Kubeflow Development Environment
0 likes72 views
This document provides instructions for setting up a development environment for Kubeflow, including installing necessary tools like Git, Docker, programming languages Go and Python, and an IDE like Visual Studio Code. It also covers setting up local testing environments with Minikube or provisioning a cloud Kubernetes cluster on IBM Cloud. The goal is to have all tools installed and configured to contribute to Kubeflow projects through version control with Git and GitHub, build container images with Docker, and test code locally or on a cluster.
Kubeflow Development Environment
- 1. Kubeflow Development Environment Weiqiang Zhuang wzhuang@us.ibm.com IBM CODAIT Github id: adrian555
- 2. Development environment • No specifically required OS • Linux, MacOS, Windows, etc. • Source code version control and repository • Git and github • Container image • Docker engine • Languages • golang • python • IDE • Unit tests • Run/Debug • Deployment environment
- 3. Git and github • Install git • Follow the instruction • On MacOS, run `brew install git` • Access to github.com • Github documentation • Create github id if not yet • Set up git • git config --global user.name "github_id" • git config --global user.email "email" • Fork a repo • Clone the forked repo and add upstream • mkdir $HOME/go/src/github.com/kubeflow • cd $HOME/go/src/github.com/kubeflow • git clone https://github.com/adrian555/kfctl.git • cd kfctl • git remote add upstream https://github.com/kubeflow/kfctl.git
- 4. Docker engine • Install Docker • Follow the link • On MacOS, Docker Desktop will be installed. If necessary, change the resource configuration. • Access to container registry • For dockerhub, follow the link to create an account • For quay.io, follow the link to sign up • Others, such as gcr.io or your organization's own • Build and push a container image • Dockerfile • docker build –t <registry>/<org>/<image>:<tag> • docker push <registry>/<org>/<image>:<tag>
- 5. Languages and IDE • Install Go Tools • Follow the instructions • On MacOS, run `brew install go` • Set up GOPATH env • export GOPATH=$HOME/go • Install Python with Miniconda3 • Follow the instructions • Create an env, run `conda create -n myenv` • Install your favorite IDE • For Visual Studio Code, follow this link • Install extensions: Go, Python, YAML
- 6. Cluster • Minikube • Good for local testing, run on VM or laptop • Follow the instructions • For MacOS, more info is covered in this document • Cloud cluster • For IBM Cloud Kubernetes clusters, follow this link to provision. Details will be covered in the hands on session. • Others, such as AWS, GCP, etc.
- 7. Demo • Code and unit test with VSC • Run and debug