Enabling MPLS breaks OSPF

Hello all,

I’m stuck on a very frustrating issue here.

I’ve added a couple of CCR routers to our Cisco network in an attempt to expand our MPLS network.
Our main routing protocol is OSPF, which works fine.
I’ve completed that setup and there are no issues.
Approx 15000 routes, routers humming along fine.

IPs changed to protect the innocent :slight_smile:


Router 2

/ip address
add address=10.85.2.4/24 interface=ether1 network=10.85.2.0
add address=113.6.25.252/32 interface=loopback0 network=113.6.25.252
add address=113.6.25.62/29 interface=vlan513 network=113.6.25.56
add address=113.6.25.26/29 interface=ether3 network=113.6.25.24

/routing ospf instance
set [ find default=yes ] router-id=113.6.25.252
/routing ospf interface
add authentication=md5 authentication-key=sssssss interface=vlan513 network-type=broadcast
add authentication=md5 authentication-key=sssssss interface=ether3 network-type=broadcast
add authentication=md5 authentication-key=1234 network-type=broadcast passive=yes
/routing ospf network
add area=backbone network=113.6.25.56/29
add area=backbone network=113.6.25.24/29



Router 1

/ip address
add address=10.85.2.3/24 interface=ether1 network=10.85.2.0
add address=113.6.25.251/32 interface=loopback0 network=113.6.25.251
add address=113.6.25.25/29 interface=ether3 network=113.6.25.24

/routing ospf instance
set [ find default=yes ] router-id=113.6.25.251
/routing ospf interface
add authentication=md5 authentication-key=1234 network-type=broadcast passive=yes
add authentication=md5 authentication-key=sssssss interface=ether3 network-type=broadcast
/routing ospf network
add area=backbone network=113.6.25.24/29



As soon as I enable LDP, everything goes crazy;
It’s currently disabled.


Router 2

/mpls ldp
set lsr-id=113.6.25.252 transport-address=113.6.25.252
/mpls ldp interface
add accept-dynamic-neighbors=no disabled=yes interface=vlan513
add disabled=yes interface=ether3



Router 1

/mpls ldp
set lsr-id=113.6.25.251 transport-address=113.6.25.251
/mpls ldp interface
add disabled=yes interface=ether3



I lose OSPF connectivity, I lose router connectivity, it reports strange OSPF errors like;
Discarding packet: no neighbor with this source address
RouterId = 113.6.25.251
source = 113.6.25.25


As soon as I disable LDP, everything starts to work again.

MT support have advised me to disable OSPF for 60s when bringing up LDP, then re-enable it, however the same thing happens.
Has anyone encountered this ? or am I doing something wrong ?

I’ve been using the link as a guide;
http://wiki.mikrotik.com/wiki/MPLS_Lab_Setup

Thanks for your time.

I think that you need your transportt-address to be the IP on eth3 not your loopback

Nick

A few things that stick out:

  • You need an OSPF network that encompasses your loopback IP address e.g. 113.6.25.0/24
  • Add an OSPF interface “all” that is set to passive, this combined with the above will advertise your loopback as an intra-area route.
  • Manually specify OSPF interfaces for your LDP enabled interfaces set to broadcast or point to point.


    Using your loopback address as the OSPF router-id is great
    Likewise with LDP lsr-id and transport address.

If you get stuck, I am happy to help out on the phone.

Could you send me a PM or email me gweebo at gmail dot com.
Happy to pay you for your time :slight_smile:

Interface “all” is that correct ?

Hi n21roadie.

You could just specify your loopback as a passive interface, but it is easier if you set the interface “all” to passive. This will then advertise any interfaces with an IP matching one of your “networks” as an intra-area route.

I am trying to get to grips with OSPF config?
What if OSPF Loopback interface is Dynamic and Passive?
And what about Wlan network type = PTP
and ether is network type = broadcast,
Does interface “all” to passive still apply

Yes it does.

We use this extensively even on wireless ptp and ptmp links.

Is it possible to post a sample of how you have setup OSPF Interfaces for both PTP and PTMP,
I must admit being somewhat baffled by how a loopback interface is dynamic in OSPF interface?

Having “all” interfaces as passive cannot be correct or is it?
Will this not effect the different network types being used by the interfaces (Wlan + Ether) and they are not passive?

It is correct. You specify the interfaces you need to be broadcast/p2p manually. The “all” interfaces passive config catches other interfaces e.g. loopbacks, pppoe, customer links that match the “networks” and sets them as passive so those prefixes get advertised as intra-area routes.

Ok - let me get this - You set “all” interfaces as passive in /routing ospf interface but then other interfaces such as Wlan and Ether which are also listed in /routing ospf interface their setting will override the “all” interfaces as passive setting?

Correct.

When I add “all” interfaces as passive, Area=“Unknown”, State=“Down”, Network Type=“broadcast”

Also should I add OSPF Loopback interface, at present its dynamic in /routing ospf interface

OK I tried interface “all” that is set to passive and this caused ospf to be confused,
I am convinced this is giving conflicting instructions to OSPF , first all ospf interfaces are passive and then for ether and Wlan ospf interfaces are not passive, the active interfaces did not over-ride passive setting and as mentioned OSPF got confused.