[Asterisk-bugs] [Asterisk 0009724]: SIP Transfers To Parking Lot From Grandstream GXP-2000 Locks Up SIP Channel

[noreply at bugs.digium.com]

A NOTE has been added to this issue. 
====================================================================== 
http://bugs.digium.com/view.php?id=9724 
====================================================================== 
Reported By:                kenw
Assigned To:                
====================================================================== 
Project:                    Asterisk
Issue ID:                   9724
Category:                   Channels/chan_sip/Transfers
Reproducibility:            random
Severity:                   major
Priority:                   normal
Status:                     new
Asterisk Version:           1.4.4 
SVN Branch (only for SVN checkouts, not tarball releases):  1.4  
SVN Revision (number only!): 63519 
Disclaimer on File?:        No 
Request Review:              
====================================================================== 
Date Submitted:             05-14-2007 14:49 CDT
Last Modified:              07-19-2007 09:26 CDT
====================================================================== 
Summary:                    SIP Transfers To Parking Lot From Grandstream
GXP-2000 Locks Up SIP Channel
Description: 
Starting a couple of months ago about once a week Asterisk would stop
responding to our SIP phones.  This has increased slowly to 3-5 times a
day.  It's taken me a long time to (hopefully) narrow the problem down to
something I could report.  I've looked everywhere online for help and I
hope I didn't miss this being reported already.

I've been able to reproduce the problem with 1.4.1, 1.4.2, 1.4.4 & the
latest SVN as of last week (63519), the version doesn't seem to affect the
timing of the problem one way or another.

I've attached logs below including SIP Debug enabled, verbose & debug set
to 4.  I've also included a console log of the state of things when it
happens.  

We are using the TRNF button on the GXP-2000 phones to transfer to a
parking lot.  It seems to be a transfer to the parking lot that causes SIP
to stop responding, though I may be mistaken.  Asterisk appears fine but
all phones get 'no response from server' when the lock up occurs.

We've got 3 boxes at 3 locations, the other 2 are working fine, this one
is having fits, however, this box is also used 10x more then the other 2. 
Last week I removed all asterisk folders and reinstalled 1.4.4 but the
problem still occurs.

When the lockup occurs if I try to reload chan_sip I get no errors, if I
try to unload it I get the following:
[May 14 11:40:47] WARNING[6314]: loader.c:458 ast_unload_resource: Soft
unload failed, 'chan_sip.so' has use count 32

The server is running CentOS 4.4, has two TDM-400p cards with 6 FXS & 2
FXO cards.


====================================================================== 

---------------------------------------------------------------------- 
 kenw - 07-19-07 09:26  
---------------------------------------------------------------------- 
*UPDATE*

So I had tried damn near everything and was about to throw the whole damn
system out when it appears I may have finally found the real problem.

First, things I have done between last post and now.

1. Remove all signs of asterisk & reinstall latest 1.4.x - no good
2. Remove all signs of asterisk & reinstall latest 1.2.x - no good
3. Setup new server, clean install of everything, latest 1.4.x, copied
config from existing server - no good (this killed me)
4. Redo configuration from scratch - no good (this REALLY killed me)

Then it dawned on me the only common problem (aside from some network
infrastructure possibility that was affecting only the phone system) was
the moving of the TDM400 cards (I've got 2 of them).  

I had an extra card, I swapped it out with the card that gets the most
usage.  I spent a solid hour last night doing nothing but transfers
different ways on version 1.4.6 without a lockup.

This is huge, as 1.4.x has always amplified the SIP lockup more.  

Another thing that I noticed when I created the new config, is a call in
on a specific Zap channel despite having the proper context & start info
was coming into the dialplan saying it couldn't find (from-zaptel,s,1)
trying from-zaptel,s,1, can't find it, trying default.  So I may only have
a bad FXO card.  Once I swapped cards this problem went away as well.

When it's been a week with no lockups I'll post back, but I'm about 95%
sure we're good to go and the problem is coming from a flaky TDM400
card/module. 

Issue History 
Date Modified   Username       Field                    Change               
====================================================================== 
07-19-07 09:26  kenw           Note Added: 0067585                          
======================================================================