Re: [mpls] AD review of draft-ietf-mpls-entropy-label

[Kireeti Kompella <[e-mail]>]

Hi Group,

In consolidating text on the setting of TTL, TC and BoS, I noticed that the current text says that the TTL of an EL SHOULD be zero.  I think this is a case for MUST (to prevent accidental forwarding based on an EL), and will reflect that in the next revision.

If there are any objections to a MUST instead of SHOULD, please reply!

Thanks,
Kireeti.

On Aug 20, 2012, at 11:40 , Adrian Farrel wrote:

> Hi,
> 
> Take it as an IETF last call comment (since I already started IETF last call).
> 
> Thanks,
> Adrian
> 
>> -----Original Message-----
>> From: Kireeti Kompella [mailto:[e-mail]]
>> Sent: 20 August 2012 19:30
>> To: [e-mail]
>> Cc: Kireeti Kompella; [e-mail];
> [e-mail]
>> Subject: Re: AD review of draft-ietf-mpls-entropy-label
>> 
>> On Aug 18, 2012, at 10:18 , Adrian Farrel wrote:
>> 
>>> Yeah, and it is great that you give a reason for doing it, but by using
> SHOULD
>>> you have to explain why you might vary. That was what my MAY sentence was
>> trying
>>> to do. If I got the reason wrong then sorry, please add your own reason.
>> 
>> Okay.  Proposed text:
>> 
>>   X MAY choose different values for the TTL and TC fields if it is
>>   known that the ELI will not be exposed as the top label at any point
>>   along the LSP.
>> 
>> If this is acceptable, can this wait, or should we spin an -06?
>> 
>> Kireeti.
>> 
>>> But if
>>> your reason is, because there might turn out to be a reason one day in the
>>> future, then use MUST and write a new I-D sometime in the future.
>>> 
>>> A
>>> 
>>>> -----Original Message-----
>>>> From: Kireeti Kompella [mailto:[e-mail]]
>>>> Sent: 17 August 2012 19:02
>>>> To: [e-mail]
>>>> Cc: Kireeti Kompella; [e-mail];
>>> [e-mail]
>>>> Subject: Re: AD review of draft-ietf-mpls-entropy-label
>>>> 
>>>> On Aug 16, 2012, at 23:43 , Adrian Farrel wrote:
>>>> 
>>>>> So...
>>>>> OLD
>>>>> X SHOULD put the same TTL and TC fields for the ELI as
>>>>> it does for TL.
>>>>> NEW
>>>>> X SHOULD put the same TTL and TC fields for the ELI as
>>>>> it does for TL to protect LSP behavior in cases where PHP is used
>>>>> and the ELI and EL are not stripped at the previous hop (see Section
>>>>> 4.4). If it is known that PHP is not used or that PHP will strip the ELI
>>>>> and EL, the TTL and TC of the ELI MAY be set differently.
>>>>> END
>>>> 
>>>> X will likely be blissfully unaware of whether PHP will be used (and
> whether
>>> the
>>>> PHP LSR will pop ELI+EL), unless this is a very short LSP.
>>>> 
>>>>> ***or***
>>>>> 
>>>>> s/SHOULD/MUST/
>>>> 
>>>> I have a pathological fear of MUST.
>>>> 
>>>> How about:
>>>> 
>>>> NEW
>>>> X SHOULD put the same TTL and TC fields for the ELI as
>>>> it does for TL.  This protects LSP behavior in cases where PHP is used
>>>> and the ELI and EL are not stripped at the penultimate hop (see Section
>>>> 4.4).
>>>> END
>>>> 
>>>> I'd better get out a new version before the list of things to do gets any
>>> longer.
>>>> 
>>>> Kireeti.=
>>> 
>