Re: BFD UDP ports

[David Ward <[e-mail]>]

See the description of the "control" session in echo mode.

-DWard

On Dec 5, 2008, at 7:25 PM, Nobo Akiya (nobo) wrote:

>
> Hello Vishwas.
>
>> Another question I had in mind was regarding the use of the
>> "echo mode". Should we allow parallel sending of echo as well
>> as asynch packets, or at one time only one mechanism needs to
>> be active?
>
> In echo mode, async packets are alos needed to support prarameter
> changes. But the base draft Sec 6.4 suggests that the rate can be
> slower since echo packets are checking the liveliness.
>
> Thanx,
> Nobo
>
>> -----Original Message-----
>> From: Vishwas Manral [mailto:[e-mail]]
>> Sent: Saturday, December 06, 2008 2:40 AM
>> To: Nobo Akiya (nobo)
>> Cc: Nitin Bahadur; David Ward; Dave Katz; [e-mail]
>> Subject: Re: BFD UDP ports
>>
>> Hi Nitin/ Nobo,
>>
>> Thanks for your replies. So I guess we will have to use the
>> same port for IPv4 and IPv6 then, hack the OS or whatever other way.
>>
>> Nobo, I agree we should make the recomendation as a MAY
>> instead of a may as I had suggested.
>>
>> Another question I had in mind was regarding the use of the
>> "echo mode". Should we allow parallel sending of echo as well
>> as asynch packets, or at one time only one mechanism needs to
>> be active?
>>
>> Thanks again,
>> Vishwas
>>
>>> Hello Vishwas, Nitin.
>>>
>>>>
>>>> On 12/4/08 12:30 PM, "Vishwas Manral"
>> <[e-mail]> wrote:
>>>>
>>>>> Hi
>>>>>
>>>>> 1. I am a bit confused about using the same BFD destination port
>>>>> for
>>>>> IPv4 and IPv6. Shouldn't we have different port numbers?
>> I realized
>>>>> that even RIP which uses UDP uses different ports for
>> IPv4 and IPv6.
>>>>> Using the same port can be an issue in a few heavily
>> used OS. Has
>>>>> anyone implemented BFD for IPv6?
>>>>
>>>> Using the same port should be for v4 & v6 should be fine.
>> I know of
>>>> at least
>>>> 1 implementation ;-)
>>>>
>>>> Thanks
>>>> Nitin
>>>>
>>>
>>> I know of another implementation which same dest port is
>> used for v4 &
>>> v6 =)
>>>
>>>> 2. As BFD source port is not used for packet replies and as the
>>>> Source address/ ifindex may be a unique identifier for a
>> source, do
>>>> we need not have the condition which states "source port number
>>>> SHOULD be unique among all BFD sessions on the system". Can we
>>>> downgrade this condition to a may?
>>>
>>> I tripped over this requirement as well, and I agree that
>> measurement
>>> of the wording strength is a bit confusing.
>>>
>>> One soft correction on your statement is that source port
>> MAY be used
>>> as a demultiplexing *aid*.
>>>
>>> http://tools.ietf.org/html/draft-ietf-bfd-v4v6-1hop-08 Sec4.1:
>>>
>>>   An implementation MAY use the UDP port source number to aid in
>>>   demultiplexing incoming BFD Control packets, but ultimately the
>>>   mechanisms in [BFD] MUST be used to demultiplex incoming
>> packets to
>>>   the proper session.
>>>
>>> Thanx,
>>> Nobo
>>>
>>