Re: [Dime] Issue#32 status

Hi Jouni,

if d is the maximum validity duration a reporting node ever sent, then d is sufficiently long.
(there may be better i.e. shorter estimations but those would be subject to more complex conditions)  

Ulrich

-----Original Message-----
From: ext Jouni Korhonen [mailto:jouni.nospam@gmail.com] 
Sent: Friday, February 28, 2014 7:22 PM
To: Steve Donovan; Wiehe, Ulrich (NSN - DE/Munich)
Cc: Ben Campbell; dime@ietf.org list
Subject: Re: [Dime] Issue#32 status

A question.. how much is "a sufficiently long period of no overload"?
I'd like to see some guidance or minimum values documented.

- Jouni

On Feb 25, 2014, at 6:36 PM, Steve Donovan <srdonovan@usdonovans.com> wrote:

> Ulrich,
> 
> Yes, with that period being long enough for the reporting node to be confident that all previously sent overload reports have expired.
> 
> Steve
> 
> On 2/25/14 10:21 AM, Wiehe, Ulrich (NSN - DE/Munich) wrote:
>> Steve, Ben,
>> 
>> for my clarification, your proposal is to say
>> 
>> ***
>> Sequence number is of type Unsigned64.
>> 
>> When generated, a new sequence number must be greater than the sequence number contained in the active overload report to which it applies (including over reboot of that node).  Note: this allows sequence numbers to start at 1 for the initial occurrence of an overload condition at a reporting node.
>> ***
>> 
>> If so, what is meant by "initial occurrence of an overload condition"?
>> 
>> I guess it means something like moving from no overload to overload after a sufficiently long periode of no overload
>> 
>> Please clarify
>> 
>> Ulrich
>> 
>> 
>> From: DiME [
>> mailto:dime-bounces@ietf.org
>> ] On Behalf Of ext Steve Donovan
>> Sent: Tuesday, February 25, 2014 4:02 PM
>> To: Ben Campbell
>> Cc: 
>> dime@ietf.org
>>  list
>> Subject: Re: [Dime] Issue#32 status
>> 
>> I agree with the suggested change.
>> 
>> Steve
>> On 2/24/14 5:00 PM, Ben Campbell wrote:
>> + 1, except as noted:
>> 
>> On Feb 24, 2014, at 2:33 PM, Steve Donovan 
>> <srdonovan@usdonovans.com>
>>  wrote:
>> 
>> Ulrich,
>> 
>> Would you agree to the following to replace the first two statements:
>> 
>> Sequence number is of type Unsigned64.
>> 
>> When generated, a new sequence number must be greater than the sequence number contained in the active overload report to which it applies (including over reboot of that node).  Note: this allows sequence numbers to start at 1 for any occurrence of overload at a reporting node.  This, I think, allows us to ignore wraparound issues as wraparound will never happen.  Unless we are worried about a server staying in overload for billions of years (assuming reports with a ten minute validity period refreshed every five minutes).
>> 
>> s/ any occurrence of overload / the initial occurrence of an overload condition
>> 
>> 
>> The other two statements are good.
>> 
>> Steve
>> 
>> 
>> 
>> 
>> 
> 
> _______________________________________________
> DiME mailing list
> DiME@ietf.org
> https://www.ietf.org/mailman/listinfo/dime