In cases where a host is in the TIMEOUT state, the dispatcher keeps issuing STOP commands to the entire detector without bothering to check if these commands are getting acknowledged. It would be better to issue STOPs in a more intelligent way where the dispatcher doesn't keep adding unack'd STOP on top of unack'd STOP.