Wonderful detective work there! I hope you don’t mind but I flagged your post as the solution. Firewall rules can be goofy - I had to dump COMODO on one of my machines because it absolutely killed my unRAID interface for some odd reason!
By default, I believe the --retry-count is set to 5 and the --retry-delay is set to 10 seconds - so you get just under 1 minute’s worth of retries before the job errors out. Once that happens it gets put back in the queue for the next scheduled run time.
Note that if you only have a single job then using the above two parameters should work well for you. Just be aware that if the destination NEVER becomes available then the total duration of --retry-count * --retry-delay would be treated as a single job run, so if use --retry-count=24 and --retry-delay=60MIN then a potential 2nd job would never get a chance to run.
Am I correct in assuming you’re looking for something similar to what I proposed here?