Prerequisites to Configure Non-containerized UDFs
The following are the prerequisites to configure non-containerized Python UDFs:
Linux:
1. Install Python 3.10 libraries using the Python 3.10 installer appropriate for your operating system and version.
Note: Do not change the default version of Python on your operating system. Install 3.10 in a separate area and set this library path for Vector.
2. Create a symbolic link in the /usr/lib directory of your installation, pointing libpython3.10.so.1.0 to the Python 3.10 library to be used. For example,
ln -s /usr/lib64/libpython3.10.so.1.0 /opt/Actian/VectorVW/ingres/lib
/libpython3.10.so.1.0
3. Install Numpy. If you are using the default Python 3.10 for Vector, use the following command:
pip3 install numpy
Note: The Linux environment variable PYTHONHOME must either be empty or set to a valid path.
Windows:
1. Set the Vector environment variable UDF_PYTHONHOME to the path where Python is installed. For example:
ingsetenv UDF_PYTHONHOME C:\Python310
2. Set the Windows System environment variable PYTHONHOME in System Properties (select Control Panel > System > Advanced system settings > Environment variables > System variables) to the path where Python is installed. For example,
PYTHONHOME=C:\Python310
3. Install Numpy. If you are using the default Python 3.10 for Vector, use the following command:
pip3 install numpy
IMPORTANT! In the production environment, UDFs should run containerized.
Last modified date: 04/15/2025