On Oct 9, 2005, at 02:04, Steve Langasek wrote:
* There needs to be another buildd. There have been numerous offers inthe past. I have no idea what to do to actually make it happen.Well, buildds are one of the more involved bits of our ports because they represent a time committment from a number of people, and when they go wrong they go very, very wrong. So we do want to be careful that we act on the *right* hosting offer; the last thing we want is to put a lot of work into a box that's going to disappear a few months later for whatever reason, andhave to start over.
I've never paid much attention to what goes into the buildd system, so these may be very naive questions, but I'm curious:
Why is so much attention required? Is it for stuff that can't be automated, or just hasn't been? (Aside from the really obvious, like hardware issues.)
Can multiple, lower-powered machines help? I'll probably be getting my hands on another alpha system soon, a 400AU I think, but it sounds like that may not be fast enough? (I've also got spare cycles on another alpha-linux box, and if we had an Alpha equivalent to Xen, I could spare cycles from my alpha-netbsd box too.) No GS320s on the horizon for me.
I'd like to help out, but don't have the biggest amount of resources to contribute, nor lots of my time on an ongoing basis, so it never sounds like it's worth getting into it. But if I it were just a matter of "oh, we wedged the machine again, please hit the reset button and boot off the Buildd Live CD when you get a chance so you can rejoin the build cluster", then I'd probably pursue it....
Ken