Guys, I'm currently routing some calls from one VoIP platform to another OpenSIPS based platform using two ISDN to SIP gateways that are connected back to back. This setup is quite recource heavy, expensive and has a limited capacity. So I'm thinking of connecting both platforms using a (well protected) SIP interconnection. That part is easy, we've done that before.
The problem is in the RTP some phones, connected to the old platform, are sending out. Every 30 seconds or so they send out 0 length RTP messages that some SIP UAs really don't like. Some hardware will hangup a call when it receives 0 length UDP frames in an RTP stream, others will stop handling the incoming RTP traffic allthogether resulting in one way audio. The 0 length UDP messages appear to conform to RFC6263 (http://tools.ietf.org/html/rfc6263) which is really new... I've tried talking to the manufacturer of the phones, talked to the supplier of the VoIP platform, talked to everyone and their neighbour and all say it's not their problem. I've identified two places where *I* can solve it. - In our core routers - At every mediaproxy machine The first option is sub-optimal, I don't want all our routers having a drop-this-packet "firewall" line for various reasons. The second option I've started to like more and more. There's two ways to resolve this: - I just make sure I add an iptables call somewhere in the startup script, or - I/We add an RFC6263 configuration option to Mediaproxy that does more or less the same The iptables call would drop all 0 length UDP messages sent to the mediaproxy ports. Am I wrong in my thinking? -- Andreas Sikkema _______________________________________________ Users mailing list [email protected] http://lists.opensips.org/cgi-bin/mailman/listinfo/users
