On Wed, Feb 3, 2021 at 11:22 PM Jakob Heitz (jheitz) <jhe...@cisco.com>
wrote:

> <snip RFC4271>
>
>    Syntactic correctness means that the BGP Identifier field represents
>    a valid unicast IP host address.
>
> </snip>
>
>
>
>      Gyan> I do see that verbiage in section 6.2
>
>
>
>    If the BGP Identifier field of the OPEN message is syntactically
>
>    incorrect, then the Error Subcode MUST be set to Bad BGP Identifier.
>
>    Syntactic correctness means that the BGP Identifier field represents
>
>    a valid unicast IP host address.
>
>
>
> BGP with IGP call back NH tracker checks the NH but how does BGP code 
> validate the RIB that the router-id is a connected loopback but
>
> and also advertised by IGP.  I have not tried it but if you set a bogus 
> router-id would all the BGP peers go down.
>
> I will try that in the lab.
>
>
>
> IOS-XR does not have this check. Nothing breaks by violating this rule.
> IOS-XR implements RFC 6286.
>
> I think you'll be hard pressed to find a router that checks this.
>
>  Gyan> Agreed.  That is exactly what I thought.  I was going to try on IOS
> XR but you saved me some time and results as I expected.  I will try test
> RFC 6286 on XR.  Have you tried doing IPv6 only peers on XR and with BGP
> identifier set unique to 4 octet IP address and see if that works.  I am
> guessing it would work as XR does not have the check.
>

    I  am not crazy about the RFC 6286 AS wide BGP identifier with 4 octet
unsigned non zero integer.  Most operators are more comfortable having
unique 4 octet IP address as BGP identifier and I think would much rather
do that as long as the check does not exist as even with enabling RFC 6286
and having AS wide unique identifier seems odd and scary to me as normally
the BGP identifier must always be unique within the domain or breaks BGP.

dual stack edge over v6 core RFC 5565 is becoming more common for operators
every day with SRv6 push and thus IPv6 only routers and running into this
issue where now you have to enable RFC 6286.

I am thinking it maybe well worthwhile to write a draft that updates RFC
4271 check as vendors don’t follow it anyway and as we all know not
checking is not going to break anything and making so that for IPv6 only
routers such as in a SRv6 core that the BGP identifier can remain a 4 octet
IP and then operators now could keep the same unique BGP identifier IP you
had on the router before you ripped it out of the core when transitioned to
SRv6.

> Regards,
>
> Jakob.
>
>
>
-- 

<http://www.verizon.com/>

*Gyan Mishra*

*Network Solutions A**rchitect *



*M 301 502-134713101 Columbia Pike *Silver Spring, MD
_______________________________________________
BESS mailing list
BESS@ietf.org
https://www.ietf.org/mailman/listinfo/bess

Reply via email to