Issues attempting to use a ubiquiti Edgerouter ER-4

[#1]

Hi all, 

Hopefully someone can help me. I feel a bit like i'm losing my mind

I am a new customer to plusnet having moved from BT. I used to use a ubiquiti USG on BT with a netgear modem. Earlier this year i noticed my sync speed dropping. This correlated with poor ping reliability - around 0.3% lost pings on average from my PC to openDNS and google. After much to-ing and fro-ing with BT and openreach, and many engineer visits, they claimed there was no issue with the line and i eventually discovered the issue appeared to be the router (not the modem). The connection was fine with the homehub, but sure enough dropped all the time with the USG. This was causing my sync speed to drop. The USG had been find for the 6 months prior to this. 

To cut a long story short, we were pretty unsatisfied with the support from BT and the customer service and had pretty much already decided to jump ship. I signed up to plusnet and just used the supplied hub for a few weeks, but now desire to return to a more flexible setup (particularly to deal with buffer bloat and to use QoS) and so decided to address the issue with my "faulty" router. I have RMA'd the USG, but i figured i'd use this as an excuse to get an Edgerouter ER-4, which has much better throughput with QoS on due to it's more powerful processor. 

The ER-4 turned up today and i have just finished getting it all setup. To my astonishment and dismay, i am having exactly the same reliability issues as i was with the USG on BT. I just ran the basic setup wizard on EdgeOS, put in my PPPoE details and left everything else on default. I tried using the edgeOS toolbox to ping openDNS too and found the same proble: after 500-ish pings, 4 had been lost. 

Am I missing something incredibly obvious here? Has something changed in the way VDSL PPPoE connectivity works recently that has rendered the default config broken in some way? Do i need to configure a VLAN ID or something like that?

Any help you guys can offer would be greatly appreciated - I'm tearing my hair out over this whole fiasco!

EDIT: I should probably add that my modem is current a huawei openreach model and that i only thought to check the router after the switch to plusnet, so did establish that i had exactly the same problem on plusnet with the USG as I am having now.

[#2]

Moderators Note

This topic has been released from the Spam Filter

[#3]

An additional piece of info:

I found a thread on these forums recommending a change to the MTU setting for eth0 and the PPPoE interface. This was: 1500 for PPPoE and 1508 for eth0 to allow for 8 bytes of PPPoE encapsulation. The poster seemed to have exactly the same problem as me, suggested that the modem was not happy fragmenting ethernet packets at lower MTU values. However, in my case this has unfortunately not fixed the connection. I'm still getting a failure of around 0.3% of pings to open DNS (32 bytes), but to google overnight the rate was 0.04%. 

[#4]

I don't see what your USG router has to do with your sync speed - that is set purely by the modem - which Netgear model is it?  If you have problems with your sync speed dropping then you need to try a different modem not router.

[#5]

Well, quite. You're exactly where i was a month ago. Unfortunately, since then i have more or less proven that the issue is not with the modem. Or at least, not with the modem in isolation. To answer your question - apologies if i forgot to mention this in my original post - i have tried two different modems: openreach and netgear. Both the plusnet hub, or the bthomehub, coupled with either modem, yields a perfectly stable connection with no detectable packet loss. Either modem connected to either the USG or the ER-4, yields an unreliable connection with at least 0.3% packet loss. 

I have come to the conclusion that there are indeed ways that the router can influence the perceived reliability of the connection from the DLM's point of view, and that i'm just not knowledgeable enough about telecoms or networking to know what all of these ways are. For example, the additional thread i alluded to that i found on the plusnet forums described just such a mechanism. The modem is asked to use a different MTU when preparing packets for the interface to the router, than it has received on its own external connection. This involves fragmenting packets still further. If the modem has some undefined behaviour associated with this that causes it to drop packets, the PPPoE interface will know. This information may well feed into the DLM system - why wouldn't you pay attention to the PPPoE interface telling you packets are going missing? - which could cause a reduction in sync speed. 

But tbh, i could speculate all day without hitting upon the right answer. The facts are the facts. Out of 4 routers and 2 modems, there is one constant: any config that involves one of the UBNT routers experiences packet loss which negatively impacts DLM. Any config that involves either the BT or plusnet hubs, doesn't.