[Serusers] How to handle retrasmissions?

Klaus Darilion klaus.mailinglists at pernau.at
Wed Dec 7 15:29:07 CET 2005


Hi Federico!

I have not followed the whole thread, sorry. Maybe we should go back on 
step. What problems do you have with retransmission? Do you mean 
retransmissions from the client to ser, or from ser to the next hop?

If you forward requests stateful (t_relay), incoming retransmission will 
be routed like the first message and will be absorbed by t_relay.

What is different in your setup that will cause problems?

regards
klaus


Federico Giannici wrote:
> Klaus Darilion wrote:
> 
>> Take a look at the documentation of openser. The functions of the tm 
>> module are well documented and I think the meaning in ser is still the 
>> same. I guess your are looking for t_forward_nonack()
>> http://openser.org/docs/modules/1.0.x/tm.html#AEN351
> 
> 
> I already tryed to use t_forward_nonack(), but I got errors like this:
> 
> Dec  5 18:47:29 eowyn OpenSER[26647]: ERROR:tm:t_forward_nonack: no 
> branch for forwarding
> 
> What am I missing?
> 
> Thanks.
> 
> 
> P.S.
> What about question number 2?
> 
> 
> 
>> Federico Giannici wrote:
>>
>>> What is the correct way to handle message retrasmissions?
>>>
>>> Usually t_relay() seems to hadle them correctly. But if a 
>>> transactions was previously created by some module function, I found 
>>> that t_relay() (and subsequent sl_reply_error()) generates the 
>>> following errors:
>>>
>>> Dec  5 18:49:25 eowyn OpenSER[5654]: ERROR: t_newtran: transaction 
>>> already in process 0x502a56f8
>>> Dec  5 18:49:25 eowyn OpenSER[5654]: ERROR: sl_reply_error used: I'm 
>>> terribly sorry, server error occurred (1/SL)
>>>
>>> So, I have the following questions:
>>>
>>> 1) What is the correct script code to relay a message, even with 
>>> retrasmissions and transaction already created?
>>>
>>> 2) Is it OK to simply discard retrasmissions (not forwarding or 
>>> replying them)?
>>>
>>> 3) How can I recognize retrasmissions? Is t_lookup_request() the 
>>> correct function for a test?
>>>
>>>
>>> Thanks.
> 
> 
> 




More information about the sr-users mailing list