I went ahead and purged those server restrictions. The way the upstreamness is set up it actually does hadoop2 first and shouldn't build hadoop1 unless it's a success, I think.
At this point, it looks like builds are failing in the infra tests. Go figure. On Mon, Sep 9, 2013 at 2:15 PM, Billie Rinaldi <[email protected]>wrote: > Might as well add those servers back in, see if they're healthier now. > > I also noticed and was wondering about the upstream build. I didn't > configure it that way, so I don't know if someone did it by accident or > purposely. Perhaps it's because there's no point in trying to build with > hadoop 2 if the regular hadoop build fails? > > > On Mon, Sep 9, 2013 at 10:52 AM, John Vines <[email protected]> wrote: > >> So, the trunk builds had been failing due to rat failures from the >> examples >> change it seems. I went ahead and change the behavior to purge the >> workspace before it pulls it, and so far it seems to be working. >> >> However, I did notice a few things that I might have missed or might >> warrant a change- >> >> 1. We skip ubuntu5 for regular trunk and ubuntu4 for hadoop2. I know we >> started skipping certain boxes because they were consistently misbehaving. >> Is this still the case or should we go ahead and remove those checks >> >> 2. The hadoop2 build is upstream of the regular hadoop build. Is there a >> reason for this? >> > >
