Summary:ASTERISK-04497: Stale nonce with multiline registration.
Reporter:Alan Thomas (athomas)Labels:
Date Opened:2005-06-28 23:08:23Date Closed:2011-06-07 14:03:13
Versions:Frequency of
Environment:Attachments:( 0) Stale_Nonce.txt
( 1) Stale_nonce_on_csv_head.txt
Description:Registration of more than one line from a Carrier Access Adit 3104 causes stale nonce and fails.


Attached is debug info.  I'm guessing but it looks like the nonce is being tracked by the IP and not my the user.  If I have just one line active everything works fine.  But when more than one line is active all fail to reg. Looking at the packets, line one starts to reg and is sent a nonce, but by the time it responds, line to starts and is handed a diffrent nonce.  When line one's reg is recived by Asterisk I think it is using the nonce handed out to line two.  I disabled the ckecking for stale nonce in chan_sip to see if this would help. With this disabled line two would register, but line one would not.  I'm not a programer and unable to trouble shoot further.  I have also tried with CVS downloaded Jun 28, 2005.  
Comments:By: Michael Jerris (mikej) 2005-06-29 00:15:35

Can we have debug from head as well please.

By: Alan Thomas (athomas) 2005-06-29 08:43:57

File Uploaded.

By: Kevin P. Fleming (kpfleming) 2005-07-05 18:18:39

Pedantic SIP checking will probably solve this problem, since the SIP peer is re-using the same Call-ID for the multiple REGISTER requests. This is technically a protocol violation, but we can work with it.

Please try it again with 'pedantic=yes' in the [general] section of sip.conf and let us know what the results are.

By: Alan Thomas (athomas) 2005-07-06 21:33:57

I will try this on Friday.  I did not even notice the reuse.  I will talk to my eng. dept about that too.  BTW I work for Carrier Access.

By: Alan Thomas (athomas) 2005-07-09 11:56:59

I have tested your solution and it is working well.  I have asked why we are using the same call id, but have not recived and answer from eng.  I will continue to work on this with my eng group and see if I can't get us to comply with the RFC.  I will update the bug report when GA code is released that does not need pedantic checking.

By: Olle Johansson (oej) 2005-07-20 13:24:59

Problem with the other side's implementation.

By: Olle Johansson (oej) 2005-07-20 13:25:28