Major highlights: dropping GridFTP/GSI, consolidating Gratia probes, transitioning to a rolling release, container tags that closely track OSG Yum repositories
Lots of Python 3 work: osg-configure, gratia-probe, HTCondor-CE, osg-build
Next steps: release XRootD and GlideinWMS
Created per-release series “upcoming” Yum repositories: 3.5-upcoming contains software with GSI + SciToken support
Added SciToken-based job submission tests (also added to 3.5-upcoming)
Concerns
SciToken support not yet there in the rest of the infrastructure; a handful of instances of sites updating/asking about updating to 3.6
GSI-based job submission tests missing from 3.5-upcoming
2
3 of 7
Enterprise Linux 8
Software Team attended a few HEP meetings about their OS plans: RPM is still a safe bet
WN software (CVMFS, client) is well-supported today
Built and tested GridFTP
HTCondor-CE 5.1.0 and HTCondor 9.0.0 releases upstream will open the floodgates for releases of the other EL8 packages
Concern: RHEL academic licensing is being discussed as an option by CERN/FNAL
3
4 of 7
Compute Federation
Compute Entrypoints:
Software along with Ops/Facilitation demonstrated end-to-end job submission to Bridges2 and Expanse with pilot usage counted against user allocations
SIUE: running a self-hosted CE in front of their Ubuntu 20 cluster with CVMFS!
Images now rebuilt at least weekly to follow Software container policies
Working on adding CVMFS support directly into the container
Concern: The certificate/DNS-free CE work project is still effort starved
4
5 of 7
Containers
Hosted 6 hackathons to work on Kubernetes/container related tasks and to educate internal staff/collaborators on Kubernetes and the GitOps model used on the Tiger cluster at Morgridge
Deployed an EL8 Koji RPM building worker on Tiger (necessary for EL8 builds, thanks to UChicago for holding us over until now!); working on a spare instance to test patches and more frequent updates
Assisted Ops with transitions of Topology and YUM repo services to Tiger
Starting to see more adoption of containers, especially for Frontier Squids: ATLAS is starting to use a federated operational model for ATLAS T2 site Squids
Next steps: develop internal tooling/strategies to produce release candidate images, improve ease of kicking off manual rebuilds, and add the ability to inspect images for important underlying software versions
5
6 of 7
Other
Software made major contributions to HTCondor
Provided development, coordination, testing, and documentation effort for HTCondor-CE 5.1.0
Assisting with improving build and release processes, such as improved GitHub integration and Ubuntu/daily build containerization efforts
Working with CMS and Operations to standardize Topology, Hosted CE, and Factory configuration for HPCs
Assisted the XRootD team with drafting a version support policy
Continued to provide technical support in campus engagements (GP-ARGO, UTC, Villanova)
More tool/workflow development for various PATh metrics