[SNMP4J] Engine time synchronization question

[Kevin J. Schmidt]

It would appear that the CommandResponder interface does indeed need
to be implemented, i.e. you provide your own processPdu method. This
is the only way I was able to respond to engine ID discovery requests.
It seems you also have to do something like:

snmpObject.setLocalEngine(authoritativeEngineID.getValue(),0,0);

if you want your own engine ID to be discovered, or call

MPv3.createLocalEngineID()

if you want a random one created. But this is speculation on my part.

Can someone in the know comment on this?

Thanks,


On Mon, 28 Mar 2005 10:42:03 -0600, Ronald.Madsen at utstar.com
<Ronald.Madsen at utstar.com> wrote:
> 
> I have never written an application to send SNMPv3 traps - mainly because I
> don't know of any "Manager" ready to accept them.  But here is my
> understanding of the situation...
> 
> From a SNMPv3 USM point of view, your application will be considered an
> "authoritative" SNMP Engine.  As such, the application is required to
> response to snmpEngine discovery queries - BUT - I believe the SNMP4J stack
> takes care of all of this for you.
> 
> I'm not real clear on port usage when just sending traps.  You might need
> to use port 161 when sending the traps - allowing the manager application
> to do the engine discovery as necessary.  If the manager application forces
> engine discovery to port 161 - then you will have to "listen" on 161.
> 
> Note that all engine discovery transactions are handled by SNMP4J.  You
> application will never see the queries.  I do not believe you need to
> actually implement the CommandResponder interface.
> 
> Best regards,
> Ron
> 
> "Kevin J. Schmidt" <kjschmidt at gmail.com> on 03/28/2005 09:37:49 AM
> 
> Please respond to "Kevin J. Schmidt" <kjschmidt at gmail.com>
> 
> To:    Ronald Madsen/UTStarcom at UTStarcom
> cc:    snmp4j at agentpp.org
> Subject:    Re: [SNMP4J] Engine time synchronization question
> 
> Hi,
> 
> This brings up a good question for me. I'm writing an application that
> can send SNMP v1, v2c, and v3 traps/notifications. It's meant to
> interface with our product and provide external alerting capabilities
> to OpenView, et al. I wasn't planning on developing agent-like
> capabilities, i.e. have a process listen on port 162 and respond to
> SNMP requests. Can I get away with doing this or do I need to be able
> to respond to requests for engineID, boots, and time?
> 
> On Wed, 23 Mar 2005 16:24:04 -0600, Ronald.Madsen at utstar.com
> <Ronald.Madsen at utstar.com> wrote:
> > I'm experimenting writing a SNMPv3 Agent using SNMP4j.
> >
> > My "Manager" sends a normal Request to discover the authoritative
> engineID
> > - SNMP4j responds with a REPORT with its engineID just fine.
> >
> > The Manager then sends an authorative Request to discover the engineBoots
> > and engineTime - SNMP4j responds with a REPORT with both engineBoots and
> > engineTime set to zero.
> >
> > Subsequent transations eventually Report "not in time window".
> >
> > I searched the SNMP4j archives and did not find a similar discussion.
> >
> > I may very well be using SNMP4j incorrectly.
> >
> > Does anyone have a small snippet fo code that properly initializes Snmp,
> > USM, etc.  correctly?
> >
> > Thanks,
> >
> > _______________________________________________
> > SNMP4J mailing list
> > SNMP4J at agentpp.org
> > http://lists.agentpp.org/mailman/listinfo/snmp4j
> >
> 
> --
> Kevin J. Schmidt
>  <kjschmidt at gmail.com>
> 
> 


-- 
Kevin J. Schmidt
<kjschmidt at gmail.com>