Re: [dhcwg] dhcpv6-24: vendor-specific issues

Thomas Narten <> Wed, 15 May 2002 20:34 UTC

Received: from ( [] (may be forged)) by (8.9.1a/8.9.1a) with ESMTP id QAA21630 for <>; Wed, 15 May 2002 16:34:49 -0400 (EDT)
Received: (from daemon@localhost) by (8.9.1a/8.9.1) id QAA29922 for; Wed, 15 May 2002 16:35:04 -0400 (EDT)
Received: from (localhost []) by (8.9.1a/8.9.1) with ESMTP id QAA29870; Wed, 15 May 2002 16:33:31 -0400 (EDT)
Received: from (odin []) by (8.9.1a/8.9.1) with ESMTP id QAA29841 for <>; Wed, 15 May 2002 16:33:29 -0400 (EDT)
Received: from ( []) by (8.9.1a/8.9.1a) with ESMTP id QAA21575 for <>; Wed, 15 May 2002 16:33:13 -0400 (EDT)
Received: from (narten@localhost) by (8.11.6/8.11.6) with ESMTP id g4FKVtm04658; Wed, 15 May 2002 16:31:55 -0400
Message-Id: <>
To: Ralph Droms <>
Subject: Re: [dhcwg] dhcpv6-24: vendor-specific issues
In-Reply-To: Message from Ralph Droms <> of "Wed, 15 May 2002 13:23:33 EDT." <>
Date: Wed, 15 May 2002 16:31:55 -0400
From: Thomas Narten <>
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: <>

Ralph Droms <> writes:

> At 11:14 AM 5/8/2002 -0400, Thomas Narten wrote:
> > >    Servers can interpret the meanings of multiple class specifications
> > >    in an implementation dependent or configuration dependent manner, and
> > >    so the use of multiple classes by a DHCP client should be based on
> > >    the specific server implementation and configuration which will be
> > >    used to process that User class option.
> >
> >This sure reads badly in the sense that it smells like
> >interoperability will not result. But I'm not sure what the intent of
> >the wording is here... Could someone clarify?

> The idea is to support additional flexibility and precision in identifying 
> clients in some deployments.  This situation anticipates that clients will 
> be configured to send additional classing information and servers will have 
> policies to select configuration information based on those 
> classes.  Because both the clients and server have been "managed" for this 
> deployment, they will interoperate.  A client that moves and tries to use a 
> server that does not have policies for the classes specified by the client, 
> or a client that does not supply the class specifications will get 
> configuration information that is not selected on the basis of the client 
> classes.

I think my issue is more with the wording than the intent.

TO be sure I understand, the User Class option is one that is
configured by the end user. I.e., to say "accounting", "engineering",
etc. In that case, the server processing shouldn't be "implementation
specific" or "vendor specific".

Another confusing part. The text is talking about multiple occurances
of the option. How this is dealt with shouldn't be configuration
specific... We don't get interoperability that way... 

On rereading this, one thing that is missing is discussion about how
the client sets the value. Is it really "opaque data"? Shouldn't  this
be a utf-8 string (for ease of configuration)?

Also, the server should treat them as opaque values in conjunction
with its configuration-specific data. This is not stated.

Does this help clarify my concern?
> > > 22.18. Vendor-specific Information option
> > >
> > >    This option is used by clients and servers to exchange
> > >    vendor-specific information.  The definition of this information is
> > >    vendor specific.  The vendor is indicated in the enterprise-number
> > >    field.  Clients that do not receive desired vendor-specific
> > >    information SHOULD make an attempt to operate without it, although
> > >    they may do so (and announce they are doing so) in a degraded mode.
> >
> >This also reads badly and against interoperabilty. DHC clients should
> >never depend on vendor-specific options in order to function
> >correctly.

> This text is essentially copied directly from section 8.4 of RFC2132 
> (which, of course, doesn't necessarily make it a good idea).  Perhaps the 
> differentiation should be "enhanced operation" with the vendor-specific 
> information versus "functioning operation" without?

yes. It should not be implied that things "may not work well" if the
server doesn't support the option. Rather, when it is supported, there
maybe enhanced operation.


dhcwg mailing list