Installation

Environment Requirements

Note Python version (3.5, 3.6, 3.7 or 3.8) is required for PyFlink. Please run the following command to make sure that it meets the requirements:

$ python --version
# the version printed here must be 3.5, 3.6, 3.7 or 3.8

Environment Setup

Your system may include multiple Python versions, and thus also include multiple Python binary executables. You can run the following ls command to find out what Python binary executables are available in your system:

$ ls /usr/bin/python*

To satisfy the PyFlink requirement regarding the Python environment version, you can choose to soft link python to point to your python3 interpreter:

ln -s /usr/bin/python3 python

In addition to creating a soft link, you can also choose to create a Python virtual environment (venv). You can refer to the Preparing Python Virtual Environment documentation page for details on how to achieve that setup.

If you don’t want to use a soft link to change the system’s python interpreter point to, you can use the configuration way to specify the Python interpreter. For specifying the Python interpreter used to compile the jobs, you can refer to the python client executable For specifying the Python interpreter used to execute the python UDF worker, you can refer to the python.executable

PyFlink is available through PyPi as a regular Python package and can be installed as follows:

$ python -m pip install apache-flink

You can also build PyFlink from source by following the development guide.

Note Starting from Flink 1.11, it’s also supported to run PyFlink jobs locally on Windows and so you could develop and debug PyFlink jobs on Windows.