Started a new topic since this was getting off-topic from my BGP post. I have 3 core routers, all mikrotik, and 2 border routers, both cisco 7000 series.
Running OSPF internally, BGP externally and IBGP Between the borders.
Now on to the problem… (Any ips here are fictitious)
Both borders advertise a 0.0.0.0 via ospf, but the cores(mikrotiks) will only have one at a time in the table…except sometimes, i would see the 0.0.0.0 route pop up with “reachable border 1 x.x.x.x, reachable via border 2 x.x.x.x”…but then it goes away and lately it will only show it reachable via one, usually border1.
Email support. If both LSAs are there with the same metric as type 2, they should both get installed into the routing table. If they’re not, that’s a bug.
I think I solved the problem. Something kept bothering me about the forwarding address being in one but not the other.
I did some research about that and found that by default a cisco has OSPF enabled on ALL interfaces when you enable OSPF. You then need to turn off the ones you dont want it on. Since this “forwarding address” was the ip of one of those interfaces…i figured let me try turing it off on the links (Since it shouldnt be on on them anyways!).
Then I began thinking, its a GOOD thing the mikrotik doesnt add that route. Because it would say 0.0.0.0 is reachable via a network the tik doesnt have a connection to. Thanks Mikrotik people!!!
So i disabled it on those interfaces…and wham, now i have 0.0.0.0 forwarding address for both LSA’s and it installed both defaults.
I know this isnt a cisco forum, but for anyone mixing mikrotik and cisco here are the cisco commands to turn off OSPF on interfaces and eliminate that forwarding address that stops the mikrotik from keeping both routes:
router ospf 1 (enter ospf process config mode))
passive-interface FastEthernet2/0 (turns off OSPF on FastEthernet 2/0)
passive-interface FastEthernet2/1 (turns off OSPF on FastEthernet 2/1)
…
Just keep adding the interfaces you DONT want ospf to run on.
FYI: When you have ciscos Advertising to tiks, make sure their OSPF config is as IDENTICAL as possible. The smallest difference can make one route not get into your tiks table (Yet still show the same LSA info so you will be stumped).
Basiclly what seemed to be causing the issue on mine was one border had "passive-interface default and and “no passive-interface GigabitEthernet0/0” and the other border had all interfaces turned off seperatly…instead of the default. Logic tells you this shoud accomplish the same thing. It did, but for some reason the tik doesnt think they are the same then.
Possibly, this had nothing to do with it and just the change restarted the OSPF process on the cisco and that restart fixed the issue. Not sure yet, but will keep posting to the thread if it starts happening again.
I have multiple Cisco’s, all using ‘passive-interface default’ in the OSPF configs, and the MikroTik’s will still lose the default route. It’s as MikroTik problem for sure.
