I have been experiencing an issue where I have a job that errors and immediately restarts. A few moments later, attempts 3 and 4 happen, again immediately next to each other. Then, 5 and 6 happen together.
Cutting a long story short, the problem was discussed in issue #63 which I was unable to re-open, and the solution suggested therein fixed my problem. Here is a link to my fix:
I will put together a pull request in case it gets accepted, but I'm logging the issue here for anyone who searches for the problem.