Devel buildsys broken?

[Julian Sikorski]

Thorsten Leemhuis pisze:
> On 27.02.2009 10:55, Andrea Musuruane wrote:
>> On Thu, Feb 26, 2009 at 10:51 PM, Kevin Kofler
>> <kevin.kofler at chello.at> wrote:
>>> No, the problem is that you need to use RPM 4.6 on the host system to
>>> build
>>> Rawhide packages, see the discussions on the fedora-devel-list.
>> If the problem is that our build system doesn't yet use RPM 4.6, why
>> don't we update it?
> 
> One reason: Because the PPC builder uses F9 and is used for other things
> as well -- so simply updating it to a RPM version might break other
> crucial services on that machine. Thus is needs to be properly discussed
> and planed how to move forward before we can do something. I tried to
> reach the owner of the PPC machine (who's traveling right now) yesterday
> to discuss a solution, but didn't get a response yet.
> 
> Another reason: Somebody needs to sit down and actually do install the
> new rpm (a sane one; not one that is found unsigned somewhere on the
> net!) on the two x86-builders that both run CentOS 5.
> 
> 
> But the real reason is a different one:
> 
> *Our organization in regards to infrastructure maintenance sucks. A lot.*
> 
> We afaics need a real infrastructure team with at least two or three
> *active* (e.g. not only driven by events like breakage) people that take
> care of the builders and all the other infra we have. Only then RPM
> Fusion will last. I had hoped Xavier would form such a team, but that
> didn't happen afaics and he afaics seems to be quite busy with real life
> and Fedora-tasks right now, otherwise I guess he would have foreseen
> this problem and acted already.
> 
> CU
> knurd
> 
> P.S.: Please note that I don't want to get counted as part of the
> "infrastructure team". Yes, I have access to some (nearly all) of the
> builders (sometimes I'm the only one that has access), but I really want
> to hand the responsibility over to a infra team. But there is nobody I
> can hand it over to afaics. Reminder: I just did infra work in the past
> because somebody had to do it to get RPM Fusion running.
> 
> And another reminder: Our infra is quite fragile afaics. We have lots of
> single-points of failure and there are lots of hacks here and there to
> make things work. That could become a real problem in the long run if we
> don't form a real active infra team.
> 
I wonder how redhat folks have solved this problem, since afaik the koji
builders are running rhel.

Julian