I am guessing that I am missing something.

I created two new events (node_down_emailB and node_up_emailB).

I copied the Node Down and Node Up event processing policies and replaced with the events above.  I also added new email actions.

I then went to the node status DCI and added a new threshold using the above events with 1 sample < 0.  (I already have a sys_node_down event that fires an email to a certain group and this part works).  I then checked process all thresholds.

I reboot the server and userA gets an email (belongs to the sys_node_down event) but userB does not get an email (belongs to the node_down_emailB event).

wrong

all events is standart. you need sys_node_down.

1. create two server action mail_a mail_b
2. go to event processing policy.
3. copy policy №1 (Show alarm when node is down)
4. go to properties for this copied policy:
    a) add to source object your server_B
    b) add to server action your mail_b action
   also you can use filtering script instead source objects.

I created two new events (node_down_emailB and node_up_emailB).

I copied the Node Down and Node Up event processing policies and replaced with the events above.  I also added new email actions.

I then went to the node status DCI and added a new threshold using the above events with 1 sample < 0.  (I already have a sys_node_down event that fires an email to a certain group and this part works).  I then checked process all thresholds.

I reboot the server and userA gets an email (belongs to the sys_node_down event) but userB does not get an email (belongs to the node_down_emailB event).

Create another event processing with filter on server_B