I've used my router (a little PC running a Red Hat Linux) with Rogers for over 25 years. Now I've switched to Bell to get FTTP (AKA FTTH).
Here are some quirks (AKA discoveries) Bell supplied a Giga Hub modem / router / access point / landline / ... There are several ethernet ports on it. - One goes directly to the TV set-top-box that Bell supplied. I don't know if I could put it behind my router. - three are 1G - one is 10G (I think) but I use as 2.5G with my router-PC. My promised speed is 1.5G down and 960M up, so 2.5G is sufficient There are two ways to use your own router with the GigaHub - the "Advanced DMZ" passes traffic directly to your router. You just need to tell the GH the MAC address of your router - the PPPoE passthrough method. The Advanced DMZ seems like a great idea BUT your router and the GigaHub share the same IP so their might be conflicts. For example, both will be doing NAT and thus might assign the same ports for conflicting purposes. I don't see how this can work (but failure would be rare and confusing). The PPPoE passthrough gives your router and the GH separate IP addresses! But it means your router needs to do PPPoE. Apparently many consumer routers cannot handle PPPoE at this speed. Good news: my little PC can. I can still use the GH's extra ethernet ports and WiFi AP, but they are outside my LAN (not behind my router). That has been useful for testing/diagnosis and for accessing the GH's control panel. I could probably add a route to my router to allow access to the GH's control panel, but I have not done so. My little PC is a TopTon Celeron N5105 box with four 2.5G ethernet ports <https://www.aliexpress.com/item/1005004360072281.html> There are newer alternatives. I run Fedora 38 as the OS. Other folks choose pfSense or OpnSense, but I haven't needed to learn about those. PPPoE nicks a few bytes from each packet. That should be fine: PMTU Discovery should handle it, but it doesn't. Some internet sites fail. Cure (at least empirically): add MSS clamping to the filewall. Here is the "policy object" that I added to firewalld as file /etc/firewalld/policies/policy_int_to_ext.xml: <?xml version="1.0" encoding="utf-8"?> <policy priority="100" target="ACCEPT"> <ingress-zone name="FedoraServer"/> <egress-zone name="external"/> <rule> <tcp-mss-clamp value="1452"/> </rule> </policy> I don't remember having to do this with my (ancient) VDSL connection to Vybe.net. Not sure why. Theory: Roaring Penguin PPPoE seems to do MSS Clamping itself. Other possible cure: there is an obsolete and never adopted RFC that specifies how to negotiate slightly-jumbo packets with PPPoE. There are rumours that Bell supports this. a (1500 + 8) MTU would eliminate this whole problem. I would think RP PPPoE would support it but I don't know how RP PPPoE fits into Fedora 38. --- Post to this mailing list [email protected] Unsubscribe from this mailing list https://gtalug.org/mailman/listinfo/talk
