Deckhouse python library

Simplifies writing module hooks for Kubernetes operators:

• Deckhouse
• Addon Operator
• Shell Operator (values are unaccessible)

NOTE:

• The API is in alpha stage

Install

pip install deckhouse

Sample hook

# hello.py
from deckhouse import hook

def main(ctx: hook.Context):
    # Manipulate kubernetes state
    # ... object = { "kind" : "Pod", "apiVersion" : "v1", ... }
    ctx.kubernetes.create_or_update(object)

    # Export metrics
    # ... metric = { "name" : "power", "group": "my_hook", "set" : 9000, ... }
    ctx.metrics.collect(metric)

    # Use in-memory values for helm chart. Shell Operator does not support values, but Addon Operator and Deckhouse do.
    ctx.values.myModule.deployment.replicas = 5


if __name__ == "__main__":
    hook.run(main, configpath="hello.yaml") # 'config' arg is also supported for raw string

# hello.yaml
configVersion: v1
onStartup: 10

How to test

An example for pytest

# hello_test.py

from hello import main
from deckhouse import hook

# Inputs
#   initial_values = { ... }
#   binding_context = [ { ... } ]
# Outputs
#   expected_metrics = [ ... ]
#   expected_kube_operations = [ ... ]
#   expected_values_patches = [ ... ]
#   expected_values = { ... }

def test_hello():
    out = hook.testrun(main, binding_context, initial_values)

    assert out.metrics.data == expected_metrics
    assert out.kube_operations.data == expected_kube_operations
    assert out.values_patches.data == expected_values_patches

    assert out.values.myModule.deployment.replicas == 5

Development

• Create virtual environment with python >= 3.8
• Install poetry
• poetry install
• make test