I've setup snmpv2 for communicating with Zenoss server. It's working. But now I have some question. 1. Is SNMPv2 client broadcasting its data all the time? Is there any way to prevent that? Through configuration file? Can I use SNMP configuration file to limit it broadcast to certain ip only? 2. If I'm trying to push my snmp info to a server with dynamic ip, is it possible? How should it know the server? By domain name? Use SNMPv3? Anyone mind to guide me?
Short version: SNMP doesn't broadcast.
SNMP can mean two things. If you're talking about sending traps, then it sends it to a specific address that you would have to set up.
But normally what people are talking about is the box responding to a server asking for certain SNMP info. You can limit who can query this info in two ways:
As Chris says, "SNMP" works in 2 ways. TRAPs are generated by a box to tell your Zenoss server about a problem. This just happens once when the problem occurs. More pervasively, an SNMP "server" such as Zenoss will periodically poll for SNMP information from SNMP "clients".
The default with Zenoss is to assume that every discovered device supports SNMP and a fairly small, standard set of SNMP requests will be issued to each box every 5 minutes. That said, you control which boxes are "discovered". You control the 5-minute polling interval. You control whether a discovered device "supports" SNMP (there is a device class of /Ping that is good for devices that don't support SNMP). You control the SNMP parameters used for classes of devices and/or individual devices - like SNMP version, community name, SNMP V3 parameters if you use V3.
Zenoss will use Domain Name Server (DNS) if it is available to the Zenoss server. If you have a DNS that understands your dynamic IP, then make sure that when you discover the device that you give the DNS name for it. In general, it isn't a good idea to have important servers with IP addresses as it does make them harder to monitor.