notifier can be started manually. Check out the man page for notifier. You should put it in the Cluster Service Group because it is a special purpose group and will survive when onthers would stop, for example it cannot be autodisabled and will failover even when you perform an hastop -local on the system where CSG is running. If you do not configure the notifier in a service group, you could start it manually, I have no idea why you would want to do that because it would not be highly available. Since "had" maintains replicated message queues, any message received by had will be sent to the notifier regardless of which system is running notifier. Only one notifier can run per system (it will fault if another copy has already connected to the port on had). Any reason you do not want it in CSG?