[Spce-user] SIP-Response INJECTION

[Klaus Peter v. Friedeburg]

Hi Jon,

I am not sure but I have found something in the kamailio-doc's:
Module SIPUTILS
There exist an function "ring_insert_callid()"
From the doc:
4.1.  ring_insert_callid()
Inserting the call-id in the internal list, which is checked when further replies arrive. Any 183 reply that is received during the timeout value will be converted to a 180 message. Please note that you need to set a positive timeout value in order to use this function.
The function returns TRUE on success, and FALSE during processing failures.
This function can be used from REQUEST_ROUTE and FAILURE_ROUTE.

Do you have some experiences with this? Can this be a solution for my problem?

I have spoken with ZYXEL about this problem, but they need some large time to fix it in the firmware of the cpe.

Some other answers in the text below.

          Klaus Peter

> -----Ursprüngliche Nachricht-----
> Von: Jon Bonilla (Manwe) [mailto:jbonilla at sipwise.com]
> Gesendet: Mittwoch, 21. November 2012 14:02
> An: spce-user at lists.sipwise.com
> Cc: Klaus Peter v. Friedeburg
> Betreff: Re: [Spce-user] SIP-Response INJECTION
> 
> El Wed, 21 Nov 2012 13:08:44 +0100
> "Klaus Peter v. Friedeburg" <friedeburg at aco.de> escribió:
> 
> > Hi Jon,
> >
> > 1. in sems we have disabled whitelisting for headers at all.
> 
> Good.
> 
> > 2. I have tested to change the IP Port for the peering-servers to 5070 but
> > it  don't change anything in this behavior. On the side of BT our peer is
> > only configured using UDP.
> 
> Good. You'll see it when they send incoming calls.
I only see incomming UDP packets from this peer 


> 
> >3. Our session timers for the Gateway BT are
> > disabled, because it seems to me that only one side should control the
> > sessions.
> 
> If you don't enable session timers for the gw, no session timers will be used
> as the system won't announce the support for them. You need to enable them and
> both servers will negotiate who's the refresher. Actually, the version of sems
>  we'll release this week has a fix to the session timer refresher negotiation
>  problem we found in our last bt interconnection.
I see the following:
When I disabled the session timer in sems the session timer works directly between the PSTN-Gateway and the SIP-End-Node. I see reINVITE messages every 1800 secs, or whatever is configured in the endpoint,  coming from PSTN-GW and forwarded to the client and from there are handled with 200 OK and vice versa. For OnNET Calls the session timer works directly from one CPE to the other CPE.  I am not sure but in our scenario we don’t need to use session-timer. 


> 
> 
> >
> > An Question of the planning update of seems: It is possible to update only
> > sems? Because we have made some customizations in OSS, www-admin and www-csc,
> > so that is difficult to update the hole system.
> >
> >
> 
> apt-get update
> apt-get install ngcp-sems
> 
> 
> This btw is not related to your PRACK problems. So we'll need to disable them
> somehow in the system. Anyways, the PRACK usage when peering against bt is
> quite absurd because they will send you about 4 provisional responses and you
> need to PRACK all of them. That's a nightmare for ngrep debugging :)
> 
> 
> Can't you just disable 100rel support in the CPEs?
I have spoken with ZYXEL about this problem, but they need some large time to fix it in the firmware of the cpe.