[Kamailio-Users] Need to handle an upstream carrier error - Duplicate INVITE Messages

Tue Feb 24 03:17:18 CET 2009

Daniel,
Thanks for the info.  I am already starting my server with -m 2048, so
I should be more than fine on the memory side :)  My Kamailio
deployment is a 64bit server with 8G of RAM.

How does this look for an implementation of this check?  I based it
off the following snippet of the RFC

8.2.2.2 Merged Requests

   If the request has no tag in the To header field, the UAS core MUST
   check the request against ongoing transactions.  If the From tag,
   Call-ID, and CSeq exactly match those associated with an ongoing
   transaction, but the request does not match that transaction (based
   on the matching rules in Section 17.2.3), the UAS core SHOULD
   generate a 482 (Loop Detected) response and pass it to the server
   transaction.

      The same request has arrived at the UAS more than once, following
      different paths, most likely due to forking.  The UAS processes
      the first such request received and responds with a 482 (Loop
      Detected) to the rest of them.

loadmodule("ht.so")
modparam("htable","htable","loop_check=>size=6000;autoexpire=30")

##
## Check to make sure we don't already have an active
## transaction for this call-id, c-seq, and from-tag
## RFC3261 - 8.2.2.2
##
if($sht(loop_check=>$ci::$cs::$ft) == null){
	$sht(loop_check=>$ci::$cs::$ft) = 1	
}else{
	sl_reply_error("482","Loop Detected - Duplicate Session Presentation");
	exit;
}

Thanks for the assistance!!
Geoff

On Mon, Feb 23, 2009 at 9:13 AM, Daniel-Constantin Mierla
<miconda at gmail.com> wrote:
>
>
> On 02/23/2009 03:50 PM, Geoffrey Mina wrote:
>>
>> Sorry about the cross post.  I wasn't sure how many people were on
>> both the OpenSIPs and Kamailio mailing lists... and since this is a
>> 'core' issue, I figured it would be good to get input from the most
>> people.
>
> ok, but separately is better, as solutions in one side do not apply always
> in the other side and will create confusion. Kamailio works on the same
> mailing lists as openser, so we have quite large and mature community here,
> over 1000 subscribed people to this mailing lists only.
>
>
>>  In the future I will only post to one.
>>
>> I will go down the path of the htable, what kind of performance/memory
>> hit am I going to take?
>>
>
> performance should not be an issue, as everything is in memory, just set the
> size of the hash table to a reasonable number. You can count the memory via
> the calls setups per second rate. For example, if you have 100 calls setups
> per second, that mean 100 INVITEs per second.
>
> So if you let the key in hash table for 30 second, that means you need
> memory for 3000 item in hash table. One entry has the size of the key name
> and value plus a 40 bytes (or so) overhead. If you take an average of 32 per
> call id and you store an integer, that means about 80bytes per key (rounded
> up). So, you would need 240kB for it, which is not such big and can live
> with default share memory size.
>
> Anyhow, if you need more share memory, increase it via -m command line
> parameter. Default is 32MB.
>>
>> This system has a lot of memory available to it, how would I increase
>> memory appropriately to ensure the htable had enough to live happily?
>>
>
> If you have enough, then is better to start with 128 MB of shared memoru: -m
> 128
>
> Cheers,
> Daniel
>
>> thanks,
>> Geoff
>>
>> On Mon, Feb 23, 2009 at 7:14 AM, Daniel-Constantin Mierla
>> <miconda at gmail.com> wrote:
>>
>>>
>>> Hello,
>>>
>>> please do not cross-post on many mailing lists. Will create confusion
>>> about
>>> available solutions.
>>>
>>> Theoretically, this is valid in SIP (e.g., 2 invites with same call-id)
>>> --
>>> it is same scenario as parallel forking in upstream.
>>>
>>> However, if you know that this shouldn't happen, you can try to fix it
>>> from
>>> config.
>>>
>>> Fist is to identify why the BYE is routed to the wrong server. It should
>>> follow the Route set and contact address. Can you provide the pcap file
>>> of
>>> such call?
>>>
>>> As solution to deny new invites with same call id is to use the htable
>>> module. Set a key there based on call id (eventually plus from user, from
>>> tag, etc.) and check it before processing the invite, if there is one,
>>> drop
>>> it.
>>>
>>> You just set key auto-expire for 30-60sec so it gets automatically
>>> deleted.
>>>
>>> Note that htable is in devel version (upcoming 1.5.0), but should work
>>> out
>>> of the box with 1.4:
>>> http://kamailio.org/docs/modules/1.5.x/htable.html
>>>
>>> Cheers,
>>> Daniel
>>>
>>>
>>> On 02/22/2009 03:24 AM, Geoffrey Mina wrote:
>>>
>>>>
>>>> Hello,
>>>> I have a carrier who provides PSTN gateway services.   They have
>>>> multiple redundant sip gateway devices in their network.  The problem
>>>> occurs when one of their devices starts to have issues.  I will
>>>> receive an INVITE request from both gateways with the same call-id.
>>>> The problem is that my Kamailio system doesn't detect that I already
>>>> set a call up for the INVITE once, and forwards the request to another
>>>> server in the dispatcher list.  What I end up with is a call on two
>>>> asterisk servers, but only one has the actual RTP stream.  The BYE
>>>> request gets routed to the wrong server, and everything just gets
>>>> screwy.  If anyone could provide any hint on how I might be able to deal
>>>> with
>>>> this scenario, I would really appreciate it.
>>>>
>>>> I have attached my current config file, and the following is a link to
>>>> a google spreadsheet which
>>>> shows the SIP trace.
>>>>
>>>> http://spreadsheets.google.com/ccc?key=pU5i2J6Ck3b519-_M6Et3cw
>>>>
>>>> I have masked my IP addresses for my own sanity.
>>>>
>>>>  XX.XX.XX.179 - Kamailio SIP Gateway
>>>>  XX.XX.XX.189 - Asterisk1
>>>>  XX.XX.XX.186 - Asterisk2
>>>>
>>>>  Thanks!
>>>>  Geoff
>>>>
>>>>  ------------------------------------------------------------------------
>>>>
>>>> _______________________________________________
>>>> Kamailio (OpenSER) - Users mailing list
>>>> Users at lists.kamailio.org
>>>> http://lists.kamailio.org/cgi-bin/mailman/listinfo/users
>>>> http://lists.openser-project.org/cgi-bin/mailman/listinfo/users
>>>>
>>>
>>> --
>>> Daniel-Constantin Mierla
>>> http://www.asipto.com
>>>
>>>
>>>
>>
>> _______________________________________________
>> Kamailio (OpenSER) - Users mailing list
>> Users at lists.kamailio.org
>> http://lists.kamailio.org/cgi-bin/mailman/listinfo/users
>> http://lists.openser-project.org/cgi-bin/mailman/listinfo/users
>>
>>
>
> --
> Daniel-Constantin Mierla
> http://www.asipto.com
>
>