Just looking for some confirmation for my brain on this one.
Let's say you have a network running OSPF on one side, and EIGRP on the other. There are two points of redistribution (for redundancy, lets call them ASBR1 and ASBR2) performing mutual redistribution (EIGRP into OSPF, and OSPF into EIGRP). This is controlled via distribution lists both ways to ensure OSPF sourced routes are not picked up in EIGRP and redistributed back into OSPF, etc, etc...
Let's say we add a static route in EIGRP and redistribute that. It becomes an external EIGRP route with an admin distance of 170 (along with the OSPF derived routes). We pick this up on the ASBR's, and happily redistribute this route into OSPF with it's default admin distance of 110.
Here's where the problem begins (i think)
Now we have an EIGRP static route floating around EIGRP with an admin distance of 170, and the same route floating around OSPF with an admin distance of 110. Due to OSPF's link state database, each point of redistribution MUST know about the other, hence these ASBR's both see the EIGRP sourced route, AND each others redistributed OSPF route for the same network, except now the OSPF route is preferred due to it's lower admin distance.
Does this make sense? Seems to only be a problem redistributing EIGRP external routes into OSPF, however there doesn't seem to be much info on it, and it seems to me this would be a fairly well documented issue. I've checked all the redistribution dox and not one mentions this could be a potential problem.
Use the distance command on each ASBR raising the admin distance on those EIGRP static networks to something higher than 170 so they prefer their EIGRP routes.
ie. ASBR1 would have the distance command configured to raise the admin distance on all external routes coming from it's neighbor ASBR2.
Sound about right?
The filter_redist_eigrp needs to look something like this:
route-map filter_redist_eigrp deny 10
match tag 55
route-map filter_redist_eigrp permit 20
The rest looks good.
Hope this helps,