The HP has two policies for OM alerts operators of failures: FwdMsgAsEMail and FwdMsgAsHtmlEMail. The policy message is sent when a particular severity. What need to modify them or create new to have sent a message when an alert with a specific node or group nodes (for example from domain controllers)?
You need to modify or create rules in either of the two policy so it will match your condition. Out of box the policy is set to match any 'critical' message and send out an email. You can modify it to add the node name (the policy editor will list the available fields under the drop down list for TargetInstance) and that will narrow it down.
Check the OMW Online Help topic for the policy to see more details as well.
For the NodeName the value wil be the UniqueID of the node (it is the long GUID for the node that you can find by looking at the prorties of the node). This way the rule will match for messages only from that node and then trigger an email.
You should not modify the check for "ThisPreventsLoops" - this condition is there to prevent from the chaining where the fact that FwdMsgAsEmail policy send a message and it matched by one of the rules in the FwdMsgAsEmail policy and hence triggers another message (and as you can see a chain reaction).
Yes you are right for the human readable name it is the PrimaryNodeName. You can use WBEMTEST tool (microsoft tool) and connect to "root\hewlettpackard\openview\data" and run the query
select * from OV_Message
and check the details of the specific message to see the actual name/value pairs.
If you trying to setup for message from specific nodegroup - i don't think you can for OV_Message does not hold that value and so you will not be able to do it at the 'coarse' level. You will only have to do it individual node level.