Trovi Collections and Citations. Trovi is Chameleon's repository of shared artifacts, allowing you to browse through hundreds of experiments that run on Chameleon. This month, we are excited to announce artifact collections. Now, you can create an artifact and then link it to other related artifacts to form a collection. These links can be used to group together artifacts that are part of a series, such as part of a course, or artifacts that are a part of a conference. We are looking forward to seeing all of the collections that our users will share on Trovi over the next few months!

We've also added a panel in Trovi artifact detail page that shows how to cite the current artifact. Previously, this information was only in our FAQ, which was difficult to find. Now you can easily copy the citation or bibtex information for any artifact from the dashboard directly.

Improved Storage at CHI@TACC and KVM@TACC. We've installed new hardware for the CHI@TACC and KVM@TACC storage cluster, which is used by the object store, filesystem shares, and VM volumes. Previously, this cluster was running on a combination of HDDs and SSDs, and now it entirely operates on NVMe SSDs. Additionally, we've improved the network connection on this cluster, and overall expect improvements to latency and bandwidth. Since this is a shared environment, we don't have guarantees on the performance you'll see in e.g., your KVM instances, but these changes should mean overall smoother IO performance and lower latency with fewer spikes.

Improved Resource Discovery API. This month, we've made significant improvements to our Resource Discovery API. This API is used internally by the Hardware Browser but is also open for your use. We've significantly improved the API response time, meaning that the hardware browser loads much faster. If you were previously using the REST API, please note that the API response has changed slightly. Mainly, we've removed parts of the response that were nonfunctional, such as the cluster status endpoint.

Vendordata Improvements. Vendordata is one of the services that runs on the testbed. It's used to generate the openrc file and the "message of the day" text when you login to a node. This month, we've improved the API response time of the service. We also fixed an issue on KVM, where the generated openrc file would not work properly.