Migrating to OpenVox at CERN
At CERN we are currently switching our whole infrastructure to Openvox all over the place and would like to contribute by giving a talk during the VoxConf 2026. This switch, although simple for some other organisations (not a simple repo and package switch for us), showed some non-negligible technical debt and challenges. We would like to present our journey, past, present and future on our Puppet to Openvox transition
For many organizations, the migration from Puppet to OpenVox might be a matter of swapping repositories and running a package update. For CERN (home to the Large Hadron Collider and tens of thousands of heterogeneous nodes spanning data centers, accelerator controls, and physics analysis grids) it has been an archaeological dig through a decade and a half of institutional configuration history.
This is a post-mortem (and mid-mortem) of a massive enterprise pivot for a system that supports +15000 machines and +400 administrators. In our pursuit of a fully OpenVox-driven infrastructure, we discovered that the technical debt was not in the software itself, but in the abstractions we had built on top of the software. This is a description of the challenges we surpassed and will have coming later on this year (and beyond) to align with CERN's long-term opensource strategy.