I have an RB1100 running v4.10 acting as a PPPoE server for several apartment blocks. A few days a go a problem crept in where my nightly backup (an export) didn't complete and the router stayed at 100% CPU. I tracked the issue down to the following:

There is a /24 for the PPPoE users. I'm using OSPF, the PPP addresses are all in their own area and route summarisation works a treat. I now have 144 OSPF interfaces, 134 are PPP interfaces. When I view them in the winbox GUI everthing seems fine. When I jump onto the console and run "/routing ospf interface pr" I get from 0 to 136 listed and then the list starts again at 4 and repeats indefinitely. If I run an export in for this, it just hangs after the first 4 lines are displayed and the router stays at 100%.

I tried kicking 10 users and running the print command again, all fine, 126 lines reported, no looping. Is there a max number of OSPF interfaces the system can handle? Is there something I can do so that OSPF doesn't 'see' these as interfaces but I can still have their /24 advertised out?

I have an even larger site (over 300 users) that I'm looking to port to a PPPoE setup (for accounting) that will be using OSPF as well.

Cheers,

Dave..

on your ospf network. do you have 0.0.0.0/0 or do you have the specific subnet listed? it sounds to me like you have 0.0.0.0/0 and so it tries to run a ospf session on every single interface including your pppoe.

cheers

As for the OSPF network tab, I've only got the /24 for the PPP users, the router's loopback address and a few /28 used for backhaul.

So, yes, it's defining every PPPoE interface as an OSPF one, but of type point-to-point (or passive if I include the definition for all-interfaces passive).

The problem seems to arise from the number of them when listed on the command line. OSPF itself doesn't seem to have any problems.