In this section:
"The cluster manager will automatically fail over jobs that were claimed by a node, if that node goes down"
We should give more detail as the actual behaviour is:
- if instance A claims a job but does not start it, then goes down, instance B can reclaim that job.
- if instance A has already started a job, and then goes down, that particular job will never be completed. Instance B will claim the next "iteration" of that job