According to the documentation, setting max_retries to 0 should allow jobs to be retried infinitely. However, upon reviewing the code in runner.py, it appears that jobs with max_retries=0 are mistakenly set to 'JobFoundDead' instead of being requeued.

This seems to be due to the logic in the _query_requeue_dead_jobs method, where the condition for exceeding max_retries does not properly account for the 0 value as "infinite retries." As a result, jobs are marked as failed even though they should be requeued indefinitely.

Expected behavior:
Jobs with max_retries=0 should be requeued without limit when found dead, in line with the documentation.

Actual behavior:
Jobs with max_retries=0 are being marked as 'JobFoundDead' and not requeued.

Code reference:
See runner.py, method _query_requeue_dead_jobs.