I have a Mesos cluster that handles spark jobs.
At some points there are n active frameworks, but some get allocated 0 resources.
Something li
