We intend to use MT in a big way as a firewall/QoS/Access Control solution for our clients.
Problem is that the PPPoE client in Mikrotik behaves strangely. The setup is as detailed.

MailClient has a Valid IP and is routed through MT BOX-A.

MT BOX-A gets connected to MT BOX-B as a PPPoE client using the PPPoE client built into MT. The MTU and MRU are set to 1492 all around.


Now when MailClient accesses his ISP's Mail server through POP3 he is able to authenticate his account but all comes to a halt as soon as downloading of first mail starts.

We have deleted the first mail, in case it was malformed, on the ISP's Mailserver but now irrespective of what is the first mail as soon as transfer is initiated all POP3 activity times out. The Mailclient is FetchMail.

As soon as we set MT BOX-A to Connect to MT BOX-B through a STATIC IP instead of a PPPoE client all problems disappear!!

Obviously the default IP ROUTES are being adjusted as the WAN link changes from PPPoE to Standard IP assigns.

Any ideas why is MT PPPoE client behaving like this with POP3 or is it to do with MTU/MRU size??

I cant keep experimenting continously. So need some fairly reliable solutions.

the first thing is that if the problem is just with POP? have you tried browsing or other downloads on the system where client is installed.

if the problem is with all stuff, then obviously its an MTU or MSS problem.. otherwise u'll have to check the pop3 port..try monitoring the realtime connections of that ip through torch.

regds,
Asad

the first thing is that if the problem is just with POP? have you tried browsing or other downloads on the system where client is installed.

No problems as of now with anything else on the client.

if the problem is with all stuff, then obviously its an MTU or MSS problem.. otherwise u'll have to check the pop3 port..try monitoring the realtime connections of that ip through torch.

Done that. This IP when it connects to the remote IP (on the POP3 port) shows some intital traffic and then nothing. Whereas when monitored with a NON-PPPoE wan connect (Standard IP/Gateway system) then traffic flows real smooth to the tune of 1Mbps burst and then sustained traffic of @ 300Kbps whereas with a PPoE wan it trickles at an intial 25Kbps and then peters down to zero.

I had the exact same issue with 2.9.38 and pppoe clients. I had to remove queue tree and everything worked great. Customers still have their queue set by radius. MSS and MTU were all set. I went through everything on the router. After hours of testing it all came down the queuing.

I had the exact same issue with 2.9.38 and pppoe clients. I had to remove queue tree and everything worked great. Customers still have their queue set by radius. MSS and MTU were all set. I went through everything on the router. After hours of testing it all came down the queuing.

We use queuing for opening to port80 as we have a NetAppliance Cache server. Could this be the reason? What queues were you using?

Its only when i have queue tree enabled that the problems start to happen. I had to disable everything in the queue tree to get it to work properly. Is your NetAppliance on a queue tree and being mangled or just a regular queue? If you setup a queue for the NetAppliance and disable all your queue tree, that should fix you up. I don't have any caching servers that i use on my network. But if i was you i would start there and see if that helps out.

I was about to pull my hair out until i found this solution. I emailed MT and they said proxy-arp which i had on at the time i disabled that and still had the issue. After tons of hours of research, it came down to that.

I would also recommend upgrading to the newest routeros also.