[sr-dev] Status of outbound

[Peter Dunkley]

Hello,

Issues 1 and 2 on the list should now be fixed in git master.

I think there is some more testing (and possibly tweaking) to be done 
around the unregister()/$ruid stuff I've done for issue 2. This will be 
done once Hugh has fixed issue 3 and I can build a proper outbound 
configuration to test all the different branching scenarios with.

I plan to add some example configuration files for outbound (edge proxy 
and registrar) once all of the issues have been sorted.

Regards,

Peter


On 14/03/13 12:17, Peter Dunkley wrote:
> Hi,
>
> As 4.0.0 is now out I thought it would be a good time to post a 
> summary for those who are interested in outbound support.
>
> I have listed the known issues below, their status, and what is likely 
> to be available as bug-fixes in the 4.0 branch.
>
> _*1 Registrar module issues surrounding Require:/Supported: headers 
> and outbound
> *_*Detail:* The registrar module doesn't quite follow the procedures 
> in RFC 5626 with regards to checking and setting Require: and 
> Supported: headers (issue identified by Olle during SIPit)
> *Status**:* Fixed in pd/outbound (some tidying of the parser code 
> still required)
> *4.0 branch:* won't be fixed in this branch as some extensive changes 
> have been made to the Kamailio parser code to support these fixes
>
> *_2 Registrar module cannot unregister on broken flo_*_*w
> *_*Detail:* When a broken flow is detected the contact binding 
> associated with that flow should be removed to prevent it being used 
> again.  Failure to do so could result in a large number of old, 
> broken, contact bindings ending up in usrloc (especially if long 
> registration expiry times are used)
> *Status:* Fix in progress in pd/outbound (updated API needs to be 
> tested and ruid needs to be added to the destination set and made 
> available as a PV)
> *4.0 branch:* won't be fixed in this branch as the changes are 
> architectural in nature
>
> _*3 Forking issues with outbound
> *_*Detail**:* When a broken flow is found during forking we need to be 
> able to re-send the request to that client instance immediately 
> without waiting for all other forked legs to complete (issue 
> identified by Olle during SIPit)
> *Status:* Hugh Waite is working on this. An email describing the 
> proposed fix will be sent to sr-dev soon for others to comment on.
> *4.0 branch:* won't be fixed in this branch as this will require 
> extensive architectural changes relating to the handling of branch 
> failures
>
> _*4) Broken in-dialog routing when using outbound
> *_*Detail:* In-dialog requests (including ACK and BYE) are incorrectly 
> routed when outbound is used (issued identified by Olle during SIPit)
> *Status:* The bug is in the loose routing code in the rr module.  I 
> now have a test system for this here and will work on this issue once 
> issues 1 and 2 are finished.
> *4.0 branch:* will be fixed in this branch.  It should be a relatively 
> small change in one function in the rr module.
>
> The upshot of this is that I expect Kamailio 4.0.1 (when it is 
> released) to be suitable for use as an outbound edge proxy, but anyone 
> wanting to use Kamailio as an outbound registrar will need Kamailio 
> 4.1.0 and above.
>
> Regards,
>
> Peter
>
>
> _______________________________________________
> sr-dev mailing list
> sr-dev at lists.sip-router.org
> http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-dev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.sip-router.org/pipermail/sr-dev/attachments/20130315/0badc6d8/attachment.htm>