As discussed in a support case there is a bug in the Retry Time Cycle causing it not always to work.

The reason is that the implementation check for the revision - but this is not in all circumstances "2":
// check if this is jobs' first execution, then change default retries
if (job.getRevision() == 2)

{ job.setRetries(durationHelper.getTimes()); }

We agreed to refactor the whole functionality somewhere soon - but for the moment I change the behavior to check for an exception entered in the Job. No exception = no retry, this is why this is safe.

Expect a pull request.