Re: Architecture qualification
* Adam D. Barratt (adam@adam-barratt.org.uk) [120515 22:26]:
> On Tue, 2012-05-15 at 21:45 +0200, Julien Cristau wrote:
> > On Tue, May 15, 2012 at 20:42:14 +0100, Adam D. Barratt wrote:
> >
> > > On Tue, 2012-05-15 at 20:42 +0200, Julien Cristau wrote:
> > > > I'd add a concern about the mips buildds to the arch qual page (not sure
> > > > how to phrase it).
> > >
> > > Assuming it's the stability issue, then maybe re-using weasel's
> > > "unstable under load" which we used for the porter box?
> > >
> > There's that, and there's some packages that can only be built on ball
> > because lucatelli/corelli fail every time. azeem has a bunch of those.
>
> Hmmm, ball's lower spec than the octerons iirc?
Different spec. swarm and octeons have different advantages (octeons
have many but a slower processors, and less memory per used processor
than swarms (because we run two buildds in parallel per hardware, each
using up to 6 cpus)).
Anyways, if the most concering issue is that there is currently only
one swarm-type mips buildd, we could just use the spare machine we
have and add another one. (Normally packages can build on any
hardware, but sometimes it's more favourable to distribute packages
accordingly between different hardware types - nothing too bad, but
sometimes e.g. RAM is more important than many processors).
Andi
Reply to: