[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: The unofficial buildd effort and its shutdown - my POV



Ingo Juergensmann <ij@2004.bluespice.org> writes:

> On Tue, Sep 07, 2004 at 03:42:43AM -0700, Steve Langasek wrote:
>
>> > > Part of "trusting" people is trusting that they'll act in the interests
>> > > of the Debian community, which means agreeing to the SC.
>> > Again, does IBM agree with the SC? Does HP?
>> Does IBM have root access to the buildd?  Does HP?
>
> Hmmm, this is weird. 
> As a long term Linux user and especially a DD, I thought you would know that
> having physical access to the machines (nearly always) means having root
> access as well. 
>
>> > So you are saying that the ~800 packages backlog (counting needs-build
>> > and build but unhandled) on arm was all caused by one single package?
>> The backlog I see consists of less than 200 packages in needs-build.
>
> Yes, and what's one of the reasons why it's no longer >400-600 packages?
>
>> "Built but unhandled" is hardly solvable by adding hardware, is it?
>
> That's what I (and others) are telling since last year. Some archs don't
> have a hardware problem, but a personal problem. 
> But it's nice to see you agreeing with my POV in this regard... :)
>  
>> > It would be nice to have some preapproved people that can jump in on a
>> > moments notice and setup a buildd for a few days. A fast acting
>> > disater relieve team. And by that I mean are allowed to jump in. 
>> No, it would be nice to have n+1 redundancy for our buildd hardware (in
>> the form of excess buildd capacity that can keep up with the package
>> load in spite of any single hardware failure), so that people don't
>
> It might surprise you that for some strange (Murphys law) reasons buildds
> tend to have problems at once, not one by one in row... So, having just n+1
> redundancy might not be enough. 
>
>> *need* to "jump in on a moment's notice".  If having people "jump in" is
>> the best we can do, then it's the best we can do, but I still don't
>> consider it "nice".
>
> I don't know who pays your power bills, but I have to pay mine by myself and
> with the increasing power costs here in Germany I tend to power off machines
> that are not needed atm. So, why should I run a power consuming machine that
> just builds 1-2 packages a day, being idle the most time?
> But it makes sense to power them on when there's a bigger backlog. 
> Nobody needs to tell me when it's necessary to turn my machines on/off. I
> have eyes in my head and can track the graphs on buildd.net or buildd.d.o by
> myself. Yes, I've obtained that skill during my 4 years work running an
> official buildd, so I can judge on my own when an additional machine makes
> sense and when not. 
>
> -- 
> Ciao...              // 
>       Ingo         \X/

I have to agree with both of you and both Ingo and I have said
multiple times in the past that Debian needs more buildds (and Debian
has gotten more) and more variance in the buildd admin (which is still
way to centered around single persons).

Also buildds that act like a "hot-spare" disk in a raid and can be
turned on in a days notice are a great way to take over in case of
hardware or person problems. They can be at peoples homes where they
normaly wouldn't run 24/7 without any problem since they would only
bee needed in emergencies for a few days.

Call them official buildds in reserve if it makes you feel better.
They could still be the fast acting desaster relieve I talked about.

MfG
        Goswin



Reply to: