Re: [mpls] [MPLS] draft-ietf-mpls-ldp-ipv6

[Vishwas Manral <[e-mail]>]

Hi Lizhong,

Sorry for the delay in replying to your email.

On Wed, Dec 15, 2010 at 6:20 AM,  <[e-mail]> wrote:
>
> Hi Rajiv, Vishwas, all,
> There are three issues in the draft as I mentioned before.
> 1. If there are two LSRs with two independent link, one is for IPv4, the
> other is for IPv6. Let's say LSR1 and LSR2 is connected by IPv4 enabled
> link1 and IPv6 enabled link2. It is very likely that LSR1 will receive the
> IPv4 hello first, and choose the IPv4 transport address, and LSR2 will
> receive the IPv6 hello first, and choose the IPv6 transport address, then
> the TCP connection can not be setup.
This is not correct.
        - An LSR should prefer the TCP connection over IPv6 for a new LDP
        session with a remote LSR, if they attempted two TCP
        connections using IPv4 and IPv6 transports simultaneously.

IPv6 would be prefered and TCP transport connection will be setup. If
an IPv4 connection is in process, it does not mean an IPv6 session
cannot be initiated.

> We have not work out a perfect way. One way is to assign a active/passive
> role for the LSR, and only the active role can choose IPv4 or IPv6 transport
> address, and the active role will be also the TCP session active role. The
> active/passive role can be choosed based on LDP-ID which is different with
> currently defined RFC5036.
We can have active/ passive role, however that is uselessly
overloading current protocol.

> 2. It should be possible to allow IPv6 hello to establish IPv4 session.
> Because IPv4 session can also be used TO setup IPv6 IP-prefix LSP. In that
> case, the IPv4 session should choose another route to reach peer node. That
> means IPv6 hello will carry IPv4 transport address, and the address will be
> choosed to established the session.
We can allow that but it does not help in anyway.

> 3. Section 6.1: An LSR should include only the transport address whose
> address family is the same as that of the IP packet carrying Hello. But in
> RFC5036, it says: An LSR MUST advertise the same transport address in all
> Hellos that advertise the same label space. The two requirements will be
> confict in the case of two LSR connected with one IPv6 link and one IPv4
> link.
Yes, this is a good point and we need to clarify that further in this draft.

> So we still have to solve these problems for LDPv6, we could discuss in this
> thread.
Ok.

Thanks,
Vishwas

> Thanks!
> Lizhong
>
>
> --------------------------------------------------------
> ZTE Information Security Notice: The information contained in this mail is solely property of the sender's organization. This mail communication is confidential. Recipients named above are obligated to maintain secrecy and are not permitted to disclose the contents of this communication to others.
> This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify the originator of the message. Any views expressed in this message are those of the individual sender.
> This message has been scanned for viruses and Spam by ZTE Anti-Spam system.
>
> _______________________________________________
> mpls mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/mpls
>
>