Difference between revisions of "Talk: CCNA Explorer 2 Distance Vector Routing Protocols"
m (Created page with "Actually, depending on the IOS release, your results may not be the same as someone else. The flush timer should run in parallel with the invalid timer. If you use an emulator, o...") |
(No difference)
|
Revision as of 22:52, 26 October 2011
Actually, depending on the IOS release, your results may not be the same as someone else. The flush timer should run in parallel with the invalid timer. If you use an emulator, or PT, things don't always work correctly.
The holddown timer should be 6 times the update timer, which is by default 30s. The holddown timer is different, it starts when the router receives an update for a route that includes a higher metric, or when the invalid timer expires (if this happens, the flush timer still should expire 60s after invalid expires).
One thing to note, when the invalid timer expires, the route will go into holddown. So by default the router will not accept any new updates for the route until the flush timer expires and removes the route. This behavior is interesting, in the sense that the holddown follows the invalid, so even if the route came back and a new update was received before the flush timer expired, it would be ignored.
With a newer IOS, if a router notices a connection fails, it will flash an update (or triggered update) to its neighbors with the route poisoned, the neighbors will in turn send out a poison reverse route back to the router, and send a triggered update out their other interfaces.
It will then remove the route from its routing table. The timers do not come into play here (except waiting for next update). It just does this based on the nature of receiving a flash update that has a poisoned route. So, in this case convergence is quicker than the timers.
Now, if you have a router that stops sending updates to its neighbors, then those neighbors will go by the timers. There is a good example of this given in Appendix E of the CCIE Cert Guide 4th Edition.
Like I said earlier, your mileage may vary depending on what version of IOS your using, and what version of RIP your using. When I repeated a lot of the examples from Routing TCP/IP Vol 1 2nd Edition, in Ripv1 and Ripv2 on PT and GNS3, results varied. With a newer IOS on GNS3 and running Ripv2, flash updates work pretty well, and the results are what Wendall shows in the Cert Guide and what Keith presents in the thread.
The best way to see this is to get some routers with a newer IOS and implement it in a lab, then runs some debugs.
HTH DelVonte