PPPoE client, bridged modem and switched routerboard ports

I have a routerboard 493G and it doesn’t seem to be routing over my PPPoE connection.

I have:

ADSL modem (bridged) ----- RB493G Ether2
Server1-------------------- RB493G Ether3 - slave to ether2
server2-------------------- RB493G Ether4 - slave to ether2
server3-------------------- RB493G Ether5 - slave to ether2

I need all servers to be switched between each other to benefit from gigabit and use they use the PPPoE connection as the gateway, In the current setup they all switch fine and can connect to the router.

Starting at the PPPoE -

name="PPPoE" max-mtu=1480 max-mru=1480 mrru=disabled 
      interface=Bridge-LAN user="username" password="password" 
      profile=default service-name="DSL" ac-name="" 
      add-default-route=yes dial-on-demand=no use-peer-dns=no allow=pap,chap

I have my PPPoE connection setup and it authenticates, gets assigned IP and adds in the route and the default route. however once established there’s no routing out of the WAN, routeros reports the nexthop is available however a ping from the terminal comes back with timeouts. I have added in the nat masquerading (although it shouldn’t matter when it cant even ping out of the PPPoE interface. I have just upgraded the routeros to 5.0rc5 and it is still occurs.

Something odd i noticed also was i had a ping in progress to the nexthop with the PPPoE client disabled. i enabled it and it established a session. in this process i received one reply from the nexthop and then the subsequent pings timed out. Seems quite weird.

I have a RB750G that was being used as the exact same setup, i can disconnect the ADSL bridge from the RB493G and connect it back into the RB750G and it works as expected. I have no idea why this new router does not want to work !

Why you run PPPOE client over your local network bridge interface?

Connect ether3,4,5 to master port ether2 and ether7,8,9 to master port ether6 (493G has two separate switch groups thus cannot be switched all together)
Create the lan bridge, attach master ports to it (ether2 and ether6) and assign an IP to this bridge interface.
Run PPPOE client over the non-bridged ether1 interface (wan) without assigning an IP to it.
Use the usual masquerade nat rule to pass output packets through pppoe interface.
Configure DNS settings to “allow remote requests”.

That’s all. I hope I didn’t forget anything. :slight_smile:

-My switch is currently setup that way (im only using ports 2-5 for the moment on this routerboard)
-The logical bridge interface has ether2 assigned to it.
-I haven’t tried assigning the physical interface for the PPPoE connection, although on my other routerboard i use the bridge interface and it works.
The DNS and the masquerading i believe are a non issue, masquerading shouldn’t be required when trying to communicate directly out of the public interface. Although they are both setup regardless.

Edit: I am powering the unit via ether1

The issue was with incorrect password :frowning: so it was my fault… shameful…

I didn’t realize until i ran a ping and reestablished the connection, there was an icmp reply of admin prohibited followed by timeouts. It was only at this point i realised…

The ISP provides a ‘playpen’ for incorrect credentials which was what was throwing me off, they prevent all communication except to a webserver which they redirect http traffic to in order to display a page advising of the incorrect credentials. Whatever happened to a simple PPP auth failure!

Sadly it makes sense from their perspective - if you’re getting 20 hours worth of support calls a month about PPP failures for invalid credentials that turn out to be typos then it makes sense to buy a $200 web server to deal with people in an automated fashion.

Makes it harder to troubleshoot when you know what you’re doing, but that isn’t their time they’re losing.