headers
Dan Wing | 3 Dec 2011 07:41
Picon
Favicon

RE: SCTP-over-UDP and port 9899

> -----Original Message-----
> From: Michael Tuexen [mailto:tuexen <at> fh-muenster.de]
> Sent: Wednesday, November 02, 2011 4:35 AM
> To: Dan Wing
> Cc: tsvwg <at> ietf.org; randall <at> lakerest.net
> Subject: Re: SCTP-over-UDP and port 9899
> 
> On Nov 2, 2011, at 3:27 AM, Dan Wing wrote:
> 
> >>> For the server, yes, that's okay.  That is what IANA-registered
> >>> default ports are all about.
> >> But the SCTP stack is neither a client nor a server...
> >
> > "Bullpucky".  Next you're going to say that "peer to peer"
> > doesn't have a client and a server.
> No, I mean it seriously: The local UDP encapsulation port number
> is a variable of the stack (for example in the FreeBSD kernel
> configurable via a sysctl variable). If you want to use one
> number if the application will ever call listen() and a different
> one it it calls never listen(), how should the stack decide that?
> Same for a userland stack...

Probably just a terminology disagreement.  To my mind, everything
is client/server, based on who sends the first packet (the client)
and who receives the first packet (server).

> >>>> That is why I used a MAY. But I'm open to use any other
> formulation.
> >>>> Do you have a suggestion?
> >>>
(Continue reading)

Permalink | Reply | 17 comments |
headers
Hannes Tschofenig | 4 Dec 2011 10:37

Re: SCTP-over-UDP and port 9899

I agree with Dan. Using a fixed source port does not appear to be useful.

On 12/3/11 8:41 AM, "Dan Wing" <dwing <at> cisco.com> wrote:

>> -----Original Message-----
>> From: Michael Tuexen [mailto:tuexen <at> fh-muenster.de]
>> Sent: Wednesday, November 02, 2011 4:35 AM
>> To: Dan Wing
>> Cc: tsvwg <at> ietf.org; randall <at> lakerest.net
>> Subject: Re: SCTP-over-UDP and port 9899
>> 
>> On Nov 2, 2011, at 3:27 AM, Dan Wing wrote:
>> 
>>>>> For the server, yes, that's okay.  That is what IANA-registered
>>>>> default ports are all about.
>>>> But the SCTP stack is neither a client nor a server...
>>> 
>>> "Bullpucky".  Next you're going to say that "peer to peer"
>>> doesn't have a client and a server.
>> No, I mean it seriously: The local UDP encapsulation port number
>> is a variable of the stack (for example in the FreeBSD kernel
>> configurable via a sysctl variable). If you want to use one
>> number if the application will ever call listen() and a different
>> one it it calls never listen(), how should the stack decide that?
>> Same for a userland stack...
> 
> Probably just a terminology disagreement.  To my mind, everything
> is client/server, based on who sends the first packet (the client)
> and who receives the first packet (server).
>
(Continue reading)

Permalink | Reply | 15 comments |
headers
Michael Tuexen | 4 Dec 2011 11:34
Picon

Re: SCTP-over-UDP and port 9899


On Dec 3, 2011, at 7:41 AM, Dan Wing wrote:

>> -----Original Message-----
>> From: Michael Tuexen [mailto:tuexen <at> fh-muenster.de]
>> Sent: Wednesday, November 02, 2011 4:35 AM
>> To: Dan Wing
>> Cc: tsvwg <at> ietf.org; randall <at> lakerest.net
>> Subject: Re: SCTP-over-UDP and port 9899
>> 
>> On Nov 2, 2011, at 3:27 AM, Dan Wing wrote:
>> 
>>>>> For the server, yes, that's okay.  That is what IANA-registered
>>>>> default ports are all about.
>>>> But the SCTP stack is neither a client nor a server...
>>> 
>>> "Bullpucky".  Next you're going to say that "peer to peer"
>>> doesn't have a client and a server.
>> No, I mean it seriously: The local UDP encapsulation port number
>> is a variable of the stack (for example in the FreeBSD kernel
>> configurable via a sysctl variable). If you want to use one
>> number if the application will ever call listen() and a different
>> one it it calls never listen(), how should the stack decide that?
>> Same for a userland stack...
> 
> Probably just a terminology disagreement.  To my mind, everything
> is client/server, based on who sends the first packet (the client)
> and who receives the first packet (server).
> 
>
(Continue reading)

Permalink | Reply | 0 comments |
headers
Michael Tuexen | 4 Dec 2011 12:23
Picon

Re: SCTP-over-UDP and port 9899

On Dec 4, 2011, at 10:37 AM, Hannes Tschofenig wrote:

> I agree with Dan. Using a fixed source port does not appear to be useful.
I'm not suggesting a fixed port. I'm only suggesting that when the stack
(possibly under the assumption that there is only one on the host, i. e.
a kernel based) binds a UDP socket, it MAY use 9989.

Best regards
Michael
> 
> 
> On 12/3/11 8:41 AM, "Dan Wing" <dwing <at> cisco.com> wrote:
> 
>>> -----Original Message-----
>>> From: Michael Tuexen [mailto:tuexen <at> fh-muenster.de]
>>> Sent: Wednesday, November 02, 2011 4:35 AM
>>> To: Dan Wing
>>> Cc: tsvwg <at> ietf.org; randall <at> lakerest.net
>>> Subject: Re: SCTP-over-UDP and port 9899
>>> 
>>> On Nov 2, 2011, at 3:27 AM, Dan Wing wrote:
>>> 
>>>>>> For the server, yes, that's okay.  That is what IANA-registered
>>>>>> default ports are all about.
>>>>> But the SCTP stack is neither a client nor a server...
>>>> 
>>>> "Bullpucky".  Next you're going to say that "peer to peer"
>>>> doesn't have a client and a server.
>>> No, I mean it seriously: The local UDP encapsulation port number
>>> is a variable of the stack (for example in the FreeBSD kernel
(Continue reading)

Permalink | Reply | 13 comments |
headers
Randy Stewart | 4 Dec 2011 17:02

Re: SCTP-over-UDP and port 9899

I agree with Michael here. Having the port assigned by IANA as a MAY
is only a weak reference. It does *not* restrict anyone from using any
port. His argument is sound, in the lack of any other knowledge  you should try
the IANA port.

I would imagine the http document *probably* mentions the IANA assigned port 80, not
saying you *must* use it… but mentioning it so that apps have a choice of port of
last resort so to speak.

I see no difference here.. and think the current wording makes sense…

R
On Dec 4, 2011, at 6:23 AM, Michael Tuexen wrote:

> On Dec 4, 2011, at 10:37 AM, Hannes Tschofenig wrote:
> 
>> I agree with Dan. Using a fixed source port does not appear to be useful.
> I'm not suggesting a fixed port. I'm only suggesting that when the stack
> (possibly under the assumption that there is only one on the host, i. e.
> a kernel based) binds a UDP socket, it MAY use 9989.
> 
> Best regards
> Michael
>> 
>> 
>> On 12/3/11 8:41 AM, "Dan Wing" <dwing <at> cisco.com> wrote:
>> 
>>>> -----Original Message-----
>>>> From: Michael Tuexen [mailto:tuexen <at> fh-muenster.de]
>>>> Sent: Wednesday, November 02, 2011 4:35 AM
(Continue reading)

Permalink | Reply | 12 comments |
headers
Randy Stewart | 4 Dec 2011 17:24

Re: SCTP-over-UDP and port 9899

one follow up here, when I say the current wording, I mean
whats in the current draft as put by Michael in this thread…

R
On Dec 4, 2011, at 11:02 AM, Randy Stewart wrote:

> I agree with Michael here. Having the port assigned by IANA as a MAY
> is only a weak reference. It does *not* restrict anyone from using any
> port. His argument is sound, in the lack of any other knowledge  you should try
> the IANA port.
> 
> I would imagine the http document *probably* mentions the IANA assigned port 80, not
> saying you *must* use it… but mentioning it so that apps have a choice of port of
> last resort so to speak.
> 
> I see no difference here.. and think the current wording makes sense…
> 
> R
> On Dec 4, 2011, at 6:23 AM, Michael Tuexen wrote:
> 
>> On Dec 4, 2011, at 10:37 AM, Hannes Tschofenig wrote:
>> 
>>> I agree with Dan. Using a fixed source port does not appear to be useful.
>> I'm not suggesting a fixed port. I'm only suggesting that when the stack
>> (possibly under the assumption that there is only one on the host, i. e.
>> a kernel based) binds a UDP socket, it MAY use 9989.
>> 
>> Best regards
>> Michael
>>>
(Continue reading)

Permalink | Reply | 0 comments |
headers
Brian F. G. Bidulock | 5 Dec 2011 00:15
Favicon

Re: SCTP-over-UDP and port 9899

+1

Hannes Tschofenig wrote:                                 (Sun, 04 Dec 2011 11:37:28)
> I agree with Dan. Using a fixed source port does not appear to be useful.
> 

--

-- 
Brian F. G. Bidulock
bidulock <at> openss7.org
http://www.openss7.org/
Permalink | Reply | 0 comments |
headers
Chris Benson | 6 Dec 2011 21:39

Re: SCTP-over-UDP and port 9899

Hi folks,

I believe that Michael's statement regarding port number is 
not restrictive, and contains a possibly useful suggestion 
for a port number.

I think Michael's I-D is clearer for containing a suggestion
of using its own IANA assigned port number than it would be 
without mentioning it.  To me it is clear that the specified
port number isn't REQUIRED, but is just a SUGGESTION.  It is
necessary to choose one local encapsulation port number 
(assuming one local SCTP).  I'd be happy to use the IANA 
assigned port 9989, because there possibly a better chance 
that it will be available to me, and not already in use as
a fixed port number of some other unrelated protocol.

There are plenty of protocols with similar or analogous 
statements, usually for recommended listener server port 
selection, which is another way of saying that HTTP servers 
SHOULD use port 80 as their source port etc.  M3UA, SUA, M2PA 
etc. all specify [recommend] one "source port" for data sent
from the role of server.

It shouldn't cause too much grief, because as a suggestion 
only, it can safely be ignored by any 9989-haters.

With thanks, from Chris.

On Sun, 4 Dec 2011, Randy Stewart wrote:
(Continue reading)

Permalink | Reply | 10 comments |
headers
Dan Wing | 7 Dec 2011 00:10
Picon
Favicon

RE: SCTP-over-UDP and port 9899

> -----Original Message-----
> From: Chris Benson [mailto:cbenson <at> adax.com]
> Sent: Tuesday, December 06, 2011 12:39 PM
> To: Michael Tuexen
> Cc: tsvwg <at> ietf.org; Dan Wing; Randy Stewart
> Subject: Re: SCTP-over-UDP and port 9899
> 
> Hi folks,
> 
> I believe that Michael's statement regarding port number is
> not restrictive, and contains a possibly useful suggestion
> for a port number.
> 
> I think Michael's I-D is clearer for containing a suggestion
> of using its own IANA assigned port number than it would be
> without mentioning it.  To me it is clear that the specified
> port number isn't REQUIRED, but is just a SUGGESTION.  It is
> necessary to choose one local encapsulation port number
> (assuming one local SCTP). 

Maybe we are finally getting to the root of the problem.  I
hope so.

I will ask:  why is it necessary to have one local encapsulation
port number?

For example, why can't I use source port 5555 from my IPv4
interface and use source port 8888 from IPv6 interface?  What
breaks?
(Continue reading)

Permalink | Reply | 9 comments |
headers
Michael Tuexen | 7 Dec 2011 10:35
Picon

Re: SCTP-over-UDP and port 9899

On Dec 7, 2011, at 12:10 AM, Dan Wing wrote:

>> -----Original Message-----
>> From: Chris Benson [mailto:cbenson <at> adax.com]
>> Sent: Tuesday, December 06, 2011 12:39 PM
>> To: Michael Tuexen
>> Cc: tsvwg <at> ietf.org; Dan Wing; Randy Stewart
>> Subject: Re: SCTP-over-UDP and port 9899
>> 
>> Hi folks,
>> 
>> I believe that Michael's statement regarding port number is
>> not restrictive, and contains a possibly useful suggestion
>> for a port number.
>> 
>> I think Michael's I-D is clearer for containing a suggestion
>> of using its own IANA assigned port number than it would be
>> without mentioning it.  To me it is clear that the specified
>> port number isn't REQUIRED, but is just a SUGGESTION.  It is
>> necessary to choose one local encapsulation port number
>> (assuming one local SCTP). 
> 
> Maybe we are finally getting to the root of the problem.  I
> hope so.
> 
> I will ask:  why is it necessary to have one local encapsulation
> port number?
Interesting question.
> 
> For example, why can't I use source port 5555 from my IPv4
(Continue reading)

Permalink | Reply | 0 comments |

Gmane