We're seeing sporadic situations where fones a) send an INVITE b) ignore the 100-Trying response, and instead c) send another INVITE with the same sequence number.
The question is, What should happen here? Currently, we send back a 500 response (server error). Is this correct? (i think not, cause it invariably causes the fone to go fast busy). Is this some other response that should occur?
poring over 3261 resulted in a headache, and no additional clarity...
Mahesh Paolini-Subramanya wrote:
We're seeing sporadic situations where fones a) send an INVITE b) ignore the 100-Trying response, and instead c) send another INVITE with the same sequence number.
The question is, What should happen here?
If the message is completely identical then it is a retransmission. Retransmissions should be detected by tm module (e.g. t_relay()) and absorbed,
regards klaus
Currently, we send back a 500 response (server error). Is this correct? (i think not, cause it invariably causes the fone to go fast busy). Is this some other response that should occur?
poring over 3261 resulted in a headache, and no additional clarity...
Users mailing list Users@openser.org http://openser.org/cgi-bin/mailman/listinfo/users
Well, thats exactly what I thought. What I see, however, is the following -
When t_relay receives the retransmission, it complains about the transaction already being in process (ERROR: t_newtran: transaction already in process 0xb672c4e8 ) and returns a negative value I then proceed to send an sl_reply_error upstream (which is causing the issues)
So, is there something I am missing? Maybe a flag i'm unaware of that says 'absorb retransmissions' or some such thing (he enquires hopefully)
cheers
----- Original Message ----- From: Klaus Darilion klaus.mailinglists@pernau.at To: mahesh@aptela.com Cc: users users@openser.org Sent: Thursday, December 14, 2006 2:06:01 AM GMT-0600 Subject: Re: [Users] Duplicate INVITEs
Mahesh Paolini-Subramanya wrote:
We're seeing sporadic situations where fones a) send an INVITE b) ignore the 100-Trying response, and instead c) send another INVITE with the same sequence number.
The question is, What should happen here?
If the message is completely identical then it is a retransmission. Retransmissions should be detected by tm module (e.g. t_relay()) and absorbed,
regards klaus
Currently, we send back a 500 response (server error). Is this correct? (i think not, cause it invariably causes the fone to go fast busy). Is this some other response that should occur?
poring over 3261 resulted in a headache, and no additional clarity...
Users mailing list Users@openser.org http://openser.org/cgi-bin/mailman/listinfo/users
Well, i read the documentation again, and found t_lookup_request, and t_check_trans. Question 1 - t_lookup_request() creates the transaction if it doesn't exist, correct? However, t_check_trans only checks to see if the transaction already exists, correct?
Question 2 - Assuming that t_check_trans only checks to see if the transaction already exists, then is the following correct?
Currently I do the following (after stripping out a bunch of stuff)
if (!t_relay()) { if (!is_method("ACK")) { sl_reply_error(); } }
should this, instead, actually look like the following? if (is_method("INVITE")) { if (if !t_check_tran()) { t_relay(); } } else if (!is_method("ACK")) { sl_reply_error(); }
----- Original Message ----- From: Klaus Darilion klaus.mailinglists@pernau.at To: mahesh@aptela.com Cc: users users@openser.org Sent: Thursday, December 14, 2006 2:06:01 AM GMT-0600 Subject: Re: [Users] Duplicate INVITEs
Mahesh Paolini-Subramanya wrote:
We're seeing sporadic situations where fones a) send an INVITE b) ignore the 100-Trying response, and instead c) send another INVITE with the same sequence number.
The question is, What should happen here?
If the message is completely identical then it is a retransmission. Retransmissions should be detected by tm module (e.g. t_relay()) and absorbed,
regards klaus
Currently, we send back a 500 response (server error). Is this correct? (i think not, cause it invariably causes the fone to go fast busy). Is this some other response that should occur?
poring over 3261 resulted in a headache, and no additional clarity...
Users mailing list Users@openser.org http://openser.org/cgi-bin/mailman/listinfo/users
Hi Mahesh,
I do not think you need t_check_tran() as t_relay() automatically absorbs the retransmissions.
are you sure you are not calling t_relay() twice for the same requests? or calling a tm function (t_newtran) before the t_relay ??
regards, bogdan
Mahesh Paolini-Subramanya wrote:
Well, i read the documentation again, and found t_lookup_request, and t_check_trans. Question 1 - t_lookup_request() creates the transaction if it doesn't exist, correct? However, t_check_trans only checks to see if the transaction already exists, correct?
Question 2 - Assuming that t_check_trans only checks to see if the transaction already exists, then is the following correct?
Currently I do the following (after stripping out a bunch of stuff)
if (!t_relay()) { if (!is_method("ACK")) { sl_reply_error(); } }
should this, instead, actually look like the following? if (is_method("INVITE")) { if (if !t_check_tran()) { t_relay(); } } else if (!is_method("ACK")) { sl_reply_error(); }
----- Original Message ----- From: Klaus Darilion klaus.mailinglists@pernau.at To: mahesh@aptela.com Cc: users users@openser.org Sent: Thursday, December 14, 2006 2:06:01 AM GMT-0600 Subject: Re: [Users] Duplicate INVITEs
Mahesh Paolini-Subramanya wrote:
We're seeing sporadic situations where fones a) send an INVITE b) ignore the 100-Trying response, and instead c) send another INVITE with the same sequence number.
The question is, What should happen here?
If the message is completely identical then it is a retransmission. Retransmissions should be detected by tm module (e.g. t_relay()) and absorbed,
regards klaus
Currently, we send back a 500 response (server error). Is this correct? (i think not, cause it invariably causes the fone to go fast busy). Is this some other response that should occur?
poring over 3261 resulted in a headache, and no additional clarity...
Users mailing list Users@openser.org http://openser.org/cgi-bin/mailman/listinfo/users