On Tue, Jun 14, 2011 at 2:12 PM, Konstantin Boudnik <c...@apache.org> wrote: > Steve, > > On Tue, Jun 14, 2011 at 4:51 AM, Steve Loughran <ste...@apache.org> wrote: >> I've added more on the limitations of the current process (not synchronised >> releases, not enough automated testing on multiple-host clusters), and on a > > actually there's a stand-alone Hudson job in Apache > https://builds.apache.org/hudson/view/G-L/view/Hadoop/job/Hadoop-22-cluster-test/ > which does exactly that: testing 0.22 against a cluster deployed from every > build (deployed automatically as well).
To add to that -- part of the allure of Bigtop (or any software management stack technology for that matter) is that it gives you a mechanism for building and qualifying software stacks, but doesn't dictate a policy. In fact, it is perfectly reasonable to have quite a few different stacks serving different customer needs as long as they are reproducible and quantifiable (of course, as a community we really would benefit from less stacks since it makes things like interoperability way easier to manage). It is a bit similar to how different Linux distros manage the upstream releases and cater to different use cases (although the great unicorn of a stable AND cutting edge distro is yet to be spotted). > The job is using a different system harness https://github.com/c0s/v4stack > though, because the framework proposed here was pulled out from public access > (e.g. from github) for whatever reason. Technically, speaking the commnunity > could've benefit from a large set of automated cluster tests for months now. I > guess late is better than never ;) Manuscripts don't burn, you know ;-) Thanks, Roman. --------------------------------------------------------------------- To unsubscribe, e-mail: general-unsubscr...@incubator.apache.org For additional commands, e-mail: general-h...@incubator.apache.org