Version 0.3.0

This version includes many bug fixes and improvements from 0.2.0. Below are brief summaries of changes across the various components - please refer to 0.3.0 release notes found in individual repos for more details.

Breaking changes

If you are a user of Renku, there are two major changes that will impact your notebook sessions:

Data in LFS no longer automatically pulled

If your project contains data in LFS, it will no longer be automatically pulled
to save on disk space and to launch servers more quickly. Once in the interactive
session, you can pull data using

renku storage pull <path>

If your project contains relatively little data (~1GB) you can select the option
to automatically pull data in the notebook server launch dialog page.

Base image upgrade

Importantly, you should consider updating your Dockerfile and .gitlab-ci.yml
to use the new base images in your projects. You can find the two configuration
files in the 0.3.0 project template.

Dockerfile

In the Dockerfile in your repository, make sure that the top of the file reads

FROM renku/singleuser:0.3.2

.gitlab-ci.yml

In .gitlab-ci.yml, make sure that the beginning of the dot section reads:

dot:
stage: build
    image: renku/renku-python:v0.3.2

This will make sure that your interactive sessions use the newest release of the Renku CLI.

Helm charts

The helm charts have been refactored and changed significantly so a simple upgrade from 0.2.0 to 0.3.0 is not possible without changes to the deployment values used. The most critical breaking changes are:

• the move of jupyterhub from the main renku chart to the renku-notebooks chart -- see the release notes for details on the new deployment values.
• the gateway service now stores the user tokens -- see the release notes and the corresponding PR.

Component Updates

UI (renku-ui)

• pagination on Project pages
• improved JupyterLab launch page with resource requests
• use of gateway API proxy for all GitLab and JupyterHub requests
• improved landing page

Gateway (renku-gateway)

• The gateway is now stateful, storing the OAuth tokens for GitLab and JupyterHub in redis.

CLI (renku-python)

• The graph part of the CLI (renku-python) has been completely rewritten to use linked data as the source of provenance information throughout.

Notebooks service (renku-notebooks)

• accepts JupyterHub user tokens for API requests
• logic for checking repository access moved to the notebooks service away from the spawner
• logic for determining which image to launch moved to the notebooks service from the spawner
• added server options that are parsed by the UI and can be used for resource requests

Version 0.2.0

This is the first minimal stable release of the Renku platform since the complete redesign that began in late January 2018. We have designed the current platform with the working data scientist in mind, providing a first implementation of features to promote data science that is reproducible, reusable, and shareable.

This is the main repository, which contains all the necessary "plumbing" for getting the platform up and running. The 0.2.0 tags in other core repositories all correspond to this same release. In the future, the versions will likely not stay in lock-step. The other repositories included in this first release are:

• renku-gateway: API gateway
• renku-notebooks: external JupyterHub service for linking GitLab repositories and interactive environments
• renku-python: python client and CLI
• renku-ui: Web user interface

In addition, Renku relies on several off-the-shelf components, most notably GitLab for project management, Keycloak for authentication and JupyterHub for managing interactive environments.

User features:

• Tracking of analysis lineage on the command line via the CLI (see the renku-python docs for details or peruse our first steps tutorial
• Automatic generation of images for interactive environments based on specifications in the repository, currently supporting python out of the box
• Project creation and basic navigation in the web UI
• Spawning of hosted Jupyter notebook servers based on versioned and automatically created
• Simple merging of notebooks via the UI
• Media-rich discussions (Ku) including notebook embedding

Deployment features:

• Deploy on any kubernetes cluster via helm charts
• Each service is individually deployable, with the umbrella chart residing in the main renku repository
• Modular deployment with the possibility to deploy against existing instances of backend services like e.g. gitlab.com