Multilink PPPoE bundle failure

dcabanaw · Thu Jun 28, 2012 7:20 pm

I'm having some multlink issues with several installations using RB750 and RB750G running 5.17. Periodically (1-2 times a day) the bundles will drop (as seen by our LNS) but the Mikrotik router will not react to the dead bundle and still show all the links up. At this point no traffic will pass and I have been using a script to check the gateway and restart the pppoe client if it is down... While this works, it is a less than an ideal solution.

Port 1 normally goes to the customer internal network, and ports 2+ are used for DSL modems. Modems are stable, have good sync. I have tested the connection without multilink and it is very stable with several weeks uptime. I have tested with a router flashed with tomato/MLPPP and this issue does not happen. PPPoE terminates to Cisco 7206vxr(mpe-g2) with IOS 12.4(24)T1.

example interface config:

/interface ethernet
set 0 arp=enabled auto-negotiation=yes bandwidth=unlimited/unlimited \
    disabled=no full-duplex=yes l2mtu=1520 mac-address=00:0C:42:AD:20:2A \
    master-port=none mtu=1500 name=ether1-lan speed=100Mbps
set 1 arp=enabled auto-negotiation=yes bandwidth=unlimited/unlimited \
    disabled=no full-duplex=yes l2mtu=1520 mac-address=00:0C:42:AD:20:2B \
    master-port=none mtu=1500 name=ether2-wan1 speed=1Gbps
set 2 arp=enabled auto-negotiation=yes bandwidth=unlimited/unlimited \
    disabled=no full-duplex=yes l2mtu=1520 mac-address=00:0C:42:AD:20:2C \
    master-port=none mtu=1500 name=ether3-wan2 speed=1Gbps
set 3 arp=enabled auto-negotiation=yes bandwidth=unlimited/unlimited \
    disabled=yes full-duplex=yes l2mtu=1520 mac-address=00:0C:42:AD:20:2D \
    master-port=ether2-wan1 mtu=1500 name=ether4-local-slave speed=1Gbps
set 4 arp=enabled auto-negotiation=yes bandwidth=unlimited/unlimited \
    disabled=yes full-duplex=yes l2mtu=1520 mac-address=00:0C:42:AD:20:2E \
    master-port=ether2-wan1 mtu=1500 name=ether5-local-slave speed=1Gbps

/interface pppoe-client
add ac-name="" add-default-route=yes allow=pap,chap,mschap1,mschap2 \
    dial-on-demand=no disabled=no interface=ether2-wan1,ether3-wan2 max-mru=\
    1454 max-mtu=1454 mrru=disabled name=isp password=password profile=default \
    service-name="" use-peer-dns=yes user=mppp%user@isp.net

I have been trying to debug the connection and setup logging on all pppoe and sending it to a syslog server. Unfortunately the syslog server was not getting all of the data... I have started logging again directly to disk and have setup a script to ftp in and archive the logs as well as keep enough disk space free for new ones. In the logs I was able to parse, just before the connection goes down, there is a long string of LCP EchoReq with no replies being sent.

Any ideas as what the cause could be? In the meantime I will keep the debug log for ppp/pppoe going and hopefully find some more information.

Thanks!

dcabanaw · Sun Jul 01, 2012 10:18 pm

I had another failure and found the following:

- One of the Ethernet interfaces was no longer receiving traffic from pppoe (pinging the modem worked fine as I set each one up on their own network)
- The logs show the PPPoE client only sending LCP EchoReq and EchoRep on only one of the bundles.
- the PPPoE client would still show 2 links up.
- disabling and re-enabling the Ethernet interface that was no longer participating was enough to get the connection back up.

I find it odd the the PPPoE client would not detect that the bundle was down, especially since it was only sending requests over one of the links...

I have attached a sample of the log which I have marked up with comments

giesen · Tue Jul 10, 2012 9:24 pm

I'm seeing the same issue with 5.16 and 5.17. I'll eventually end up with more than 10 sessions connected when there should only be 3 (2 links + bundle).

Multilink PPPoE bundle failure

Multilink PPPoE bundle failure

Re: Multilink PPPoE bundle failure

Re: Multilink PPPoE bundle failure

Who is online