Re: [Isms] Ismssnmp: URI scheme transport extensions

[Wes Hardaker <wjhns1 at hardakers.net>]

>>>>> On Mon, 27 Jul 2009 10:47:36 +0200, "David Harrington" <ietfdbh at comcast.net> said:

DH> Why do we need to specify the transport in the URI? We are already
DH> specifying port 5161, which is snmp/ssh.

You're presuming TCP there.  You're also presuming a standard port is
being used and you can determine from that what is running on it, which
isn't a safe assumption.

DH> Why do we need to specify the securityModel in the URI? It is already
DH> specified in the SNMP header.

It's configuring a client.  Consider a URI the equivalent of a
TARGET-MIB entry that specifies how the client should connect.

The point of the URI stuff (I don't use it either, but Juergen will
correct me if I'm wrong) is to be able to do the equivalent of:

  snmpget snmp://tester5 at example.com:5161;transport=ssh sysContact.0

No other command line options needed because it's all in the URI string.

[on a side note, I *did* write a plugin to KDE once to make their web
browser do something almost exactly like this...  Connect via SNMP to a
remote agent given a URI and display the results.]

DH> In fact, wouldn't we need to specify everything that could be
DH> specified in the set of tables in the notification-mib?

I think so, yes [but really it's a target].

-- 
Wes Hardaker
Cobham Analytic Solutions