OSPF network creating two interfaces and other OSPF question

We have recently started looking into using OSPF for our small network. I have few questions about OSPF configuration.
We have almost all ROS version 3.30. First thing is that it seems that documentation http://wiki.mikrotik.com/wiki/OSPF-reference and actual ROS differs a lot. I.e. documentation states: "Since v3.17 it is possible to run multiple OSPF instances. General OSPF configuration now is moved to instances. " but my router (V 3.30) does not know command /routing ospf instance.

However I’ve encountered other major problem that I can not handle. When I add OSPF network on one of our main routers it automatically adds two OSPF interface definitions for the same interface (or only one if there is already one created manually). The router gets connected to other router on that network and they create the adjacency. They even do excange LSAs but the every 10 secs router logs error like this:
Database Description packet has init bit set in the middle of an exchange.
And the routes are removed from routing table. After 10 secs the routes get imported again and it all goes again. I am suspecting the two interfaces running two OSPF processes but I can not disable one of the by any means. And I am not able to recreate the same behavour on any other router. The truth is that the configuration of tat router is a little complicated with bridges multiple IPs on interface (even from same range) but I tried all these factors on other V3.30 router and it creates only one OSPF interface.
Last question is about running OSPF in some safe way. I know that quagga has option to import routes into different routing table. There is this option mentioned in the reference for mikrotik but there is no such option in the actual ROS. I am using route filters to mark all routes in ospf-in chain with some unused routing mark. I would like to see all our routers with proper routing tables created by OSPF before I enable it for real :slight_smile:

It knows it if you are using RouterOS v4, or RouterOS v3 with routing-test package.

However I’ve encountered other major problem that I can not handle. When I add OSPF network on one of our main routers it automatically adds two OSPF interface definitions for the same interface (or only one if there is already one created manually).

This happens when you have two or more IP subnets (assigned IP addresses with different network fields) on the same interface. The solution is specify OSPF network with mask that only matches one of those addresses.

I know that quagga has option to import routes into different routing table. There is this option mentioned in the reference for mikrotik but there is no such option in the actual ROS.

Use RouterOS v4, or v3 with routing-test package :slight_smile:

It seems I will try the routing-test package :slight_smile:

I did specify the network with the mask but both IPs are in the same subnet (for some historical reasons, the router did get IP of one other “server”). More strangely when I try to recreate the problem on other ROS it creates only one interface no matter how many IPs I specify. Its the same HW and same version (3.30). I’ll try the routing-test package to see if it helps though…

Try putting the main IP that you want OSPF to speak on, into routing->ospf->networks as a /32 and see if that makes a difference.

It will not create two interfaces if there are two (or more) addresses with the same netmask, on the same interafce (e.g. 10.0.0.1/24 and 10.0.0.2/24). In this case one address is considered “primary” and the rest are “secondary”. Try adding such addresses on plain Linux and see their description (run ‘ip addr show’). MT OSPF ignores such “secondary” addresses.

Also, awsmith, putting /32 in networks will have no effect, unless there really is a matching /ip address with /32 netmask. The “network” address, not the “local” address is used by OSPF to check whether to run the protocol on an interface.

Atis, I do not know, what is the reason, but it does create two interfaces with identical options. I know that quagga on linux would ignore such “secondary” address, but this is mikrotik and I do not know how it exactly behaves. Maybe the problem is somewhere else but I can not find it.
Problem is that I was not able to recreate this on any other RB and the RB I am speaking about is “production” grade and I can not experiment with it, however I need the OSPF to run on it if I should deploy it network-wide.
Next thing I want to try is to duplicate all configuration on other RB to see if it happen there too. I’ll post any other findings.

I have two ADSL modem with PPPOe connection. I use RB 750 (5 LAN Port). Want the two connection in two interface in router and out from a single interface for user.

Please inform me how to configure

Mithu

I have partially solved my issue. The problem has been linked to that two IPs on one interface. I had to remove the network and brodcast option on the “less” important IP address. The settings have been correct but I think they forced the OS to run two OSPFs. However the settings reappeared but OSPF is still running. I do not know what would be wrong with these options as they are correct. I would consider it a bug or a glitch in matrix :slight_smile:
Important is that OSPF is running now, I can evaluate its functionality on our network and preferably I would upgrade mikrotik before using OSPF in production…