I am trying to migrate an EIGRP network to OSPF. I thought I would do this by enabling OSPF across the network with admin dist of 180, then drop the AD on the Area 0 routers to get OSPF working in the core, then do the same in the leaf areas. What I found was that when I droped the admin dist of OSPF to 80 on the core/area 0, OSPF was also active for all routes on the leaf areas, which was a surprise. I thought OSPF and EIGRP were ships in the night routing, so changing admin dist locally on one router, shouldn't affect the EIGRP routing on another router.
I give a specific example below.
R1---area 1---R2----area 0-----R3---area 2---R4----network10.1.1.0
1. EIGRP is running across all routers (with no redistribution)
2. Configure OSPF on all routers with areas as shown in diagram, with admin dist 180.
3. Check to confirm that all routes are still in EIGRP.
4. Set admin distance to 80 for OSPF on R2,R3.
At this stage I expect the route connected to R4 (10.1.1.0) to be in OSPF on R2,R3, but in EIGRP on R1. What I find is that R1 has the route in OSPF.
Debugging shows that when I set the R2,R3 OSPF distance to 80, R1 says there has been a route change, it does the EIGRP DUAL query for the successor, but R2 says I don't have an EIGRP route in my main routing table for 10.1.1.0 (the route is in OSPF), so R2 says I provide a successor, so R1 sets the route to an infinity metric and moves it to the DUAL passive state.
If this was true ships in the night routing, R2 would still have an EIGRP route in its local EIGRP routing table (not the main routing table) to reply to R1 with, it is just that this route is in the EIGRP table, but not in the main routing table. However R2 seems to check do I have a route, and is my EIGRP route in the main routing table, and if it is not in the main routing table, then don't respond to any EIGRP queries.
Has anyone seen this behavior before?
Is there any hidden command to make EIGRP and OSPF behave like ships in the night?