Re: [tcpm] draft-ietf-tcpm-tcp-uto-02

So, a summary of my take and responding to some of this (which I hope we 
now have right). I'd be happy with new text saying something along the 
lines of:

* MIN RTO is system-defined (with a warning that small requires 
caution;-) ).
* The RECOMMENDED value for MIN RTO is follow RFCxxx.

* UTO negotiation SHOULD be OFF by default.
* UTO negotiation MAY be via a system config.
* RECOMMEND negotiation is enabled via an API call from the application.
* Upper and lower bounds may be set, but MUST NOT go below system MIN RTO
* MUST use MIN RTO, if UTO negotiation is not enabled.

?

Gorry

Fernando Gont wrote:

> At 03:56 p.m. 12/12/2005, Ted Faber wrote:
> 
>> If I recall correctly, they do have a min/max kind of set up that
>> encourages systems not to push their defaults down.  I can see a good
>> argument that UTO negotiation should *never* reduce the default min UTO
>> of a system to avoid violating POLA.  Is this restriction what you're
>> arguing for, Gorry?  (If so, Lars/Fernando, do you have
>> objections/counter-arguments?)
> 
> 
> Of the top of my head, I don't recall we proposed any value for the 
> lower limit.
> 
> Anyway, that's what we had in mind when including the equation for 
> choosing the UTO: "As long as it is within acceptable limits, choose the 
> larger of the two".
> 
> The idea is that if you don't set the UTO explicitly, then your local 
> UTO should be set to the default UTO (which may even be larger than the 
> minimum allowed UTO, i.e., the lower limit). This means that, your user 
> timeout will be, at least, the default. And, at most, that of the upper 
> limit.
> 
> If you do set the UTO explicitly (i.e., you know what you're doing), 
> then the user timeout will be, at least, that of the lower limit, and, 
> at most, that of the upper limit.
> 
> (I realize that this is probably not explained in the draft, though).
> 
> Does it make sense?
> 
> Kindest regards,
> 
> -- 
> Fernando Gont
> e-mail: fernando@gont.com.ar || fgont@acm.org
> 
> 
> 
> 
> 
> 

_______________________________________________
tcpm mailing list
tcpm@ietf.org
https://www1.ietf.org/mailman/listinfo/tcpm