python

spark-k8s-test

The main purpose of the spark-k8s-test library is to provide a seemless, user-friendly interface to implement tests for Charmed Apache Spark. spark-k8s-test provides a number of pytest fixtures that allow to create, manage and destroy various kind of resources needed for testing, such as:

• K8s resources (e.g. namespaces, pods, service accounts)
• S3 Buckets
• spark8t managed service accounts / secrets

Moreover, spark-k8s-test provides simple API to run spark-jobs within a Python testing environment

Quick start

Installation

The package can be simply installed from PyPI using

pip install spark-k8s-test 

or from the source by cloning this repo and then running:

pip install .

The package assumes that a MicroK8s installation is present in the system, and that the deployment is configured with MinIO to provide object storage capabilities.

You can use the following scripts to setup a MicroK8s deployment along with MinIO:

./spark_test/resources/bin/microk8s.sh setup

./spark_test/resources/bin/microk8s.sh configure dns rbac minio

Basic Usage

You can simply import the fixtures and use them straight away in your tests:

from spark_test.fixtures.pod import admin_pod
from spark_test.fixtures.service_account import registry, service_account

def test_pod_admin(admin_pod, service_account):

    output: str = admin_pod.exec(
        ["spark-client.service-account-registry", "list"]
    ).decode("utf-8")

    assert output.strip("\n") == service_account.id

Development

For development, we recommend to use Poetry. Once Poetry is installed, you can simply setup the project by

poetry install

Tox also provide easy short-cut to provide CI/CD capabilities such as

• Auto-Formatting: tox -e fmt
• Linting and Static Typing: tox -e lint
• Unittest: tox -e unittest
• Integration tests: tox -e integration

User acceptance tests and benchmarks

The tox environments can be run against a live deployment to assert its compliance. Here is an example running the benchmarks environment to stress-test a deployment in a spark-model Juju model:

tox run -e benchmarks -- --backend terraform --backend-storage azure --no-deploy --keep-models --model spark-model

Contributing

Canonical welcomes contributions to the spark-k8s-test library. Please check out our guidelines if you're interested in contributing to the solution. Also, if you truly enjoy working on open-source projects like this one and you would like to be part of the OSS revolution, please don't forget to check out the open positions we have at Canonical.

License

The spark-k8s-test toolkit is free software, distributed under the Apache Software License, version 2.0. See LICENSE for more information.

See LICENSE for more information.