[OpenAFS-devel] Re: RHEL5 buildslaves

[Andrew Deason]

On Fri, 16 Aug 2013 16:34:08 -0400
Jason Edgecombe <jason@rampaginggeek.com> wrote:

> I think that the "identically configured" delaying the other builders
> part are the big blockers. I suspected that the suggestion was a
> non-starter, but wanted to double-check.

While I agree it's better that way and should be goal, I don't think
that should block any new slaves outright. We've had, and will continue
to have, non-deterministic build failures, and it's not the end of the
world. They are problems, but they are problems that can identified and
fixed. If a few builds failed solely depending on which slave they were
on, it would be even easier to identify.

-- 
Andrew Deason
adeason@sinenomine.net