Hello!
Need help with debugging OSPF stub on Mikrotik.
Ready to provide output of any commands or test something in lab or “in the fields” =)
Scheme is following:
MPLS cloud - Mikrotik as PE (main ospf/MP-bgp/LDP) - and customer on one interface, OSPF peer in one of VRF’s, stub area.
RouterOS 6.11, RB2011UiAS
After i set same in lab using mine router as customer equipment i saw that Mikrotik isn’t announcing default.
I can see it in “routing ospf lsa print” but not in customers database.
When i switch to default area type both sides - default is announced and received as planned.
On cisco i don’t even need to do anything, it just announces default, because router is ABR (MPLS-backbone).
On mikrotik i’ve found “distribute default” option but no effect for stub area type.
Any tips appreciated!
Configuration of Mikrotik in trouble part:
ip address print
172.20.214.1/29 172.20.214.0 eth2_RRL_Man
Maybe this is a stupid statement, since I am not that proficient in OSPF and BGP, but AFAIK a stub area has only one entry/exit point by definition, and that is the ABR.
So in my view it can not and should not provide any route except those belonging to the stub because the link via the ABR is the only entry/exit for that area by definition and a distinct external route other then the ABR makes no sense from BGPs point of view for a stub area (and is discarded).
Maybe a nssa is the solution in your case, which allows redistribution of non-BGP routes from a stub, in your case the one obtained via OSPF.
In topology like this (MPLS with L3VPN and OSPF peering in that L3VPN) router is ABR and should function accordinlgy.
Any MPLS in such situation equals to backbone area 0.0.0.0
So behavior is unlogical.
@dallas:
nope, as i said above, i could not make it working. So i switched area type to “standart” and so did the customer.
I think it is software bug. But i’m quite shure that it isn’t high priority bug.
I tried tones of version. I think when they programmed the stub and nssa. They just forgot to have the default route to propagate. So, yeah. I would call it a bug. I want to use mpls with with redudancy. How can I use mpls and do what I wanted with ospf. RIP or something else? I need to have areas so I dont have 4000 routers. I am too big to use mpls. Because I run out of memory and cpu maxes out. If I have ospf areas. Then my routing table is smaller so the mpls forward table is much small. Which means less memory.
I believe I got it to work. On the ABR router. You must setup a /rout ospf network to the backbone. Thats it. Then the 2nd area will get a default route. Will do some more testing and learning.