bstadlbauer/flyte-dev-setup

Tooling to nicely setup a Flyte dev cluster

Flyte dev setup

Repo used to easily and automatially end-to-end test the dask integration for Flyte.

Requirements

• poetry
• make
• kubectl
• helm
• flytectl

Usage

First, make sure to stand up a Flyte dev sandbox k8s cluster which has both all Flyte dependencies as well as the dask-k8s-operator running. You can do so using

make setup

During that you'll be asked to start Flyte locally, for more info checkout the development setup guide.

To create the python virtual environment, you can run

poetry install

To run all tests, use

poetry run pytest

The tests will:

1. Build a dockerimage based on docker/Dockerfile
2. Register that image to localhost:30000
3. Trigger a test workflow from flyte_dev_setup/workflows.py. This workflow is designed to test different facets of the plugin and will use the docker image mentioned above.

To check on the status of the workflow, you can check the Flyte UI at http://localhost:30080/console/projects/dask-testing/domains/development/workflows/flyte_dev_setup.workflows.dask_test_workflow

Versions

Versions of different components are specified in the following places:

• flytekit: ./pyproject.toml
• flytekitplugins-dask: ./pyproject.toml
• dask: ./pyproject.toml
• dask-kubernetes: ./pyproject.toml
• dask-kubernetes-operator: ./Makefile -> Will currently always use the latest (and perform a helm repo update)
• flyte: Whatever is started locally. Dependencies such as the ingress, docker registry will be whatever is used by flytectl demo start --dev.