Chameleon Changelog for November 2019

Great news in Chameleon-land!

The turkey has turned, the online deals have soured, but there’s always one thing you can count on to not go bad: a Chameleon changelog.

Cascade Lake hardware available for all users! Newly added to CHI@TACC are 24 Dell PowerEdge R740 nodes equipped with Intel Cascade Lake CPUs, 192 GB RAM, and a 240GB SSD. These nodes have taken the place of several of our older Haswell-based nodes and are available for reservations immediately! The cost in SUs is the same as the Haswell or Skylake compute nodes. Check out the detailed specifications in our hardware catalog.

Version support for shared artifacts. In September we announced the Sharing Portal: a collection of, and more importantly a mechanism for contributing, experimental artifacts for others to reuse, redeploy, and reconsider. This month we have added support for versioned artifacts. Each artifact you publish will receive a DOI from Zenodo and is considered “frozen” at that point, so that it can be reliable reference in the future. However, sometimes you may desire to publish a new version, if your artifact is e.g., a library, code sample, or reference implementation. Perhaps you simply wish to correct some errata you found after publishing the initial version. In any case, you can upload a new set of files (whatever they may be) and publish a new version of your existing artifact from your Jupyter server with the click of a button. The new version will be semantically linked to the prior versions to ensure the history is not lost. Check out our docs for more information about the Sharing Portal!

A more streamlined way to launch instances. When launching bare metal instances via the Chameleon GUI, it was always required to select the “baremetal” flavor explicitly. Keeping with our focus on bare metal provisioning, we’ve made this the default selection, removing another step from the process of launching an instance! We also made the disk image selection screen load a looooot faster for you.

Reminder: old KVM cloud approaching end-of-life in January. Last month we announced the introduction of a new KVM deployment, which includes several stability improvements. We plan to decommission the old KVM cloud in January 2020. Please work to migrate your workloads to the new environment; in many cases, this should be as simple as switching URLs. See our documentation for more information on the migration, and let us know if you encounter issues.

Upcoming special topics webinar on Jupyter! Join us on December 10 for the webinar Make the Best Use of Your Allocations: Orchestrate Your Experiments. You know how to create an experiment using the Chameleon graphics interface, how about doing that in a reproducible and documented way? In this webinar we will orchestrate an experiment -- from resource allocation to gathering and visualizing the results to tearing it down -- using Jupyter Notebooks.

Add a comment

No comments