Re: [radext] WGLC for draft-ietf-radext-bigger-packets-01

[Sam Hartman <[e-mail]>]

>>>>> "Alan" == Alan DeKok <[e-mail]> writes:


    Alan>   That might be hard work, if the client is sending 100's of
    Alan> 1000's of packets.  The packets will be sorted by socket, IPs,
    Alan> ports, etc.  Not by size.

You know the socket and port.
You're trying to sort between one accounting request and one
access-request with id 1.
That seems relatively easy.
However:

    Alan>   I'd prefer to not overload Access-Reject or
    Alan> Accounting-Response.  Those are for *normal* situations.
    Alan> i.e. they do signaling for users, not for protocol errors.

Agreed.
Also, they get proxied, and we don't want that here (I don't think)

    Alan>   That means we need a new packet code.

    Alan>   TBH, the simplest thing would be to include the original
    Alan> packet code in a new attribute.  That way the peer receiving
    Alan> the packet knows how to find the original.  It can match code,
    Alan> src/dst IP, src/dst port.

    Alan>   And change the name of the packet code. A generic name is
    Alan> much prettier than a specific one.

    Alan>   i.e.:

    Alan> Packet code == Protocol-Error Attribute-TBD = Access-Request
    Alan> Error-Cause = Packet-Too-Big Response-Length = 8192

    Alan>   The packet signing rules should be the same as for
    Alan> Access-Accept, I think.  i.e. copy the original vector to the
    Alan> response vector and sign.  Even if the original was
    Alan> Accounting-Request.

    Alan>   I think that's simple, has minimal changes to the document,
    Alan> and has minimal changes to existing implementations.

I'm happy with that solutionn.

Others?