So there I was moving right along through the lab and I hit problem 2.2. First part of it - no problem and I got down to the redistribution. I tagged routes coming in through RIP on R6 for the respective BB routers and then on the R1 and R2 redistribution I simply made them E1 routes on the side that I wanted to work. At first I was happy and thought this would work but then I remembered the excellent hot/cold potato example Brian gave and decided not to go with this solution as the internal metric could potentially cause a flaw in this method. I then decided to play with metric instead which ironically was what the SG did. I was about to pat myself on the back and move on to the next problem when I I saw the routing table. Not only did it not work - it revealed a wierdness that I couldn't explain. My metrics at first were too small - which was shown in the database for the routes - meaning the additional metric didn't make enough difference to win - and when I finally made them larger it didn't seem to matter. (even +10k)
Router 5 would choose the first router to advertise the routes - either R1 or R2 regardless of the metrics. If I restarted the process it was a toss up for who would win. (even though the metrics in the table clearly showed 50k for unpreferred routes and 12k for preferrred for both routers when the link for the other router was disabled) I even went as far as to restart each router. Again it was who first came up - won. One time I did this and every other route was to a different router - regardless of metric. One was 50k, next 12k, next 50k - it didn't matter - it was random. Turn off/on cef - no different. I know it is delivering the routes correctly from each because if I turn off one of the routers - R5 shows 50k to unpreferred routes - 12k to preferrred, switch routers and exactly the same - 50k unpreferred, 12k preferred. Turn them both on and it is anyone's guess which route it chooses - it is different every time. Anyone ever seen this? E1, E2, nor metric changes - nothing made a dent. The only way to beat it was summary routes. I am running c1841-adventerprisek9-mz.124-24.T1.bin on R5 and R1/R2 are running flash:c2600-adventerprisek9-mz.124-25d.bin. I can't explain the way it is acting. Anyone else have this problem?