Re: [sipcore] Session timer fix

["Jesske, Roland" <[e-mail]>]

Hi,
I agree on Christer’s comment.
But nevertheless I would be happy to see a clear advice for the refresh mechanism when the S-E timer is switched off.
Perhaps an table and some text in section 7.4 for the refresh mechanism would help e.g.:

This is valid for UPDATE and INVITE refreshing a session where S-E is negotiated and the supported header with timer is set.


UAC supports?    SE parameter    S-E parameter        S-E
                  in request     in response      switched off
----------------------------------------------------------
      Y              Y                N                Y
      Y              N                Y                N
      Y              N                N                Y

                        Figure X Switch off S-E

When the when required timer set the S-E cannot be switched off.

What do you think?
Of course we could do the same for the initial negotiation to make everything clear.

Best Regards

Roland


Von: Christer Holmberg [mailto:[e-mail]]
Gesendet: Dienstag, 31. Oktober 2017 19:44
An: Roman Shpount <[e-mail]>
Cc: Paul Kyzivat <[e-mail]>; Jesske, Roland <[e-mail]>; [e-mail]
Betreff: RE: [sipcore] Session timer fix

Hi,

I may reply to Roman’s comments later.

But, at the end of the day I think what Paul said is what matters most: the procedures for turning session timer etc applies AFTER the session timer has been fully negotiated.

So, if the session timer is negotiated during the initial INVITE transaction, whatever UPDATEs are sent during that transaction will not contain session timer information, and they will have no impact on the session timer.

But, once the session timer has been negotiated (once the UAC has received the 200 OK, the rules regarding turning the session timer off etc applies.

Regards,

Christer


From: Roman Shpount [mailto:[e-mail]]
Sent: 31 October 2017 20:21
To: Christer Holmberg <[e-mail]<mailto:[e-mail]>>
Cc: Paul Kyzivat <[e-mail]<mailto:[e-mail]>>; Jesske, Roland <[e-mail]<mailto:[e-mail]>>; [e-mail]<mailto:[e-mail]>
Subject: Re: [sipcore] Session timer fix


On Tue, Oct 31, 2017 at 4:35 AM, Christer Holmberg <[e-mail]<mailto:[e-mail]>> wrote:
My suggestion is that no-SE-in-response would switch off if SE was in the
request.

So, if there is no SE in the request, there is no switch off.

This will break pretty much every SE implementation. There are number of implementations where UA does not insert SE in the request. It is inserted either by proxies (actually this is the most common use case) or by the answering agent.

I think it would be much better to have a new header which will specify "do not update Session-Timer" in the offer.
This way if proxies insert Session-Timer, new header will overwrite.

Regards,
_____________
Roman Shpount