There isn’t an easy way to disable the retry handler logic. This system is difficult to disable and no one has asked for this before. About the easiest we can think of is a BeanPostProcessor that wraps any RetryClusterService instance with a proxy that just noOps any call to the #setLifecycleRequirementMet method. That way, the contract to start the retry services is never met and the retry system never gets going after application startup.