Hi All, I'm investigating http://track.sipfoundry.org/browse/XX-9189 so I'm resurrecting http://thread.gmane.org/gmane.comp.telephony.pbx.sipfoundry.general/30441 discussion...
I managed to recreate the issue on my machine by specifying the UDP range 31000 - 44000 and restarting prompted services. Basically Joe's statement "sipxrls is binding to a udp port that is on the range of our RTP (UDP) ports; which then prevents Media Relay from starting" can be explained as: sipxrls starts before Media Relay and binds to port 38428 (which is in the range of ephemeral ports for linux: http://en.wikipedia.org/wiki/Ephemeral_port), then Media Relay is started and tries to account ports in range 31000 - 44000 (including 38428) but since 38428 is already taken it fails to start (as a test I also started sipxrls and registrar after Media Relay and problem went away). As Joegen suggested we could ignore port if not available at Media Relay startup and go check next port in range, however the symmitron code works mostly with range of ports not with specific ports (meaning that if a bad port at startup it won't know further that the port is bad and will try to use it again later). While I agree this should be the correct fix I feel that radically changing the symmitron code is not what we want at the moment (or at least for 4.6). Therefore I would suggest only to improve help text on config page and to make admin aware about possible side effects when specifying RTP ports in ephemeral range. Looking for some feedback, Thanks, George _______________________________________________ sipx-dev mailing list [email protected] List Archive: http://list.sipfoundry.org/archive/sipx-dev/
