How to move on with infrastructure?

[Dan Horák]

Thorsten Leemhuis píše v St 21. 10. 2009 v 20:27 +0200: 
> Hi!
> 
> Some of you might be aware of it, other not, but we afaics have a
> (whole) lot of medium-sized and small problems in the infrastructure
> area(¹). The main ones from the top of my head:
> 
> - our infrastructure team basically consists of just one person: Xavier
> Lamien (laxathom/SmootherFrOgZ)
> 
> - multiple people offered to help in the past (²), but not even one of
> those people found its way into the project/the infrastructure team:
> Why? It can't continue like that, as that is a route to fail...
> 
> - it sometimes take a lot of time until cvs requests are done
> 
> - we have a x86-64 builder that isn't used since a few weeks (months?)
> 
> - we afaik were asked to migrate to a different ppc builder but that
> work stalled
> 
> - using koji and bodhi for RPM Fusion is something that people are
> asking for every few weeks

I have setup my private Koji instance connected to Fedora using the
external repo feature and played a bit with mash. I have also some know
how from Fedora secondary archs about using distributed Koji
architecture with remote builders. So I'm offering my help in this area.

> - there are multiple small problems (often not in bugzilla) that are
> annoying, but not pressing (no automatic rebuilds, no repoclosure
> scripts, repodata regeneration after push, incomplete delta-rpm support
> are some of them, out-of-diskspace situations, some builders use mock
> caching while others do not, rawhide out of date on of of the builders, ...)
> 
> 
> So how to we fix all of the above to make sure RPM Fusion gets more
> healthy over time?


Dan