INSTALL / BUILD instructions for Apache Airflow

Basic installation of Airflow from sources and development environment setup

============================================================================

This is a generic installation method that requires minimum standard tools to develop airflow and

test it in a local virtual environment (using standard CPyhon installation and `pip`).

Depending on your system, you might need different prerequisites, but the following

systems/prerequisites are known to work:

Linux (Debian Bookworm):

sudo apt install -y --no-install-recommends apt-transport-https apt-utils ca-certificates \

curl dumb-init freetds-bin krb5-user libgeos-dev \

ldap-utils libsasl2-2 libsasl2-modules libxmlsec1 locales libffi8 libldap-2.5-0 libssl3 netcat-openbsd \

lsb-release openssh-client python3-selinux rsync sasl2-bin sqlite3 sudo unixodbc

You might need to install MariaDB development headers to build some of the dependencies

sudo apt-get install libmariadb-dev libmariadbclient-dev

MacOS (Mojave/Catalina) you might need to install XCode command line tools and brew and those packages:

brew install sqlite mysql postgresql

The `pip` is one of the build packaging front-ends that might be used to install Airflow. It's the one

that we recommend (see below) for reproducible installation of specific versions of Airflow.

As of version 2.8, Airflow follows PEP 517/518 and uses `pyproject.toml` file to define build dependencies

and build process, and it requires relatively modern versions of packaging tools to get airflow built from

local sources or sdist packages, as PEP 517 compliant build hooks are used to determine dynamic build

dependencies. In the case of `pip` it means that at least version 22.1.0 is needed (released at the beginning of

2022) to build or install Airflow from sources. This does not affect the ability to install Airflow from

released wheel packages.

Downloading and installing Airflow from sources

-----------------------------------------------

While you can get Airflow sources in various ways (including cloning https://github.com/apache/airflow/), the

canonical way to download it is to fetch the tarball (published at https://downloads.apache.org), after

verifying the checksum and signatures of the downloaded file.

When you download source packages from https://downloads.apache.org, you download sources of Airflow and

all providers separately. However, when you clone the GitHub repository at https://github.com/apache/airflow/

you get all sources in one place. This is the most convenient way to develop Airflow and Providers together.

Otherwise, you have to install Airflow and Providers separately from sources in the same environment, which

is not as convenient.

## Creating virtualenv

Airflow pulls in quite a lot of dependencies to connect to other services. You generally want to

test or run Airflow from a virtualenv to ensure those dependencies are separated from your system-wide versions. Using system-installed Python installation is strongly discouraged as the versions of Python

shipped with the operating system often have some limitations and are not up to date. It is recommended to install Python using the official release (https://www.python.org/downloads/), or Python project management tools such as Hatch. See later

for a description of `Hatch` as one of the tools that is Airflow's tool of choice to build Airflow packages.

Once you have a suitable Python version installed, you can create a virtualenv and activate it:

python3 -m venv PATH_TO_YOUR_VENV

source PATH_TO_YOUR_VENV/bin/activate

## Installing Airflow locally

Installing airflow locally can be done using pip - note that this will install "development" version of

Airflow, where all providers are installed from local sources (if available), not from `pypi`.

It will also not include pre-installed providers installed from PyPI. If you install from sources of

just Airflow, you need to install separately each provider you want to develop. If you install

from the GitHub repository, all the current providers are available after installing Airflow.

pip install .

If you develop Airflow and iterate on it, you should install it in editable mode (with -e) flag, and then

you do not need to re-install it after each change to sources. This is useful if you want to develop and

iterate on Airflow and Providers (together) if you install sources from the cloned GitHub repository.

Note that you might want to install `devel` extra when you install airflow for development in editable env

this one contains the minimum set of tools and dependencies needed to run unit tests.

pip install -e ".[devel]"

You can also install optional packages that are needed to run certain tests. In case of local installation

for example, you can install all prerequisites for Google provider, tests, and

all Hadoop providers with this command:

pip install -e ".[google,devel-tests,devel-hadoop]"

or you can install all packages needed to run tests for core, providers, and all extensions of airflow:

pip install -e ".[devel-all]"

You can see the list of all available extras below.

# Using Hatch to manage your Python, virtualenvs, and build packages

Airflow uses [hatch](https://hatch.pypa.io/) as a build and development tool. It is one of the popular

build tools and environment managers for Python, maintained by the Python Packaging Authority.

It is an optional tool that is only really needed when you want to build packages from sources, but

it is also very convenient to manage your Python versions and virtualenvs.

Airflow project contains some pre-defined virtualenv definitions in `pyproject.toml` that can be

easily used by Hatch to create your local venvs. This is not necessary for you to develop and test

Airflow, but it is a convenient way to manage your local Python versions and virtualenvs.

Installing Hatch

----------------

You can install Hatch using various other ways (including Gui installers).

Example using `pipx`:

pipx install hatch

We recommend using `pipx` as you can manage installed Python apps easily and later use it

to upgrade `hatch` easily as needed with:

pipx upgrade hatch

## Using Hatch to manage your Python versions

You can also use Hatch to install and manage airflow virtualenvs and development

environments. For example, you can install Python 3.10 with this command:

hatch python install 3.10

or install all Python versions that are used in Airflow:

hatch python install all

## Using Hatch to manage your virtualenvs

Airflow has some pre-defined virtualenvs that you can use to develop and test airflow.

You can see the list of available envs with:

hatch env show

This is what it shows currently:

┏━━━━━━━━━━━━━┳━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓

┃ Name ┃ Type ┃ Description ┃

┡━━━━━━━━━━━━━╇━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩

│ default │ virtual │ Default environment with Python 3.8 for maximum compatibility │

├─────────────┼─────────┼───────────────────────────────────────────────────────────────┤

│ airflow-38 │ virtual │ Environment with Python 3.8. No devel installed. │

├─────────────┼─────────┼───────────────────────────────────────────────────────────────┤

│ airflow-39 │ virtual │ Environment with Python 3.9. No devel installed. │

├─────────────┼─────────┼───────────────────────────────────────────────────────────────┤

│ airflow-310 │ virtual │ Environment with Python 3.10. No devel installed. │

├─────────────┼─────────┼───────────────────────────────────────────────────────────────┤

│ airflow-311 │ virtual │ Environment with Python 3.11. No devel installed │

├─────────────┼─────────┼───────────────────────────────────────────────────────────────┤

│ airflow-312 │ virtual │ Environment with Python 3.12. No devel installed │

└─────────────┴─────────┴───────────────────────────────────────────────────────────────┘

The default env (if you have not used one explicitly) is `default` and it is a Python 3.8

virtualenv for maximum compatibility with `devel` extra installed - this devel extra contains the minimum set

of dependencies and tools that should be used during unit testing of core Airflow and running all `airflow`

CLI commands - without support for providers or databases.

The other environments are just bare-bones Python virtualenvs with Airflow core requirements only,

without any extras installed and without any tools. They are much faster to create than the default

environment, and you can manually install either appropriate extras or tools that you need for

testing or development.

hatch env create

You can create specific environments by using them in create command:

hatch env create airflow-310

You can install extras in the environment by running pip command:

hatch -e airflow-310 run -- pip install -e ".[devel,google]"

And you can enter the environment by running a shell of your choice (for example, zsh) where you

can run any commands

hatch -e airflow-310 shell

Once you are in the environment (indicated usually by an updated prompt), you can just install

the extra dependencies you need:

[~/airflow] [airflow-310] pip install -e ".[devel,google]"

You can exit the environment by just exiting the shell.

You can also see where Hatch created the virtualenvs and use it in your IDE or activate it manually:

hatch env find airflow-310

You will get a path similar to the following:

/Users/jarek/Library/Application Support/hatch/env/virtual/apache-airflow/TReRdyYt/apache-airflow

Then you will find `python` binary and `activate` script in the `bin` sub-folder of this directory, and

you can configure your IDE to use this python virtualenv if you want to use that environment in your IDE.

You can also set the default environment name by the HATCH_ENV environment variable.

You can clean the environment by running the following:

hatch env prune

More information about hatch can be found in https://hatch.pypa.io/1.9/environment/

## Using Hatch to build your packages

You can use Hatch to build installable packages from the Airflow sources. Such package will

include all metadata configured in `pyproject.toml` and will be installable with pip.

The packages will have pre-installed dependencies for providers that are available when Airflow is installed from PyPI. Both `wheel` and `sdist` packages are built by default.

hatch build

You can also build only `wheel` or `sdist` packages:

hatch build -t wheel

hatch build -t sdist

## Installing recommended version of dependencies

Whatever virtualenv solution you use, when you want to make sure you are using the same

version of dependencies as in main, you can install the recommended version of the dependencies by using

constraint-python<PYTHON_MAJOR_MINOR_VERSION>.txt files as `constraint` file. This might be useful

to avoid "works-for-me" syndrome, where you use different versions of dependencies than the ones

that are used in main CI tests and by other contributors.

There are different constraint files for different Python versions. For example, this command will install

all basic devel requirements and requirements of Google provider as last successfully tested for Python 3.8:

pip install -e ".[devel,google]"" \

--constraint "https://raw.githubusercontent.com/apache/airflow/constraints-main/constraints-3.8.txt"

Using the 'constraints-no-providers' constraint files, you can upgrade Airflow without paying attention to the provider's dependencies. This allows you to keep installed provider dependencies and install the latest supported ones using pure Airflow core.

pip install -e ".[devel]" \

--constraint "https://raw.githubusercontent.com/apache/airflow/constraints-main/constraints-no-providers-3.8.txt"

Airflow extras

==============

Airflow has several extras that you can install to get additional dependencies. They sometimes install

providers, sometimes enable other features where packages are not installed by default.

You can read more about those extras in the extras reference:

https://airflow.apache.org/docs/apache-airflow/stable/extra-packages-ref.html

The list of available extras is below.

Core extras

-----------

Those extras are available as regular core airflow extras - they install optional features of Airflow.

# START CORE EXTRAS HERE

aiobotocore, apache-atlas, apache-webhdfs, async, cgroups, cloudpickle, deprecated-api, github-

enterprise, google-auth, graphviz, kerberos, ldap, leveldb, otel, pandas, password, pydantic,

rabbitmq, s3fs, saml, sentry, statsd, uv, virtualenv

# END CORE EXTRAS HERE

Provider extras

---------------

Those extras are available as regular Airflow extras; they install provider packages in standard builds

or dependencies that are necessary to enable the feature in an editable build.

# START PROVIDER EXTRAS HERE

airbyte, alibaba, amazon, apache.beam, apache.cassandra, apache.drill, apache.druid, apache.flink,

apache.hdfs, apache.hive, apache.iceberg, apache.impala, apache.kafka, apache.kylin, apache.livy,

apache.pig, apache.pinot, apache.spark, apprise, arangodb, asana, atlassian.jira, celery, cloudant,

cncf.kubernetes, cohere, common.io, common.sql, databricks, datadog, dbt.cloud, dingding, discord,

docker, elasticsearch, exasol, fab, facebook, ftp, github, google, grpc, hashicorp, http, imap,

influxdb, jdbc, jenkins, microsoft.azure, microsoft.mssql, microsoft.psrp, microsoft.winrm, mongo,

mysql, neo4j, odbc, openai, openfaas, openlineage, opensearch, opsgenie, oracle, pagerduty,

papermill, pgvector, pinecone, postgres, presto, qdrant, redis, salesforce, samba, segment,

sendgrid, sftp, singularity, slack, smtp, snowflake, sqlite, ssh, tableau, tabular, telegram,

teradata, trino, vertica, weaviate, yandex, zendesk

# END PROVIDER EXTRAS HERE

Devel extras

------------

The `devel` extras are not available in the released packages. They are only available when you install

Airflow from sources in `editable` installation - i.e., one that you are usually using to contribute to

Airflow. They provide tools like `pytest` and `mypy` for general-purpose development and testing.

# START DEVEL EXTRAS HERE

devel, devel-all-dbs, devel-ci, devel-debuggers, devel-devscripts, devel-duckdb, devel-hadoop,

devel-mypy, devel-sentry, devel-static-checks, devel-tests

# END DEVEL EXTRAS HERE

Bundle extras

-------------

Those extras are bundles dynamically generated from other extras.

# START BUNDLE EXTRAS HERE

all, all-core, all-dbs, devel-all, devel-ci

# END BUNDLE EXTRAS HERE

Doc extras

----------

Doc extras are used to install dependencies that are needed to build documentation. Only available during

editable install.

# START DOC EXTRAS HERE

doc, doc-gen

# END DOC EXTRAS HERE

Deprecated extras

-----------------

The deprecated extras are from Airflow 1 and will be removed in future versions.

# START DEPRECATED EXTRAS HERE

atlas, aws, azure, cassandra, crypto, druid, gcp, gcp-api, hdfs, hive, kubernetes, mssql, pinot, s3,

spark, webhdfs, winrm

# END DEPRECATED EXTRAS HERE

Compiling front-end assets

--------------------------

Sometimes, you can see that front-end assets are missing, and the website looks broken. This is because

you need to compile front-end assets. This is done automatically when you create a virtualenv

with hatch, but if you want to do it manually, you can do it after installing node and yarn and running:

yarn install --frozen-lockfile

yarn run build

Currently, we are running yarn coming with note 18.6.0, but you should check the version in

our `.pre-commit-config.yaml` file (node version).

Installing yarn is described in https://classic.yarnpkg.com/en/docs/install

Also - in case you use `breeze` or have `pre-commit` installed, you can build the assets with the following:

pre-commit run --hook-stage manual compile-www-assets --all-files

or

breeze compile-www-assets

Both commands will install node and yarn, if needed, to a dedicated pre-commit node environment and

then build the assets.

Finally, you can also clean and recompile assets with `custom` build target when running the Hatch build

hatch build -t custom -t wheel -t sdist

This will also update `git_version` file in the Airflow package that should contain the git commit hash of the

build. This is used to display the commit hash in the UI.