Installation¶

Getting Airflow
Requirements
Extra Packages
Provider packages
Initializing Airflow Database
Docker image
Troubleshooting
- Symbol not found: _Py_GetArgcArgv

Getting Airflow ¶

Airflow is published as apache-airflow package in PyPI. Installing it however might be sometimes tricky because Airflow is a bit of both a library and application. Libraries usually keep their dependencies open and applications usually pin them, but we should do neither and both at the same time. We decided to keep our dependencies as open as possible (in setup.cfg and setup.py) so users can install different version of libraries if needed. This means that from time to time plain pip install apache-airflow will not work or will produce unusable Airflow installation.

In order to have repeatable installation, however, starting from Airflow 1.10.10 and updated in Airflow 1.10.13 we also keep a set of "known-to-be-working" constraint files in the constraints-master and constraints-1-10 orphan branches. Those "known-to-be-working" constraints are per major/minor python version. You can use them as constraint files when installing Airflow from PyPI. Note that you have to specify correct Airflow version and python versions in the URL.

Prerequisites

On Debian based Linux OS:
sudo apt-get update
sudo apt-get install build-essential

Installing just Airflow

Note

On November 2020, new version of PIP (20.3) has been released with a new, 2020 resolver. This resolver does not yet work with Apache Airflow and might leads to errors in installation - depends on your choice of extras. In order to install Airflow you need to either downgrade pip to version 20.2.4 pip upgrade --pip==20.2.4 or, in case you use Pip 20.3, you need to add option --use-deprecated legacy-resolver to your pip install command.

AIRFLOW_VERSION=1.10.14
PYTHON_VERSION="$(python --version | cut -d " " -f 2 | cut -d "." -f 1-2)"
# For example: 3.6
CONSTRAINT_URL="https://raw.githubusercontent.com/apache/airflow/constraints-${AIRFLOW_VERSION}/constraints-${PYTHON_VERSION}.txt"
# For example: https://raw.githubusercontent.com/apache/airflow/constraints-1.10.14/constraints-3.6.txt
pip install "apache-airflow==${AIRFLOW_VERSION}" --constraint "${CONSTRAINT_URL}"

Please note that with respect to Python 3 support, Airflow 1.10.14 has been
tested with Python 3.6, 3.7, and 3.8, but does not yet support Python 3.9.

Installing with extras (for example postgres, google)

Note

On November 2020, new version of PIP (20.3) has been released with a new, 2020 resolver. This resolver does not yet work with Apache Airflow and might leads to errors in installation - depends on your choice of extras. In order to install Airflow you need to either downgrade pip to version 20.2.4 pip upgrade --pip==20.2.4 or, in case you use Pip 20.3, you need to add option --use-deprecated legacy-resolver to your pip install command.

AIRFLOW_VERSION=1.10.14
PYTHON_VERSION="$(python --version | cut -d " " -f 2 | cut -d "." -f 1-2)"
CONSTRAINT_URL="https://raw.githubusercontent.com/apache/airflow/constraints-${AIRFLOW_VERSION}/constraints-${PYTHON_VERSION}.txt"
pip install "apache-airflow[postgres,google]==${AIRFLOW_VERSION}" --constraint "${CONSTRAINT_URL}"

Most of the extras are linked to a corresponding providers package. For example "amazon" extra has a corresponding apache-airflow-providers-amazon providers package to be installed. When you install Airflow with such extras, the necessary provider packages are installed automatically (latest versions from PyPI for those packages). However you can freely upgrade and install provider packages independently from the main Airflow installation.

Requirements ¶

You need certain system level requirements in order to install Airflow. Those are requirements that are known to be needed for Linux system (Tested on Ubuntu Buster LTS) :

sudo apt-get install -y --no-install-recommends \
        freetds-bin \
        krb5-user \
        ldap-utils \
        libffi6 \
        libsasl2-2 \
        libsasl2-modules \
        libssl1.1 \
        locales  \
        lsb-release \
        sasl2-bin \
        sqlite3 \
        unixodbc

You also need database client packages (Postgres or MySQL) if you want to use those databases.

If the airflow command is not getting recognized (can happen on Windows when using WSL), then ensure that ~/.local/bin is in your PATH environment variable, and add it in if necessary:

PATH=$PATH:~/.local/bin

Extra Packages ¶

The apache-airflow PyPI basic package only installs what's needed to get started. Subpackages can be installed depending on what will be useful in your environment. For instance, if you don't need connectivity with Postgres, you won't have to go through the trouble of installing the postgres-devel yum package, or whatever equivalent applies on the distribution you are using.

Behind the scenes, Airflow does conditional imports of operators that require these extra dependencies.

For the list of the subpackages and what they enable, see: Extra Packages Reference.

Provider packages ¶

Unlike Apache Airflow 1.10, the Airflow 2.0 is delivered in multiple, separate, but connected packages. The core of Airflow scheduling system is delivered as apache-airflow package and there are around 60 providers packages which can be installed separately as so called "Airflow Provider packages". The default Airflow installation doesn't have many integrations and you have to install them yourself.

You can even develop and install your own providers for Airflow. For more information, see: Provider packages

For the list of the provider packages and what they enable, see: Providers packages reference.

Initializing Airflow Database ¶

Airflow requires a database to be initialized before you can run tasks. If you're just experimenting and learning Airflow, you can stick with the default SQLite option. If you don't want to use SQLite, then take a look at Initializing a Database Backend to setup a different database.

After configuration, you'll need to initialize the database before you can run tasks:

airflow db init

Docker image ¶

Airflow is also distributed as a Docker image (OCI Image). For more information, see: Production Deployment

Troubleshooting ¶

This section describes how to troubleshoot installation issues.

`Symbol not found: _Py_GetArgcArgv`¶

If you see Symbol not found: _Py_GetArgcArgv while starting or importing Airflow, this may mean that you are using an incompatible version of Python. For a homebrew installed version of Python, this is generally caused by using Python in /usr/local/opt/bin rather than the Frameworks installation (e.g. for python 3.7: /usr/local/opt/python@3.7/Frameworks/Python.framework/Versions/3.7).

The crux of the issue is that a library Airflow depends on, setproctitle, uses a non-public Python API which is not available from the standard installation /usr/local/opt/ (which symlinks to a path under /usr/local/Cellar).

An easy fix is just to ensure you use a version of Python that has a dylib of the Python library available. For example:

# Note: these instructions are for python3.7 but can be loosely modified for other versions
brew install python@3.7
virtualenv -p /usr/local/opt/python@3.7/Frameworks/Python.framework/Versions/3.7/bin/python3 .toy-venv
source .toy-venv/bin/activate
pip install apache-airflow
python
>>> import setproctitle
# Success!

Alternatively, you can download and install Python directly from the Python website.