Talos machines within the releng ecosystem are maintained through puppet/opsi. We need to get outside talos slaves up to date. Either a shadow puppet/opsi system or a comprehensive list of what needs to be installed along with installation instructions. I have already filed bug 589213 for the tools staging talos boxes, but this but also covers the work to create a separate addon talos production pool.

Given that you would probably want to add/remove slaves from your setup on your own timetable, I think it makes sense to get you some setup docs and/or scripts for configuring your own staging masters for puppet and opsi.

puppet/opsi are too fragile to try to replicate elsewhere, and probably will be for some time. Could we "bake" systems in the releng network, and then ship them off? Or make an image of a fully-cooked slave?

Is this substantially different from bug 557704, or can the two be combined?

The baking for the initial set up would be fine - but there is the ongoing maintenance burden as changes are made.

found in triage. For the last 6?9? months, I believe ATeam have been using ash and cedar project branches to test talos changes, instead of running their own standalone systems needing to be kept in sync. I think this means there's no longer a need, hence closing. However, if there is still a need, please reopen with details.