One of our producers just pointed out what sounds like potential bug. Some jobs were submitted with our default Task Error Limit of 4, and he overrode that limit on the jobs after the fact to set the limit to 1. However, according to him, tasks that were rendering already when the limit was changed did not respect the new value of the job-level override, and would thus be re-queued at least once after hitting an error.
I wanted to run it by you guys to get confirmation on what the existing behavior is (or should be), and to see if there’s any possibility this could be a bug.
Thanks.