Saúl,

I've had it running for several hours now with much more carriers than normal 
using media relay.  And not a single 400 in the logs.  Excellent!


- Jeff



On Aug 18, 2011, at 9:46 AM, Jeff Pyle wrote:

> On Aug 18, 2011, at 8:40 AM, Saúl Ibarra Corretgé wrote:
> 
>> 
>> On Aug 18, 2011, at 2:26 PM, Jeff Pyle wrote:
>> 
>>> Hi Saúl,
>>> 
>>> On Aug 18, 2011, at 4:03 AM, Saúl Ibarra Corretgé wrote:
>>> 
>>>> Calling route(x) from the failure route is not the same as getting a new 
>>>> message into the proxy. When you add headers to the message in the main 
>>>> route you can't remove them in the failure route, you need to use a branch 
>>>> route. Same case applies here. In the case you described above you would 
>>>> call use_media_proxy twice for the same message.
>>> 
>>> I'm not doing anything relay-related in the failure route.  If we land in 
>>> the failure route due to a carrier's 503, we immediately return to the 
>>> request route via the route[x] call to evaluate the next carrier, deciding 
>>> at this point whether or not we need any relay functions based on the needs 
>>> this time, and to some extent what happened last time.
>>> 
>> 
>> Calling route(x) will not magically bring you to the main route (route(0) 
>> really). You may execute the same code, but you are in a different state, 
>> because the request failed.
> 
> In my case it's route(2).  Interesting distinction.
> 
> Well, your branch_route / failure_route suggestion seems to work at least as 
> well as my other approaches on the test box.  I'll get it onto the production 
> boxes during the next maintenance window and see how it does.
> 
> 
> - Jeff
> 
> 
>> --
>> Saúl Ibarra Corretgé
>> AG Projects
> 
> _______________________________________________
> Users mailing list
> [email protected]
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users


_______________________________________________
Users mailing list
[email protected]
http://lists.opensips.org/cgi-bin/mailman/listinfo/users

Reply via email to