When using a cache that is distributed and a hotrod bulk operation with a count the entries or keys are not fully iterated upon, which requires the iterator to be manually closed or else this will consume a thread per request until the server runs out of resources.