Due to some reason (implementation problem, environment problem, etc), there's a case where a handler sometimes blocks indefinitely. In this case, an administrator might want to create a new event loop thread and migrate all channels (except the blocked one) to the new thread to keep the server working.