no rtp ports remaining
Submitted by ambroisie on Wed, 04/30/2008 - 12:06am.
hello
wonder if anyone has seen this problem. i today experienced a problem with my users being unable to make or receive calls. the debug messages showed
Apr 30 15:31:14 ERROR[8272] rtp.c: No RTP ports remaining. Can't setup media stream for this call.
Apr 30 15:31:14 ERROR[8272] rtp.c: No RTP ports remaining. Can't setup media stream for this call.
Apr 30 15:31:14 WARNING[8272] chan_sip.c: Unable to create RTP audio and video session: Address already in use
i fixed the issue by restarting the system. wonder if there is an issue with ports not being released. i have my system configured for RTP to use 10001-10021 which should be more than enough for 6 extensions, usually at any one time there are no more than 2-3 calls on the system. this was true when this problem was experienced.
is there a way of checking what ports are in use ?
is there a way of releasing ports ?
is there a timeout or audit process that needs to be modified ?
any suggestions to prevent a reocurrence would be much appreciated
cheers
David
PS: i am running TB 2.2 on this system....it has bene running for months without any issue albeit i only changed the RTP port to this range about 2 weeks ago
you need RTP ports for each
Wed, 04/30/2008 - 9:37am
you need RTP ports for each phone, plus for each SIP trunk (sounds like you have SIP trunks?)
but looks like you should have enough for 5 calls - just check here for proper RTP settings and how it works to verify everything:
http://www.fridu.org/content/view/38/55/
I usually dont limit it that much, maybe 10000-10050 at most.
OP says usually 2-3 calls,
Wed, 04/30/2008 - 3:33pm
hi chaps...apologies for the delay
Wed, 05/07/2008 - 3:53am
been rather busy and have been monitoring this issue but it has not reoccurred - guess that is good and bad. i also found about half a dozen other references (google) to same issue dating back to 2003 or so when digium did do some fixes to asterisk code, but also found a number of instances since when the problem appears to have been a one off. seems to be the same here. at least i know what i can do to monitor it (netstat -na seems to be a good way of tracking the number of open ports and to see if this is increasing - some other instances of this problem seem to have resulted in people losing the entire 10000 port range they had allocated).
my rtp.conf is very simple....
/etc/asterisk/rtp.conf
rtpstart=10001
rtpend=10021
as far as I know:
-for each
Thu, 05/08/2008 - 12:12am
as far as I know:
-for each RTP port, you also open RTCP port
-a call can take up 4 ports
-first port should be even, so 10001 won't be used
-SIP trunks also use rtp ports
Maybe ports aren't released directly
What happens when calls are put on hold...? This way a phone takes up N times 4 ports!
I'd just double your RTP range
http://en.wikipedia.org/wiki/
Thu, 05/08/2008 - 11:38am
http://en.wikipedia.org/wiki/Real-time_Transport_Protocol
RTP does not have a standard TCP or UDP port on which it communicates. The only standard that it obeys is that UDP communications are done via an even port and the next higher odd port is used for RTP Control Protocol (RTCP) communications.
Member Since:
2007-10-30