README.md

This directory contains code for the components that comprise the Kubeflow Pipelines backend.

Building & Testing

To run all unittests for backend:

go test -v -cover ./backend/...

The API server itself can be built using:

go build -o /tmp/apiserver backend/src/apiserver/*.go

Code Style

Use gofmt package to format your .go source files. There is no need to format the swagger generated go clients, so only run the following command in ./backend/src and ./backend/test folder.

go fmt ./...

For more information, see this blog.

Building APIServer image locally

The API server image can be built from the root folder of the repo using:

export API_SERVER_IMAGE=api_server
docker build -f backend/Dockerfile . --tag $API_SERVER_IMAGE

Deploy APIServer with the image you own build

Run

kubectl edit deployment.v1.apps/ml-pipeline -n kubeflow

You'll see the field reference the api server docker image. Change it to point to your own build, after saving and closing the file, apiserver will restart with your change.

Building client library and swagger files

After making changes to proto files, the Go client libraries, Python client libraries and swagger files need to be regenerated and checked-in. Refer to backend/api for details.

Updating licenses info

1. Install go-licenses tool and refer to its documentation for how to use it.

2. Run the tool to update all licenses:

   make all

Visualization Server Instructions

Updating python dependencies

pip-tools is used to manage python dependencies. To update dependencies, edit requirements.in and run ./update_requirements.sh to update and pin the transitive dependencies.

Building conformance tests (WIP)

Run

docker build . -f backend/Dockerfile.conformance -t <tag>