Hey Derek- good question. Because Render’s services have zero downtime deploys, there’s no guarantee that the old worker is completely shut down before the new worker starts up. Your use case makes a lot of sense, though- I’ve created a feature request for this that you can vote on, and you’ll be notified when we have any updates.
As a workaround in the meantime, if you add a 1db disk to your background worker, zero downtime deploy will be disabled, which should guarantee that an old version of your deploy is not running.