Chameleon Changelog for December 2018

Great news in Chameleon-land!


We prepared some presents for you that might have gotten held up by the postage service. Hope you enjoy them now!

More Power (metrics) to More People. Metrics are now provided on all bare metal nodes at TACC! The metrics provided are the instantaneous power (in watts), CPU 1 and 2 temperature, baseboard temperature, and inlet air temperature (all temperatures in Celsius). These metrics are collected on a 5-minute interval and are accessible via the Gnocchi database (a.k.a. OpenStack metrics.) Each bare metal node is a resource with the Ironic node id as its resource id and has the five metrics attached, similar to the previously-released low power nodes that were updated last month. Also similar to those metrics, the CPU temperature sensors only report when powered on but the other three sensors report back whether the node is powered on or off.


Dedicated 100G network between TACC and UC available for wider use. In April of this year we announced a new pre-defined network for high-bandwidth (up to 100 Gbps) wide-area networking experiments between Chameleon’s UC and TACC sites. We deployed a new shared network (“sharedwan1”) enabling high-bandwidth network traffic to flow between the sites over an isolated layer 2 network. However, this network was only available to users in a special “beta testers” project. We are happy to announce that now this network is available to all projects! In order to control and monitor usage, we simply ask that you send us a ticket requesting access to this network. A Chameleon operator will then enable it for your project. Currently sharedwan1 is limited in practice to 40 Gbps during the rollout period, however, we can lift the limitation and provide assistance for experiments that can show need for higher bandwidth. We look forward to seeing what you can do with this!


KVM traces with machine events. We have upgraded the Chameleon KVM trace format from version 0.1 to version 0.2. The major change in version 0.2 is the inclusion of machine events. Machine events provide information on the physical capacity of each machine, including number of CPU cores, amount of memory, disk size, as well as the availability of the machine over time. With the machine events table, you can run your simulations or your analysis more accurately. For more information about cloud traces, please visit our cloud traces page on Science Clouds and contact us if you would like to either use our data or contribute your own.

CHI-in-a-Box Alpha 0.2 release. In August we announced the first alpha release of CHI-in-a-Box, our set of provisioning tools that allow associate sites to host their own Chameleon testbed. We have made several improvements to the setup process, streamlined the installation, added some operational “special sauce” such as our automatic maintenance hammers, and added some helpful bootstrapping scripts that assist you in enrolling bare metal nodes into the OpenStack cluster. As with before, the alpha release is intended for adventurous users who are interested in partnering with Chameleon to provide scientific computational resources to the research community. Please contact us at if you are interested in learning more!

Add a comment

No comments