Hello CTA team,
We are evaluating CTA + dCache as the replacement for the current tape backend (TS4500 libraries with TS1160 drives).
We set up a production-shape lab running CTA 5.11.18 with a dCache 11.2 buffer end to end, and it works very well. However, and since our fleet standard is Debian, we had to run cta-taped / cta-rmcd / cta-frontend-grpc / cta-maintd as el9 podman containers which works, but is neither ideal nor acceptable to our operations team for a production infrastructure.
The current CTA build is locked to el9 (CMake platform detection, project.json references) and the only portability request we found has been closed as “will not do” in 2024.
We also noticed the alma10 attempt is blocked on dependency availability rather than code; Debian 13 already packages every build dependency in project.json at a compatible version except the librados and opentelemetry-cpp pins.
We think the question is timely following the recent news regarding CERN moving its accelerator-control computers to Debian 13 and experimental Debian package support in Koji, so:
- Is a Debian target something you would consider on the roadmap, even as “community-supported, not run at CERN”?
- If not, would you review and merge an external contribution? We would scope it to a CPack DEB path, a non-blocking Debian CI job, and the CMake change to accept ID=debian, for the PostgreSQL + gRPC subset we run.
We are ready to put engineering time into this, test the result on real TS4500 hardware over iSCSI, and report back here. Any guidance, including “no, and here is why”, would help us decide on our target architecture.
Thanks a lot.