Chameleon Changelog for September 2019

Great news in Chameleon-land!


Welcome to October, and a new (academic) year for all of you! Over the summer we’ve been working on some cool new things and are now happy to release them into the world for your use and enjoyment.


Easier access to all Chameleon sites. Notice any changes on We added so much stuff to the testbed that we were running out of space in the navigation menu and took the opportunity to do some re-organizing to make it easier for you to access all of the Chameleon sites and features. You’ll notice that there is a new “Experiment” dropdown--this contains links for every part of your experimental workflow, from discovering resources (hardware and software), to provisioning them on a site (bare metal or KVM), to documenting and publishing your work. Each experimentation site is now listed under a “Sites” section within this dropdown for easier access. We’ve also included convenient links to both the Jupyter Interface and the new Sharing Portal (see below!). Hopefully this will make it easier for you to find things -- and if not -- or if you would like to share any other "findability" concerns please let us know. We are standing by to make things easier.


A new way to share reproducible artifacts. You might nod your head in agreement if I said it is currently unreasonably difficult to reproduce most complex Computer Science experiments, especially if they involve resources and complex topologies provisioned on an open testbed like Chameleon. We’ve been paying attention to how much you all like to use the Jupyter Interface to Chameleon and have been exploring ways to use it to solve this problem. Today we’re happy to announce a preview release of a new Sharing Portal that is deeply tied to Jupyter and Chameleon--it allows you to re-create and re-run an experiment by following along with a Jupyter Notebook. Most importantly, you can publish your own Notebooks and experiment artifacts to the portal today! To get started, find the “Sharing” link in the “Experiment” section of



Currently there are just a few examples to get you started, but we will be adding to this list soon. One thing you will notice is that we have partnered with Zenodo for storage of your experiment artifacts--this means that your Jupyter Notebooks, supporting code, visualizations and more will be backed by CERN’s long-term artifact storage. Zenodo comes with some other cool features such as automatic DOI assignment and the ability to see view/download counts for your artifact files. For example, see the JupyterHub example here.


The most powerful feature, though, is the ability to re-launch the artifacts within Chameleon. Clicking “Launch with JupyterHub” will open a new Jupyter Notebook server with the artifacts download (size permitting; currently we support artifacts up to 500MB in total size, but can work with you for bigger projects). The below animation shows how easy it is:



Whether you’re making changes to an existing experiment or creating a new one, you can publish your own artifacts to the Sharing Portal! There is a new “Share” menu item in the Jupyter interface that allows you to upload your artifact files straight to Zenodo and then create a new shared artifact on Chameleon. The shared artifact will be created under your Chameleon account and you will be able to edit its metadata directly within the Sharing Portal.



We’re excited to work more on enabling faster and more streamlined flows around reproducibility to enable you to work more efficiently with Chameleon, and hope you enjoy this new release!


Less hassle with managing users in a project. If you manage a project with a lot of members, you probably have had issues handling adding new members, particularly if they hadn’t completed registration by confirming their email address. We’ve recently made some changes that will allow you to still invite these users, which should hopefully reduce back-and-forth you might encounter with your project members (or students.)


Some big changes coming in KVM land. It feels safe to announce that we’re actively working on a new KVM cluster to provide to those of you who prefer working with virtual machines. We’re working on upgrading our subsystems and brushing away lots of cobwebs at the moment, but will be announcing a new KVM site in the near future.


Upcoming webinars:


Conference deadlines coming up:

Add a comment

No comments