I’ve tried everything. I have it working on another router/link with similar config. Works great so far on that one. Only difference is it is a cable modem this one (non working) is FIOS. 1437 pings fine, 1438 times out. That’s with or without the Don’t Fragment option checked. 6 hours straight into this and still nothing. :-/ Any ideas?
I can post my config but the other site/router is working fine with this on it’s cable connection same RB450G. RB1100 on the other end. I’m just wondering if that 1437 MTU is a magic number that can shed some light.
Thanks
According to Verizon, the Verizon FIOS uses a smaller MTU than cable.
I can ping full 1500 byte packets without fragmentation over the Fios line. Are you sure that’s not from the days of PPoE FIOS? (Mine is DHCP) Should I lower the MTU of the public interface for kicks? 1492?
If you have the DHCP FIOS, then yes it should be 1500… have you tried the router at 1500 with MRU, MTU, etc all at default? It should work.
Yes it’s all on default, but i’ve lowered it and even raised it above 1500 with no luck. If i turn off MRRU and just do reg L2TP its fine. Just can’t pass more than 1460 bytes. Just for kicks I tried setting MTU and MRU of L2TP client interface all the way down to 800 with MRRU at 1600 and was only getting 4% packet loss with unfragmented 1500 byte pings. Can’t be very efficient! I forgot to mention pings do go through with normal settings. It just gets about 90% packet loss and can’t pass any traffic at all!
Is there anything else to test besides ICMP pings? Really want to get to the bottom of this.
Also worth mentioning, I am running 5.7 on everything. RB450G on FIOS end. RB1100 at Data center in another town. RB450G is connected directly to the FIOS ONT, has been for quite some time.
Ok so the problem appears to have solved itself. I was getting about 4% packet loss on larger packets pinging to my L2TP data center server. It seems that enabling MRRU over ppp connections requires a very stable link. The packet loss subsided so I decided to give it another try and now it is working great! 1500 byte packets flow over the VPN link beautifully!
Scratch that. Worked for about 6 hours then went back to the way it was in post #1. I didn’t change anything at all and wasn’t even here when problem started again. The strange thing is there is still no packet loss and traceroutes to VPN server appear to take the same path over the internet as they did when it was working
having the same problems only 4 years later 
if you need 1500 byte package, whole path need to support increased MTu on all interfaces that transfer that packet.
in routerboards you can do that by increasing L2MTU value. bu first check Max-l2mtu value.
Note that it doesn’t work if you change it only on endpoints, whole path, all involved interfaces need to support that,