the DAG and import required modules

Import DAG, DummyOperator, and TaskGroup from airflow and airflow.operators.dummy. Create a DAG named example_task_group_dag with start_date set to datetime(2024, 1, 1) and schedule_interval set to None.

Apache Airflow

# Import DAG, DummyOperator, TaskGroup
# Create DAG named example_task_group_dag with start_date=datetime(2024, 1, 1) and schedule_interval=None
# Your code here

Hint

Use from airflow import DAG and from airflow.operators.dummy import DummyOperator. Use datetime from the datetime module for the start date.

2

Create a TaskGroup named processing_tasks

Inside the DAG context, create a TaskGroup named processing_tasks.

Apache Airflow

from airflow import DAG
from airflow.operators.dummy import DummyOperator
from airflow.utils.task_group import TaskGroup
from datetime import datetime

dag = DAG(
    'example_task_group_dag',
    start_date=datetime(2024, 1, 1),
    schedule_interval=None
)

# Create a TaskGroup named processing_tasks
# Your code here

Hint

Create the task group by calling TaskGroup('processing_tasks', dag=dag).

3

Create tasks inside the TaskGroup and set dependencies

Inside the processing_tasks group, create three DummyOperator tasks named extract, transform, and load. Set the dependencies so that extract runs before transform, and transform runs before load.

Apache Airflow

from airflow import DAG
from airflow.operators.dummy import DummyOperator
from airflow.utils.task_group import TaskGroup
from datetime import datetime

dag = DAG(
    'example_task_group_dag',
    start_date=datetime(2024, 1, 1),
    schedule_interval=None
)

processing_tasks = TaskGroup('processing_tasks', dag=dag)

# Create DummyOperator tasks extract, transform, load inside processing_tasks
# Set extract >> transform >> load dependencies
# Your code here

Hint

Create each task with DummyOperator and assign task_group=processing_tasks. Use >> to set dependencies.

4

Print the DAG structure

Print the DAG structure by calling print(dag.task_dict).

Apache Airflow

from airflow import DAG
from airflow.operators.dummy import DummyOperator
from airflow.utils.task_group import TaskGroup
from datetime import datetime

dag = DAG(
    'example_task_group_dag',
    start_date=datetime(2024, 1, 1),
    schedule_interval=None
)

processing_tasks = TaskGroup('processing_tasks', dag=dag)

extract = DummyOperator(task_id='extract', task_group=processing_tasks)
transform = DummyOperator(task_id='transform', task_group=processing_tasks)
load = DummyOperator(task_id='load', task_group=processing_tasks)

extract >> transform >> load

# Print the DAG task dictionary
# Your code here

Hint

Use print(dag.task_dict) to see the tasks including those inside the task group.

Practice

1. What is the main purpose of using TaskGroup in Airflow DAGs?

easy

A. To visually group related tasks for better organization

B. To execute tasks in parallel automatically

C. To replace the need for defining individual tasks

D. To schedule DAG runs at specific times

Solution

1. Step 1: Understand TaskGroup purpose

   TaskGroup is designed to group related tasks visually in the Airflow UI.

2. Step 2: Differentiate from other features

   It does not change execution behavior or scheduling but improves readability.

3. Final Answer:

   To visually group related tasks for better organization -> Option A

4. Quick Check:

   TaskGroup = Visual grouping [OK]

Hint: TaskGroup groups tasks visually, not for execution control [OK]

Common Mistakes:

• Thinking TaskGroup runs tasks in parallel
• Confusing TaskGroup with scheduling
• Assuming TaskGroup replaces tasks

2. Which of the following is the correct syntax to create a task group named 'processing' in Airflow?

easy

A. with TaskGroup('processing') as tg:

B. TaskGroup('processing'):

C. group = TaskGroup('processing')

D. with task_group('processing') as tg:

Solution

1. Step 1: Recall TaskGroup syntax

   The correct way to create a task group is using a context manager: with TaskGroup('name') as variable:

2. Step 2: Check option correctness

   with TaskGroup('processing') as tg: uses correct capitalization and syntax. TaskGroup('processing'): misses 'with'. group = TaskGroup('processing') misses 'with' and context manager. with task_group('processing') as tg: uses wrong function name casing.

3. Final Answer:

   with TaskGroup('processing') as tg: -> Option A

4. Quick Check:

   Use 'with TaskGroup(...) as ...:' [OK]

Hint: Use 'with TaskGroup("name") as var:' to create groups [OK]

Common Mistakes:

• Omitting 'with' keyword
• Using wrong capitalization like 'task_group'
• Not using 'as' to assign the group

3. Given the code snippet:

with TaskGroup('group1') as tg1:
    task_a = DummyOperator(task_id='a')
    task_b = DummyOperator(task_id='b')
task_c = DummyOperator(task_id='c')
task_a >> task_c

What will be the visual relationship between tasks in the Airflow UI?

medium

A. Tasks 'a', 'b', and 'c' are all grouped together

B. Tasks 'a' and 'b' are grouped; 'a' runs before 'c'; 'b' is independent

C. Task 'c' runs before 'a' and 'b' inside the group

D. Tasks 'a' and 'b' run after 'c'

Solution

1. Step 1: Analyze TaskGroup and tasks

   Tasks 'a' and 'b' are inside TaskGroup 'group1', so they appear grouped visually. Task 'c' is outside the group.

2. Step 2: Check dependencies

   Only 'task_a' has a downstream dependency to 'task_c', so 'a' runs before 'c'. 'b' has no dependencies and is independent.

3. Final Answer:

   Tasks 'a' and 'b' are grouped; 'a' runs before 'c'; 'b' is independent -> Option B

4. Quick Check:

   Group contains a,b; a >> c; b independent [OK]

Hint: Only tasks inside TaskGroup are grouped visually [OK]

Common Mistakes:

• Assuming all tasks are grouped
• Thinking 'b' depends on 'c'
• Confusing execution order

4. Identify the error in this Airflow DAG snippet using TaskGroup:

with TaskGroup('group1'):
    task1 = DummyOperator(task_id='task1')
task2 = DummyOperator(task_id='task2')
tg >> task2

medium

A. DummyOperator cannot be used inside TaskGroup

B. TaskGroup name must be uppercase

C. Task dependencies cannot cross TaskGroup boundaries

D. Missing 'as' keyword to assign TaskGroup to a variable

Solution

1. Step 1: Check TaskGroup usage

   The 'with TaskGroup('group1'):' statement should assign the group to a variable using 'as', e.g. 'with TaskGroup('group1') as tg:'

2. Step 2: Validate other options

   TaskGroup names can be lowercase. DummyOperator is allowed inside TaskGroup. Dependencies can cross groups.

3. Final Answer:

   Missing 'as' keyword to assign TaskGroup to a variable -> Option D

4. Quick Check:

   Use 'with TaskGroup(...) as var:' [OK]

Hint: Always use 'as' to assign TaskGroup in 'with' statement [OK]

Common Mistakes:

• Omitting 'as' in 'with TaskGroup(...) as ...:'
• Thinking TaskGroup names must be uppercase
• Believing DummyOperator is disallowed inside groups

5. You want to create a DAG with two task groups: 'extract' and 'transform'. Tasks in 'extract' must run before any in 'transform'. Which code snippet correctly sets this up?

hard

A. with TaskGroup('extract') as extract:
t1 = DummyOperator(task_id='t1')
t1b = DummyOperator(task_id='t1b')
with TaskGroup('transform') as transform:
t2 = DummyOperator(task_id='t2')
t2b = DummyOperator(task_id='t2b')
t1 >> t2

B. extract = TaskGroup('extract')
transform = TaskGroup('transform')
extract >> transform

C. with TaskGroup('extract') as extract:
t1 = DummyOperator(task_id='t1')
t1b = DummyOperator(task_id='t1b')
with TaskGroup('transform') as transform:
t2 = DummyOperator(task_id='t2')
t2b = DummyOperator(task_id='t2b')
extract >> transform

D. with TaskGroup('extract'):
t1 = DummyOperator(task_id='t1')
t1b = DummyOperator(task_id='t1b')
with TaskGroup('transform'):
t2 = DummyOperator(task_id='t2')
t2b = DummyOperator(task_id='t2b')
t1 >> t2

Solution

1. Step 1: Assign TaskGroups to variables

   Use 'with TaskGroup(...) as var:' to create groups and assign them to variables for referencing in dependencies.

2. Step 2: Set group-level dependencies

   Use extract >> transform to ensure every task in 'extract' (t1, t1b) completes before any task in 'transform' (t2, t2b) starts.

3. Step 3: Check options

   with TaskGroup('extract') as extract: t1 = ... t1b... with TaskGroup('transform') as transform: t2 = ... t2b... t1 >> t2 only links t1 to t2; t1b independent of transform tasks. extract = TaskGroup('extract') transform = TaskGroup('transform') extract >> transform misses 'with' context; no tasks added to groups. with TaskGroup('extract'): t1 = ... t1b... with TaskGroup('transform'): t2 = ... t2b... t1 >> t2 misses group variables and only links specific tasks. with TaskGroup('extract') as extract: t1 = ... t1b... with TaskGroup('transform') as transform: t2 = ... t2b... extract >> transform correctly uses group dependency.

4. Final Answer:

   with TaskGroup('extract') as extract: t1 = DummyOperator(task_id='t1') t1b = DummyOperator(task_id='t1b') with TaskGroup('transform') as transform: t2 = DummyOperator(task_id='t2') t2b = DummyOperator(task_id='t2b') extract >> transform -> Option C

5. Quick Check:

   tg1 >> tg2 for complete group ordering [OK]

Hint: Assign 'with TaskGroup(name) as tg:' then tg1 >> tg2 [OK]

Common Mistakes:

• Using individual task deps (t1 >> t2) with multiple tasks per group
• Missing 'with' context manager
• Omitting 'as var' when using group dependencies
• No tasks added without context manager