Use executable objects with gateway services

You can use executable objects with Python services to specify a custom Python interpreter for your Python scripts. This allows different Python services to use different Python versions or configurations.

When to use executable objects with Python services

Use executable objects with Python services when you need to:

• Run different Python services with different Python versions (for example, Python 2 and Python 3)
• Use a Python interpreter installed in a non-standard location
• Use Python from a conda environment or pyenv installation
• Apply specific command-line flags to your Python interpreter

Default behavior

If you don’t specify an executable object, Python services use the system default Python interpreter. This maintains backward compatibility with existing Python services.

Prerequisites

• An executable object configured with a Python interpreter path
• A Git repository that contains your Python script
• Operator role permissions
• Repository authentication configured (if using a private repository)

Create a Python service with an executable object

Use the iagctl create service python-script command with the --executable-object parameter to create a Python service that uses a custom interpreter.

Syntax:

$
iagctl create service python-script <service-name> \
>
  --executable-object <object-name> \
>
  --repository <repository-name> \
>
  --working-dir <directory-path> \
>
  --filename <script-filename>

Parameters:

• <service-name>: A unique name for the Python service
• --executable-object: The name of the executable object that defines the Python interpreter
• --repository: The name of the Git repository that contains your Python script
• --working-dir: The path within the repository where your script is located
• --filename: The name of the Python script file to execute

Example:

Create a Python service that uses Python 3.11:

$
iagctl create service python-script simple-python \
>
  --executable-object python311-standard \
>
  --repository gateway-resources \
>
  --working-dir pythonscripts \
>
  --filename main.py

Examples for different Python installations

Use Python from a conda environment

$
iagctl create executable-object python-conda \
>
  --exec-command /opt/conda/envs/myenv/bin/python \
>
  --description "Python from conda environment"
$
$
iagctl create service python-script conda-python-service \
>
  --executable-object python-conda \
>
  --repository data-science-scripts \
>
  --working-dir analysis \
>
  --filename analyze.py

Use Python from pyenv

$
iagctl create executable-object python-pyenv \
>
  --exec-command /home/user/.pyenv/versions/3.10.5/bin/python \
>
  --description "Python 3.10.5 from pyenv"
$
$
iagctl create service python-script pyenv-python-service \
>
  --executable-object python-pyenv \
>
  --repository automation-scripts \
>
  --working-dir python \
>
  --filename automation.py

Use a specific Python version for testing

$
iagctl create executable-object python39-test \
>
  --exec-command /usr/bin/python3.9 \
>
  --description "Python 3.9 for compatibility testing"
$
$
iagctl create service python-script test-python-service \
>
  --executable-object python39-test \
>
  --repository test-suite \
>
  --working-dir tests \
>
  --filename run_tests.py

Run a Python service with an executable object

Run Python services the same way you run other services:

$
iagctl run service python-script simple-python

IAG uses the executable object configuration to run your Python script with the correct Python interpreter.

Static environments and executable objects

You cannot use static environments and executable objects together in the same Python service. If you configure a Python service with both a static environment and an executable object, IAG returns an error.

Choose one of the following approaches for each Python service:

• Use a static environment for package management with the system Python interpreter
• Use an executable object to specify a custom Python interpreter without static environments