There's some calls that don't, and we don't have many calls that last that long (which is why we hadn't noticed until now.)
I've had a look through the documentation, and the only thing I saw that looked likely was the rtp-timeouts which I changed to this without any change:
<param name="rtp-timeout-sec" value="1800"/>
<param name="rtp-hold-timeout-sec" value="1800"/>
(changed in internal and external profiles)
Here's the log output for when the call ends:
9d39d866-5732-11e1-b1a8-05fc5f64fa97 2012-02-14 17:42:58.340991 [DEBUG] switch_core_session.c:1009 Send signal sofia/internal/
sip:12605@99.237.220.56:3907 [BREAK]
2012-02-14 17:42:58.340991 [DEBUG] switch_scheduler.c:214 Added task 4 switch_ivr_schedule_hangup (9d39d866-5732-11e1-b1a8-05fc5f64fa97) to run at 1329241378
9d39d866-5732-11e1-b1a8-05fc5f64fa97 2012-02-14 17:42:58.361011 [DEBUG] switch_core_session.c:726 Send signal sofia/internal/
sip:12605@99.237.220.56:3907 [BREAK]
9d39d866-5732-11e1-b1a8-05fc5f64fa97 2012-02-14 17:42:58.521032 [DEBUG] switch_ivr.c:589 sofia/internal/
sip:12605@99.237.220.56:3907 Command Execute respond(603)
9d39d866-5732-11e1-b1a8-05fc5f64fa97 2012-02-14 17:42:58.521032 [DEBUG] mod_sofia.c:2363 Responding with 603 [Decline]
Any other settings I should be looking at?
Cheers,
Fraser