According to my BSCI study materials, the EIGRP topology table can store up to six redundant routes.
What if, bizarrely, there are more than six routes to a destination and all the metrics are equal?
I'm guessing it would just install the first six it learns about; but that makes me wonder about the architecture of the EIGRP software: How long does it take for a received update to be examined and installed in the topo table? Can another update arrive and compete with it before the topology table is officially updated?
>> Can another update arrive and compete with it before the topology table is officially updated?
Very good note this.
Actually, in the DUAL algorithm description it is clearly stated that each input event (EIGRP hello, EIGRP update and so on) is processed one at time.
They are totally processed one each.
So the condition you have suggested should not happen regardless of the time it takes to complete the process actions on the update.
So you can say that the updates coming from the 6 oldest neighbors are installed.
Actually sh ip eigrp neighbors contains a column H that tells the order of neighborship (H=0 oldest)
newer IOS images can have increased to 8 this limit reflecting the increased paths in CEF.
Hope to help