net-snmp: how should I approach the project of making a bilingual snmpd?

Question

I apologize that this question comes from the uninformed, huddled masses: I've been away from net-snmp for three years and I've missed all of the developments.

I have to deliver a bilingual v2c/v3 snmpd for use in an embedded Linux system.

I expect to use the superb net-snmp sources. What is the right approach to marshalling these sources, where "right" == straightforward, uncomplicated, vanilla, and "it just works"?

Thanks so much, everybody. And I apologize again for my out-of-dateness.

Edit: Why do I need this input? I have a lot of experience with snmp v2c in general; some experience with net-snmp; and little experience with snmp v3. Because I'm new to v3, I'm looking for a cookbook approach from someone who's done it so I can avoid the many undocumented pitfalls and hurdles that my experience tells me lie in the path of anyone building an agent from net-snmp sources.

Answer 1

First, Net-SNMP is already tri-lingual under the hood. If you configure it to accept v1, v2c and v3 it'll happily accept and respond to all 3 protocol versions without changing a thing.

Second, to get snmpv1/2c working, all you need in your snmpd.conf file is:

rocommunity COMMUNITYNAME

Where, COMMUNITYNAME is the super-secret insecure community/password you want to use.

For SNMPv3, I'm assuming you want to use SNMPv3 with USM. I suggest you start by reading the Net-SNMP tutorial on Securing SNMP traffic and go from there.

Note also, you can run snmpconf -g basic_setup to get more help with the above as well as with other options.