[asterisk-bugs] [LibPRI 0015797]: RR storm on libpri 1.4.10.1

[Asterisk Bug Tracker]

The following issue has been CLOSED 
====================================================================== 
https://issues.asterisk.org/view.php?id=15797 
====================================================================== 
Reported By:                tzafrir
Assigned To:                
====================================================================== 
Project:                    LibPRI
Issue ID:                   15797
Category:                   General
Reproducibility:            always
Severity:                   major
Priority:                   normal
Status:                     closed
Asterisk Version:           1.4.26.1 
JIRA:                        
SVN Branch (only for SVN checkouts, not tarball releases):  1.4  
SVN Revision (number only!):  
Disclaimer on File?:        N/A 
Request Review:              
Resolution:                 open
Fixed in Version:           
====================================================================== 
Date Submitted:             2009-08-31 03:52 CDT
Last Modified:              2010-06-10 10:05 CDT
====================================================================== 
Summary:                    RR storm on libpri 1.4.10.1
Description: 
The following was tested on libpri 1.4.10.1 .

Symptom: after a while RR-s are sent frequently. More frequently than once
per 10 seconds. This is most common on BRI where providers tend to
occasionally take the connection down. But can also be reproduced by
disconnecting the line, reconnecting and allowing libpri to recover.

Note that the issue may not easily reproduce on a local [bp]ri loopback
setup, probably due to timing considerations. The problem is also
reproduced with two Asterisk servers connected through PRI.

Timer t203 is automatically scheduled at end of __q921_receive() function.
Therefore it is scheduled for example after receiving DISC, UA etc. Those
timer events are not removed and automatically rescheduled again and again
that causes eventually RR exchange each seconds instead of each 10 seconds.
Ref the attached log.

The problem was introduced in libpri 1.4 r680 (a while after 1.4.9 was
released) after call to "q921_reset(pri)" was removed from the beginning of
the "q921_dchannel_up()" function.

The problem is that the old T203 timer was never reset. Once the
connection is up, triggering it will still yield sending an RR. After a
second disconnect/reconnect we will have three timers trigerring sending
RRs, and so forth.
====================================================================== 

---------------------------------------------------------------------- 
 (0123228) lmadsen (administrator) - 2010-06-10 10:05
 https://issues.asterisk.org/view.php?id=15797#c123228 
---------------------------------------------------------------------- 
I'm going to close this for now, however the reporter is welcome to reopen
this issue if it is still a problem on libpri 1.4.11.2. Thanks! 

Issue History 
Date Modified    Username       Field                    Change               
====================================================================== 
2010-06-10 10:05 lmadsen        Note Added: 0123228                          
2010-06-10 10:05 lmadsen        Status                   feedback => closed  
======================================================================