SNMP and the UTM-1 Edge Boxes

Hello there. I am using Cacti to monitor several devices on my network and when we decided to install an Edge box at each of our remote sites to establish site-to-site VPNs, I thought I would also look into monitoring those in cacti as well. I'm able to monitor interface traffic, CPU usage and uptime on the Edge devices using standard templates that come with Cacti. I was also interested in monitoring the number of used nodes vs the number of allowed nodes on each box so I used snmputil to walk the OIDs on the box. I discovered that the .1.3.6.1.4.1.6983.1.3.4 OID shows the number of used nodes so I set up the required things in Cacti to graph it. When I was done, I created the graphs for each of our 28 Edge boxes. Oddly enough, all of them worked except for 2. When I used snmputil to query the OID on those two boxes, I discovered that it did not even exist in these 2 boxes. In fact, the entire .1.3.6.1.4.1.6983 tree was not even there. I ensured that all of my devices were running the latest version of the firmware and also that SNMP was configured identically in each one.

To make matters even crazier, after setting up these "node monitor" graphs on all of my Edge boxes in Cacti, I noticed that some of them stopped graphing.  All other graphs for the affected boxes are working fine, it's just the monitored nodes one that decides to stop working.  When I use snmputil to again query the OID, .1.3.6.1.4.1.6983 was gone!  

I guess I don't understand enough about SNMP to know why an OID would vanish. Has anyone ever seen anything like this?

I appreciate any insight.

-Chris