By design, config-rotator fails a build if a new build is schedules and there is no new baselines.
This was a decision to signal the user that this it is out of context to build with the same configuration twice (without using the reset option).

It happen when users hits Build now and there is no new baselines, but it happens more often when slaves temporarily goes offline, and Jenkins then schedules a new build to get at workspace.

Improvement will be to use nothing to do when a build finds no new baselines.