We are getting SNMP agent down alerts about once every 24 hours, and the alert clears in about a minute or less. We are getting these for several servers, but the times are staggered. Is it possible tochange this to trigger the alert if it occurs x times, and if so how?
If the times are well staggered then it suggests that the problem is a timeout at the agent, rather than something happening at your Zenoss server. Is there any pattern to these events with regard to the type of device or the network they are on?
One thing you might look at are the zSnmpTimeout and zSnmpTries parameters whose defaults are respectively, 2.5 and 2. I would try upping the timeout on an offending device and see if the problem goes away.
You find these parameters from Configuration Properties for a device.