I also posted all the details in this serverfault link.

http://serverfault.com/questions/661693 ... -full-mesh

Basically I have 4 mikrotik routers set up in a full mesh, with OSPF, MPLS, and iBGP between them. Everything works fine internally. But all the networks that have one IP on one of my routers and the rest external will not ping from any routers besides the one that is directly connected to it.

Any ideas on why this issue may be going on?

Possible issues:

Is it possible that the upstream is filtering ICMP except from certain addresses?
Which address is the ping originating from? Is it being forced to a specific value?
Can you see the ping traffic egressing towards the gateway as expected?
Can you ping external IPs such as Google DNS 8.8.8.8 from the same device and does that traffic pass through the gateway?
Do you have any forwarding filters in place?

So basically the OSPF was putting in the routing table the routes with no preferred source address.

When the router would originate packets, it would pick the best route, and since there was no preferred source address, it would pick an ip address on the adapter, the private one. The only public ip I had on each router was the loopback ip. All the other IPs I had in the private range.

I could never figure out a way of forcing a preferred source address on the machines, besides using mangling rules which I didn't want to do. So I tried again making every interface have public ips. I fooled around with a few settings, finally with making sure the downstream router had a default route as well, and having ospf redistribute connected and static routes, the combination made it work. It is using more public IPs then I wished (would have preferred to have 1 public IP per router, but now need 1 + 1 /30 per additional router in the mesh. But it seems to be working.

I also posted all the details in this serverfault link.

http://serverfault.com/questions/661693 ... -full-mesh

Basically I have 4 mikrotik routers set up in a full mesh, with OSPF, MPLS, and iBGP between them. Everything works fine internally. But all the networks that have one IP on one of my routers and the rest external will not ping from any routers besides the one that is directly connected to it.

Any ideas on why this issue may be going on?

A quick question: are you natting networks on these routers?

No, I wasn't wanting to run any NAT. Trying to keep these routers as lean and quick as possible.

So I was able to get it working using global addresses for all interfaces, instead of some interfaces using a private ip address.

I've set up my ipv6 space the same way, and I'm having an issue. Which maybe is related to the previous issue.

When I look at the routes R3 (or R2 or R4) receives from R1, they all show like this. All incoming ipv6 BGP routes show as unreachable, using 2607:1a0f:8000:101::1 as the gateway.
The thing is, you can see in that image that the ip is pingable. The OSPF route is installed a few lines below, and shows as reachable. Likewise if I just do a ping to that gateway address (2607:1a0f:8000:101::1), it pings just fine. But all the BGP routes coming in show as unreachable using that same address as the gateway.

This is a problem for me because I need to turn around and announce that whole routing table in another BGP session. And as all the routes show as unreachable the BGP session doesn't announce them.