Has anyone seen this issue before?

Drayteks both end of the link - user dial in PPP VPN access.

Every time uPNP announces anything the VPN suffers enough packet loss to drop any active sessions within the VPN, but not the VPN its self.

78630 968.369676 172.16.99.104 239.255.255.250 SSDP 175 M-SEARCH * HTTP/1.1
78631 968.436783 172.16.99.11 239.255.255.250 SSDP 1066 NOTIFY * HTTP/1.1

Are the offending parts from Wireshark. 172.16.99.x is the local end of the network with UPNP DISABLED via the router GUI...

*ANY* time this kicks up I get packet loss over the VPN.

If I sit and ping a device on the LAN side of the VPN it gets around 50 packets, then 2 fail, then works ok again. I'll try and do some more testing later in the week but this is becoming a pain... Another botched firmware perhaps (after all the issues I've had in the past it wouldn't surprise me.

The VPN endpoint is a 2820 running 3.3.7.6_232201