[Devel] Processing REGISTER requests

Klaus Darilion klaus.mailinglists at pernau.at
Thu Oct 6 18:07:28 CEST 2005


Dan Pascu wrote:
> On Thursday 06 October 2005 15:19, Klaus Darilion wrote:
> 
>>I still can't see how we may combine them. Just take your scenario 1.
>>from above. Two different clients -> 2 different call-ids. Thus, your
>>call-id algorithm does not match and you suggest the use the old
>>algorithm. Thus, we again overwrite the contact.
> 
> 
> I really don't understand what you say here.
> 
> First registration.
> 
> Phone 1:
>   callid = somecallid
>   cseq   = 101
>   ip     = 10.0.0.1
>   port   = 5060

Ok! Now the second phone will register:

> Phone 2:
>   callid = anothercallid
>   cseq   = 101
>   ip     = 10.0.0.1
>   port   = 5060

Now ser will use your algorithm:

1. Check for callid: not found

 > 2. if previous step failed to find an entry use the current algorithm 
 > to lookup by contact.

This will match the registration from phone 1. Thus, combining call-id 
and the existing algorithm won't work!

klaus



> Second registration if phones support the RFC recommandation to reuse 
> callid and increment CSeq:
> 
> Phone 1:
>   callid = somecallid
>   cseq   = 103
>   ip     = 10.0.0.1
>   port   = 5060
> 
> Phone 2:
>   callid = anothercallid
>   cseq   = 103
>   ip     = 10.0.0.1
>   port   = 5060
> 
> using callid and cseq each phone will match its previous registration.
> 
> Second registration if phones do not support the RFC recommandation to 
> reuse callid and increment CSeq:
> 
> Phone 1:
>   callid = thirdcallid
>   cseq   = 101
>   ip     = 10.0.0.1
>   port   = 5060
> 
> Phone 2:
>   callid = fourthcallid
>   cseq   = 101
>   ip     = 10.0.0.1
>   port   = 5060
> 
> using callid and cseq each phone will not match its previous registration 
> and the contacts will be overwriten. However note that this happens right 
> now with the current algorithm, so nothing changed. It behaves exactly 
> the same.
> 
> As I said, I've found that 98% of the phones follow the RFC recommendation 
> about using the same callid and incrementing cseq with each register 
> request, so this concludes that using callid will improve contact lookup 
> for 98% of the phones while for the rest will continue to behave like now 
> (not any bit worse).
> 
> Now there are claims that different phones use the same callid which 
> should interfere with this. Until some provides some data about this, in 
> my book is just an hypothesys. Even if true, how many of the 98% do you 
> think a random callid overlapping will affect? I'd say it's still an 
> improvement, even if the critics provide to be true.
> 




More information about the Devel mailing list