This is a strange, strange bug - the polling log shows it polled on ubuntu5, and even if master had executors and had the ubuntu label, that should still mean the subsequent build would run on ubuntu5, since that's where the previous build ran, and Jenkins will always opt to run where the previous build ran if possible. I wonder if there might have been a race condition, with ubuntu5 going offline due to inactivity just as the build kicked off by the polling tried to start? Not sure.
If someone can come up with a way to reliably reproduce this, I can dive in deeper. A. On Sun, Oct 23, 2011 at 2:53 AM, Niklas Gustavsson <nik...@protocol7.com>wrote: > On Sat, Oct 22, 2011 at 11:19 PM, Stack <st...@duboce.net> wrote: > > Should we list machines we would build on Niklas? Currently we have > 'ubuntu' > > That should be just fine, this is a bug in Jenkins. > > /niklas >
