Re: [v6ops] RFC 2464 - MTU >1500

[Matt Crawford <[e-mail]>]

(If this gets bounced from v6ops for my non-membership in the list, please repost.)

On Feb 7, 2011, at 3:18 PM, Daniel Roesen wrote:

> On Fri, Feb 04, 2011 at 08:42:49AM -0500, Thomas Narten wrote:
>> The MTU option in an RA can be used to override the default, in cases
>> where the default value is somehow not optimal.
> 
> Except that the IPoverEthernet RFC says to ignore MTU >1500 when received
> in RAs. So on Ethernet, RA MTU option may only decrease MTU, not increase
> compared to default.
> 
> The reasoning behind that is beyond me.
> 
> Looking at
> https://datatracker.ietf.org/doc/draft-ietf-ipngwg-trans-ethernet/history/,
> that paragraph was never substantially changed in the I-D process, just
> the paragraph referring to DHCP="manual" added.
> 
> Matt, any chance of clarification?


The text was nearly the same in RFC 1972. I have the nroff source of all drafts of both the i-d's, but I don't have all my email from 15 years ago so I cannot say who raised and/or endorsed this point. 

Reaching back almost six years, there was a relevant discussion on this question. I think the same text answers here ...




Begin forwarded message:

> From: Hugh LaMaster <[e-mail]>
> Date: July 20, 2005 7:15:52 PM CDT
> To: Matt Crawford <[e-mail]>
> Subject: IPv6 and Jumbo Frames on Ethernet
> Reply-To: Hugh LaMaster <[e-mail]>
> 
> Hi,
> 
> There is an ongoing discussion on the "[e-mail]"
> mailing list wrt jumbo frames.  Is it your opinion that
> RFC2464 prohibits (that is, if something is supposed
> to be strictly in compliance with RFC2464) the IPv6
> MTU from being greater than 1500 Bytes?  That is one
> interpretation.  I don't read it that way myself, but,
> apparently whoever implemented the FreeBSD IPv6 code
> interprets it that way also-- that the RFC prohibits
> IPv6 MTU > 1500.
> 
> Regards,
> Hugh LaMaster
> NASA ARC/NAS/NREN
> 


Begin forwarded message:

> From: Matt Crawford <[e-mail]>
> Date: July 20, 2005 8:52:48 PM CDT
> To: Hugh LaMaster <[e-mail]>
> Subject: Re: IPv6 and Jumbo Frames on Ethernet
> 
>> There is an ongoing discussion on the "[e-mail]"
>> mailing list wrt jumbo frames.  Is it your opinion that
>> RFC2464 prohibits (that is, if something is supposed
>> to be strictly in compliance with RFC2464) the IPv6
>> MTU from being greater than 1500 Bytes?  That is one
>> interpretation.  I don't read it that way myself, but,
>> apparently whoever implemented the FreeBSD IPv6 code
>> interprets it that way also-- that the RFC prohibits
>> IPv6 MTU > 1500.
> 
> Hm.  It is only supposed to mean that Router Advertisements cannot increase the MTU above whatever it was otherwise believed to be, and that in the absence of configuration information (including DHCP), it is presumed to be 1500.
> 
> I'd have to agree that it could have been a bit clearer.
> 
>                Matt Crawford   <[e-mail]>
>                FNAL/CD/CCF/Wide Area Systems
>                [phone]
> 



Begin forwarded message:

> From: Hugh LaMaster <[e-mail]>
> Date: July 21, 2005 1:15:45 PM CDT
> To: Matt Crawford <[e-mail]>
> Subject: Re: IPv6 and Jumbo Frames on Ethernet
> Reply-To: Hugh LaMaster <[e-mail]>
> 
> Matt Crawford wrote:
> 
>> Hm.  It is only supposed to mean that Router Advertisements cannot
>> increase the MTU above whatever it was otherwise believed to be, and
>> that in the absence of configuration information (including DHCP), it is
>> presumed to be 1500.
> 
> Do you mind if I repost this statement to the mailing list?
> 
> Regards,
> Hugh LaMaster
> 



Begin forwarded message:

> From: Matt Crawford <[e-mail]>
> Date: July 21, 2005 2:04:05 PM CDT
> To: Hugh LaMaster <[e-mail]>
> Subject: Re: IPv6 and Jumbo Frames on Ethernet
> 
> 
> On Jul 21, 2005, at 13:15, Hugh LaMaster wrote:
> 
>>> Hm.  It is only supposed to mean that Router Advertisements cannot
>>> increase the MTU above whatever it was otherwise believed to be, and
>>> that in the absence of configuration information (including DHCP), it is
>>> presumed to be 1500.
>> 
>> Do you mind if I repost this statement to the mailing list?
> 
> No.  Please include the admission that perhaps it could have been clearer.  Remember the old "We'll fix it in Draft" mantra that used to be the ipngwg refrain?
> 
>          Matt
>