Re: [sipcore] Session-timer: Result below an acceptable value

[Paul Kyzivat <[e-mail]>]

On 10/6/18 3:50 AM, OKUMURA Shinji wrote:
> Hi,
> 
> If a proxy decreases the S-E, the next proxy is more likely to reject
> the request by 422.
> So I think that a proxy should adjust the Min-SE rather than decrease
> the S-E.
> And ultimately, a UAS should decrease the S-E as low as possible after
> considering it's own wishes.

There is an inherent conflict here: those elements (presumably proxies) 
that need the timer for tracking session state (e.g. for billing) may 
want a small S-E value to improve precision. The other elements (notably 
the UAs) that don't *need* it want the value as large as possible in 
order to reduce their load. The negotiation of S-E and Min-SE is 
intended to seek a compromise between these. But there is no guarantee 
that there is an acceptable compromise. Somebody may decide there is no 
acceptable value and reject the session. I see no "solution" to this.

	Thanks,
	Paul

> Regards,
> Shinji
> 
> On 2018/10/04 22:33, Christer Holmberg wrote:
>> Hi,
>>
>>
>> The UAS is not allowed to include Min-SE to begin with (unless it sends a 422 response).
>>
>>
>> The UAS is currently allowed to reduce the S-E value in the response, as long as it it's larger than the Min-SE value in the corresponding request. My suggestion would be to say that a UAS SHOULD NOT reduce the S-E. Again, assuming the proxies know better how often they want session refreshes, there is no need for the UAS to reduce the value.
>>
>>
>> Regards,
>>
>>
>> Christer
>>
>>
>>
>> ________________________________
>> From: [e-mail] <[e-mail]>
>> Sent: Thursday, October 4, 2018 4:21 PM
>> To: Christer Holmberg
>> Cc: Roman Shpount; OKUMURA Shinji; SIPCORE
>> Subject: Re: [sipcore] Session-timer: Result below an acceptable value
>>
>> Hi
>> Reading this. I understand that the proxy should be allowed to increase the Min SE.
>> For my understanding.
>> Following that the UAS cannot lower it again. Correct?
>> That would be my understanding.
>>
>> Best Regards
>> Roland
>>
>> Am 04.10.2018 20:05 schrieb Christer Holmberg <[e-mail]>:
>>
>> Hi,
>>
>>
>> In order to keep track of the suggested RFC modifications, I suggest that we create GitHub issues.
>>
>>
>> As an example, I created one for Shinji's suggestion to allow a proxy to increase the Min-SE header field value.
>>
>>
>> https://github.com/cdh4u/draft-sessiontimer-race/issues/1
>>
>> [https://avatars0.githubusercontent.com/u/9333380?s=400&v=4]<https://github.com/cdh4u/draft-sessiontimer-race/issues/1>
>>
>> Proxy allowed to increase Min-SE header field value · Issue #1 · cdh4u/draft-sessiontimer-race<https://github.com/cdh4u/draft-sessiontimer-race/issues/1>
>> github.com
>> BACKGROUND: Section 8.1 of RFC 4028 says: "A proxy MUST NOT insert a Min-SE header field or modify the value of an existing header field in a proxied request if that request contains a Support...
>>
>>
>>
>>
>>
>> (As always, any formal decision whether to modify something will be made on the list)
>>
>>
>> Regards,
>>
>>
>> Christer
>>
>>
>>
>>
>> ________________________________
>> From: sipcore <[e-mail]> on behalf of Christer Holmberg <[e-mail]>
>> Sent: Thursday, October 4, 2018 2:43 PM
>> To: Roman Shpount
>> Cc: OKUMURA Shinji; SIPCORE
>> Subject: Re: [sipcore] Session-timer: Result below an acceptable value
>>
>>
>> Hi,
>>
>>
>> Sorry, I read your first e-mail wrong: I thought you said that the proxy would DECREASE Min-SE. But, now I see that you talk about INCREASE 😊
>>
>>
>> I see no reason why a proxy couldn't increase Min-SE.
>>
>>
>> As we have learned, the main users of session timer are proxies, not UAs, so a UA should not care if the minimum value is increased.
>>
>>> Technically speaking UA still care about how often they need to generate refresh since this can result in extra load on the UA.
>> Sure, but increasing the Min-SE value will not result in extra load - on the contrary it may reduce load :)
>>
>>> When session timer is negotiated, S-E specifies the max acceptable value and Min-SE specifies min acceptable value, so proxy should be able to increase Min-SE and decrease SE.
>> Correct.
>>
>> Regards,
>>
>> Christer
>>
>>
>>
> 
> 
> 
> _______________________________________________
> sipcore mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/sipcore
>