Okay.. I found the reason why my job getting moved so early.. It often get moved with the error "maxAttemps[..]".

=> Although I set the timeout inside the job to 900sec, my worker retried the job after the default time. That results in the described behaviour. The job get finished, but appeared already as failed.

The solution for me is to set the worker timeout parameter to 0sec (--timeout=0) in that way I can choose the timeout in my jobs and now it works as intended :)