Am Freitag 03 Dezember 2010 schrieb Hans-Peter Jansen:
Hmm, that heuristic sounds reasonable at first.
I think, it could be improved by adding a fairness/charity ratio: a fixed amount of build resources, let's say 5%, that that weights by project finished ratio (packages missing / packages done) preferring high values, combined with one that prefers the longest waiting one. The fixed amount of resources is only taken, if there are any aspirants in this group. By turning the knob, the overall fairness can be controlled on the fly.
While this sounds fair, it will make your situation worse. Because it will lead to all projects finishing at the same time - if I understood your proposal correct, because you seem to say that we should prefer the project that has 90 scheduled over those that has 1 scheduled. But it doesn't matter how many are scheduled - in the end you want a published repo - when _all_ are done. So what I consider more fair is that the build power is spent on stuff that people are interested at. This can be either people downloading or developers testing. Right now I notice way too many rebuilds of stuff I expect noone to care for. What the dispatcher should aim for is decreasing the average time of waiting for a package result - if noone waits for it, there is no reason to touch it. Greetings, Stephan -- To unsubscribe, e-mail: opensuse-buildservice+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-buildservice+help@opensuse.org