[Freeswitch-users] Profile failing on restart after a few hours of use

[Francis]

This had me stumped for two days. The server would reboot, FS would come 
up and profiles would work for a few seconds, then fail. Checking the 
profile status showed it missing, but manually starting it, everything 
worked?

The trigger was the ipv6 interface coming up late, causing the profiles 
to reload:


2014-05-08 03:51:39.443205 [INFO] mod_sofia.c:5751 EVENT_TRAP: IP change 
detected
2014-05-08 03:51:39.443205 [INFO] mod_sofia.c:5752 IP change detected 
[]->[] [::1]->[2632:f0d0:4211:a::2]
2014-05-08 03:51:39.803298 [NOTICE] sofia_glue.c:5885 Reload XML 
[Success]
2014-05-08 03:51:39.803298 [INFO] mod_enum.c:876 ENUM Reloaded
2014-05-08 03:51:39.803298 [INFO] switch_time.c:1191 Timezone reloaded 
530 definitions
2014-05-08 03:51:39.843230 [NOTICE] sofia_reg.c:135 UN-Registering 
footel
2014-05-08 03:51:40.843346 [NOTICE] sofia.c:2682 Waiting for worker 
thread
2014-05-08 03:51:40.843346 [INFO] switch_core_sqldb.c:1495 
sofia:external Destroying SQL queue.
2014-05-08 03:51:40.963253 [NOTICE] sofia.c:2682 Waiting for worker 
thread
2014-05-08 03:51:40.963253 [INFO] switch_core_sqldb.c:1495 
sofia:internal Destroying SQL queue.
2014-05-08 03:51:41.043214 [INFO] switch_core_sqldb.c:1439 
sofia:external Stopping SQL thread.


.... which wasn't a problem, but then this error would pop up halfway 
through the reload:


2014-05-08 03:51:43.883336 [NOTICE] sofia.c:2682 Waiting for worker 
thread
nua.c:223 nua_destroy() nua_destroy(0x90c7a98): FATAL: nua_shutdown not 
completed
2014-05-08 03:51:43.883336 [INFO] switch_core_sqldb.c:1495 
sofia:int-15000 Destroying SQL queue.
2014-05-08 03:51:44.083185 [INFO] switch_core_sqldb.c:1439 
sofia:int-15000 Stopping SQL thread.


.... resulting in this error:


2014-05-08 03:51:44.083185 [NOTICE] sofia.c:4874 Started Profile 
int-15000 [sofia_reg_int-15020]
2014-05-08 03:51:44.083185 [DEBUG] sofia.c:2413 Creating agent for 
int-15000
2014-05-08 03:51:44.103652 [ERR] sofia.c:2506 Error Creating SIP UA for 
profile: int-15000 (sip:mod_sofia at 115.89.334.121:15000;maddr=10.5.5.250)
The likely causes for this are:
1) Another application is already listening on the specified address.    
                                                                         
2) The IP the profile is attempting to bind to is not local to this 
system.


So as an experiment, I deleted all the db's in /var/lib/freeswitch/db 
and the error disappeared, for two reboots. On the third reboot, the 
db's had grown again (less than 1mb all told) and the errors resurfaced. 
I also get the same error intermittently when restarting the profile 
manually. However, stopping the profile, waiting 10 seconds, then 
starting it again works fine.

The system is FreeSWITCH (Version 1.2.23 -1 32bit) on an Atom N270 
1GBram 32GB USB3 MLC flash, which I thought would be fast enough to 
handle this. It seems like nua isnt' able to finish whatever it's doing 
before FS tries to restart the profile, causing it to assume something 
else is listening on the address and fail loading the profile. Is there 
any way I can extend the restart timeout to avoid this?

-Francis