airlift

Airlift

Introduction

Airlift is a Command Line Interface (CLI) tool designed to provide a local development environment for Apache Airflow with a simple but flexible interface. It is built on top of the official Airflow Helm Chart.

Requirements

Airlift requires the following software to be installed on your system:

• Helm
• Docker
• Kind

Below are the installation instructions for each of these tools on MacOS and Linux distributions.

Note: It is also recommended to allocate at least 4GB of RAM for Docker to run this service.

Option 1: Nix Installation

airlift is packaged and available via the nix package manager.

Install nix

Install Nix via the recommended multi-user installation:

sh <(curl -L https://nixos.org/nix/install) --daemon

Install airlift

nix-shell -p airlift

This will install airlift in a disposable shell. Once you exit the shell, it will no longer be available for use.

Note: airlift is only available on the 23.11 or unstable nix channel.

You can also run a single airlift command using this shell:

nix-shell -p airlift --command "airlift -h"

To install airlift permanently in your environment, you can use nix-env instead.

Option 2: Homebrew Installation

Homebrew is a package manager that we will use to install the necessary software. If you don't have Homebrew installed, you can install it by following these instructions:

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

Install Software

With Homebrew installed, you can now install Helm, Docker, and Kind.

Helm

brew install helm

Docker

brew install --cask docker

-OR- use a tool like Docker Desktop or Rancher Desktop

Kind

brew install kind

Note: This software was tested and validated working with kind v0.17. There are known issues with kind v0.20 and Rancher. If you are experiencing issues, please downgrade your kind installation by installing from the source/release binaries.

pip Installation

Airlift can be installed using pip:

pip install airlift

Usage

The general syntax for using the Airlift CLI tool is:

airlift [subcommand] [options]

Subcommands and Options

1. start

Starts the Airflow service.

airlift start -d /path/to/dags -p /path/to/plugins -r /path/to/requirements.txt

Note: The DAG and Plugins folders are mounted directly to the airflow service for hot-reloading. When you make a change locally, it should automatically appear in the Airflow UI.

Note: Start times for Airflow can be upwards of 5 minutes, due to the bootstrapping, installation of required PyPi packages, and the creation of the Postgres database. This all depends on your local machines power & the complexity of your Airflow setup.

2. check

Checks if all pre-requisite software is installed.

airlift check

3. pause

Pauses the Airflow service.

airlift pause

4. unpause

Unpauses the Airflow service.

airlift unpause

5. remove

Removes all containers/clusters related to the airlift service.

airlift remove

6. status

Checks the status of the service and whether or not it is reachable.

airlift status -P 8080

7. import_variables

Imports a variables.json file to a running Airflow instance.

airlift import_variables -P 8080 -V /path/to/variables.json

8. run_dag

Runs a DAG given an ID.

airlift run_dag -P 8080 -D example_dag_id

Configuration files

Helm values.yaml

This file provides the configuration for the Airflow Helm chart. This can be used for things such as:

• Setting the Secrets Backend to AWS Secrets Manager
• Adding custom environment variables (such as connections)
• Changing the executor
• Modifying the memory allocation for the webserver/scheduler/workers
• Updating any airflow.cfg value.

Here's an example:

executor: "CeleryExecutor"
config:
  core:
    load_examples: 'False'
    executor: CeleryExecutor 
    colored_console_log: 'False'
# Airflow scheduler settings
scheduler:
  #  hostAliases for the scheduler pod
  hostAliases: []
  #  - ip: "127.0.0.1"
  #    hostnames:
  #      - "foo.local"
  #  - ip: "10.1.2.3"
  #    hostnames:
  #      - "foo.remote"

  # If the scheduler stops heartbeating for 5 minutes (5*60s) kill the
  # scheduler and let Kubernetes restart it
  livenessProbe:
    initialDelaySeconds: 10
    timeoutSeconds: 20
    failureThreshold: 5
    periodSeconds: 60
    command: ~
  # Airflow 2.0 allows users to run multiple schedulers,
  # However this feature is only recommended for MySQL 8+ and Postgres
  replicas: 1
  # Max number of old replicasets to retain
  revisionHistoryLimit: ~

  # Command to use when running the Airflow scheduler (templated).
  command: ~
  # Args to use when running the Airflow scheduler (templated).
  args: ["bash", "-c", "exec airflow scheduler"]

You can find all the possible configuration overrides here: https://artifacthub.io/packages/helm/apache-airflow/airflow?modal=values

Note: By default, we disable the livenessProbe checks for the scheduler & triggerer due to conflicts with Kind. See ./src/airlift/config/helm/values.yaml for the exact config values

Airlift Configuration

The Airlift configuration file overrides all flag values to simplify starting the service.

For example, $HOME/.config/airlift/config.yaml:

# config.yaml
dag_path: /path/to/dags
plugin_path: /path/to/plugins
requirements_file: /path/to/requirements.txt
helm_values_file: /path/to/values.yaml
extra_volume_mounts:
  - hostPath=/my/cool/path,containerPath=/my/mounted/path,name=a_unique_name
cluster_config_file: /path/to/cluster/config.yaml
image: 'apache/airflow:2.6.0'
helm_chart_version: '1.0.0'
port: 8080
post_start_dag_id: 'example_dag_id'

In this example, dag_path in the yaml file overrides the -d setting, plugin_path overrides the -p setting, and so forth.

Using this configuration, you can now start the service using:

airlift start -c $HOME/.config/airlift/config.yaml

Examples

See here for examples with common configuration modifications.

FAQ

See here for Frequently Asked Questions

Motivation

The motivation behind the creation of Airlift is to simplify the process of setting up a local development environment for Apache Airflow. It aims to be a flexible tool that allows developers to easily configure and manage their Airflow instances with unlimited flexibility.

Support and Contribution

If you encounter any issues or have suggestions for improvements, feel free to open an issue on the GitHub repository. Contributions to the project are also welcome.

Contact

If you have questions or feedback about Airlift, please reach out by opening an issue on the GitHub repository.