In this lesson, you’ll learn how to install Apache Airflow on a Kubernetes cluster. We’ll first deploy Airflow on Kubernetes using Helm, run a sample DAG, and then clean up the installation. By the end, you’ll have a fully functional Airflow setup on Kubernetes.

Please note that a basic understanding of Kubernetes is required to follow this tutorial.

Prerequisites

Before you start, make sure that the following tools are installed on your local system.

Step 1: Create airflow namespace

To start, let’s create a dedicated namespace for Airflow within the Kubernetes cluster. Using a separate namespace helps isolate Airflow’s resources, making it easier to manage, monitor, and troubleshoot while preventing conflicts with other deployments.

kubectl create ns airflow

Step 2: Add Airflow Helm repository

Now that the Airflow namespace is in place, we need to add the Airflow Helm repository. Use the following commands:

helm repo add apache-airflow https://airflow.apache.org
helm repo update
helm search repo airflow

To confirm that the Airflow repository has been added, run:

helm repo list

The result should include the name of the Airflow repository.

NAME        URL
apache-airflow   https://airflow.apache.org

With the repository configured, it’s time to deploy Apache Airflow to your Kubernetes cluster Helm install command bellow.

helm install airflow apache-airflow/airflow --namespace airflow --debug

Step 3: Verify the Deployment

To ensure that Airflow has been deployed successfully, you can check the status of the pods

kubectl get pods -n airflow

You should get results similar to the following.

NAME                                     READY   STATUS    RESTARTS   AGE
airflow-postgresql-0                     1/1     Running   0          5m17s
airflow-redis-0                          1/1     Running   0          5m17s
airflow-scheduler-6f48fcad45-kkxmr       2/2     Running   0          5m17s
airflow-statsd-7d985bcb6f-b42t7          1/1     Running   0          5m17s
airflow-triggerer-0                      2/2     Running   0          5m17s
airflow-webserver-9f64f5b98-gv4sk        1/1     Running   0          5m17s
airflow-worker-0                         2/2     Running   0          5m17s

Step 4: Port Forwarding for Airflow

Once Airflow is deployed, a service named airflow-webserver is also deployed. To get the name of the service, run the command:

kubectl get svc -n airflow

The result should resemble the screen below:

NAME                   TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)               AGE
airflow-postgresql     ClusterIP   10.108.13.202    <none>        5432/TCP              6m47s
airflow-postgresql-hl  ClusterIP   None             <none>        5432/TCP              6m47s
airflow-redis          ClusterIP   10.101.6.45      <none>        6379/TCP              6m47s
airflow-statsd         ClusterIP   10.102.16.173    <none>        9125/UDP, 9102/TCP    6m47s
airflow-triggerer      ClusterIP   None             <none>        8794/TCP              6m47s
airflow-webserver      ClusterIP   10.109.38.86     <none>        8080/TCP              6m47s
airflow-worker         ClusterIP   None             <none>        8793/TCP              6m47s

Now, run the following command tu to forward the service’s port to your local machine.

kubectl port-forward svc/airflow-webserver 8080:8080 -n airflow

Step 5: Log in to Airflow

After setting up Apache Airflow, please wait for a moment. Open a web browser and navigate to http://localhost:8080. You should see the Apache Airflow login page.

Log in using the following credentials:

  • Username: admin
  • Password: admin

Warning

Please note that you should not use these for production environments.

Step 6: Cleanup

Once you’ve completed your tasks and want to uninstall the Airflow Helm Charts Release, you can use the following command:

helm uninstall airflow --namespace airflow

You might also find it necessary to keep an eye on your Airflow installation. To achieve this, you can establish a monitoring platform by following the instructions in our guide Deploy Prometheus Operator in Kubernetes.

PreviousNext