Throttling Semaphores cause wrong retry count

Hi, we're facing an issue when using Throttling Semaphores

Some of our jobs should try 4 times but getting blocked by the semaphore consume some retries and actually run just one time

here a dashboard screenshot for reference

Any suggestions?

Start by reproducing the reported behavior with Throttling Semaphores and jobs configured for four attempts, using the dashboard screenshot as the observed result. Trace when semaphore blocking and retry counting occur, then verify that blocked executions do not incorrectly consume the configured retry count.

Written by the indexing model from the issue text.