[SNMP4J] Do I need a ProxyForwarder implementation?

Alan McSherry alan.mcsherry at corvil.com
Fri Apr 6 16:51:17 CEST 2007


Hi Frank,

Thanks for the reply. 

Do I understand that I could implement a simple ProxyForwarder which
happily forwards all requests to and returns all responses from agent1
based on the community string "agent1" (for example) and uses agent2 to
service requests made with the community string "agent2"? 

I think that would solve my problem nicely. 

Alan.

On Fri, 2007-04-06 at 15:07 +0200, Frank Fock wrote:
> Hi Alan,
> 
> You will not be able to proxy requests based on the OID
> and still be SNMP standard compliant when your systems
> allow SET requests, because SNMP does not implement a
> two-phase-commit (2PC).
> 
> The standard compliant ProxyForwarder of SNMP4J
> forwards requests based on community, context, and/or
> context engine ID.
> 
> If you still want to forward subrequests (!= messages)
> based on OIDs the you will have to implement something
> that similar like the SNMP4J-AgentX API but uses SNMP
> as protocol.
> 
> Please verify carefully whether the AdventNet solution
> is standard conform and works in a real life scenario.
> 
> Best regards,
> Frank
> 
> Alan McSherry wrote:
> > Hi all,
> > 
> > I have a situation whereby two independent vendors are amalgamating
> > their products onto one hardware platform. 
> > 
> > Each has it's own (non AgentX master/sub) SNMP agent. 
> > 
> > I would like to run a proxy agent on the default SNMP port and run each
> > of the other agents on it's own port so that when an SNMP request
> > arrives at the standard port I can switch on the OID and forward the
> > request to the relevant SNMP agent. 
> > 
> > It seems that this is a job for a ProxyForwarder? 
> > 
> > Can anyone offer any pointers to examples of ProxyForwarder
> > implementations? (I've seen one in a previous months posting)
> > 
> > Does anyone know of an existing (commercial or otherwise) solution to
> > this dilemma? I'm about to check out AventNETSnmpUtilities but that's
> > the only option I've found so far. 
> > 
> > Any help greatly appreciated.
> > 
> > Alan.
> > 
> > 
> > _______________________________________________
> > SNMP4J mailing list
> > SNMP4J at agentpp.org
> > http://lists.agentpp.org/mailman/listinfo/snmp4j
> 




More information about the SNMP4J mailing list