With several UNIX servers, the RO string keeps getting ‘out of sync’ between Orion NPM and the server. All of the sudden the string will no longer verify. One of the UNIX SA’s will have to change it back to public, and then I will have to change NPM to public, then the SA will change it back to our standard string, and I will change NPM (again) to match. All is fine for a while, but eventually the problem will occur again.
We become aware of this happening by receiving alerts that these servers are constantly rebooting, even though they aren't (no pages are sent, just log entries occur). So far the only pattern seems to be that these are Solaris 9 servers, but not all of our Solaris 9 servers exhibit this symptom. Also, these systems seem to only respond as SNMP v1 (not 2c) but we’re not sure why. Again, this is not true for all of our Solaris 9 servers.
The SA’s have tried everything they can think of, but can’t explain this behavior from their end any better than I came from the Orion end. Any else had this occur, or have any ideas?