RIP Loop prevention note

johnny_jiang

                                                                                                                                                                                                Route poisoning
===============
The process of sending an infinite-metric (hop count 16) route in routing updates when that route fails.
Poison reverse
==============
The act of advertising a poisoned route (metric 16) out an interface, but in reaction to receiving that same poisoned route in an update received on that same interface.
Poison work flow
================
               
               
               
               
               
               
               
               
                ---------(1)Router A(2)-------------(1)Router B(2)--------------
|  a                         b                           c      |
|                            d                                  |
---------(1)Router C(2)-------------(1)Router D(2)--------------
Router A routing table:
a is directly connected, router a/1b is directly connected, router a/2
c [120/1] via router b/1, invalid timer, router a/2d [120/1] via router c/1, invalid timer, router a/1
Router B routing table:
b is directly connected, router b/1
c is directly connected, router b/2
a [120/1] via router a/2, invalid timer, router b/1
d [120/1] via router d/2, invalid timer, router b/2
1. When shuting down router a/1, router b will receive route poisoning update (metric 16) sent from router a. It contains two routes:
a is directly connected, router a/1d [120/1] via router c/1, invalid timer, router a/1
Since router a/1 is the exit interface for these two networks. This is a triggered update not subject to timer, it only includes changed routes.
2. After receiving route poisoning update, router b will send back poison reverse routes immediately. Meanwhile, it will send only another poison route to router d. See below:
a [120/1] via router a/2, invalid timer, router b/1
Since only network a is associated with router b/1 interface which receiving poison route. The other network d is going through another interface router b/2.
3. When router d receives the poison route from router b, it will not influence routing table, since network a is not going through router d/2 interface.
Split Horizon
=============
RIPv1/v2 will advertise all routes with AD=0 even if the route is static route created by 'ip route' command.
E.g. (Config)# ip route 192.168.1.0 255.255.255.0 s1/0
In addition, please note if we use 'ip default-network' command to tag a route, that route will be advertised by RIP without the need of 'network' command. If the 'network' command is used, there will be two routes advertised out, original network and default route (0.0.0.0). On the other hand, if we untag the route afterwards, it will not trigger flash update, that is, it depends on invalid timer to inactivate the default routes advertised by RIP.
E.g. R*   0.0.0.0/0 is possibly down, routing via 192.168.1.1, Serial1/2
Please also note as it is advertised as a default route, it will not be subject to split horizon rule as the default route does not reside in RIP domain. However, when redistributing static route, split horizon will take effect.
Note: 'ip default-network' command is mostly used with RIP protocol to generate default route. You also can redistribute 0.0.0.0 default route. During redistribution, no matter what AD(0 or 1), static route is static.
Command 'clear ip route'
=======================
The command will erase all static routes and routes indicated as 'possibly down' listed in routing table. This command also reset invalid timer. Also, after the clear command, R1 would immediately issue RIP request packets, which cause the neighboring routers to send full routing updates, instead of waiting on their next update time.
02:09:11: RIP: received v2 request from 192.168.20.2 on Serial1/0
02:09:11: RIP: sending update with long TTL
02:09:11: RIP: sending v2 update to 192.168.20.2 via Serial1/0 (192.168.20.4)
02:09:11: RIP: build update entries
02:09:11:       192.168.30.0/24 via 0.0.0.0, metric 1, tag 0
RIP packet format
=================
RIP request

Request packet does not contain IP address header.
RIP response

RIPv1

RIP Database
============
RIP database will get synced with routing table, which means, route will not be shown in rip database if they cannot be listed in the routing table.
E.g.
S    192.168.3.0/24 [1/0] via 192.168.2.2
The route above will take precedence over rip-advertised route. Let's say if the rip-advertised route is already in the rip database, (generally speaking, the static route is appended afterwards) that route will start timeout the invalid timer/flush timer/holddown timer and then get removed from rip database.
R4#sh ip rip database | be 192.168.3.0192.168.3.0/24 is possibly down192.168.3.0/24 is possibly down
00:47:15: RIP: sending v2 flash update to 224.0.0.9 via Serial1/0 (192.168.1.1)00:47:15: RIP: build flash update entries00:47:15:       192.168.3.0/24 via 0.0.0.0, metric 16, tag 000:47:15: RIP: sending v2 flash update to 224.0.0.9 via Serial1/2 (192.168.2.1)00:47:15: RIP: build flash update entries00:47:15:       192.168.3.0/24 via 0.0.0.0, metric 16, tag 000:47:15: RIP: sending v2 flash update to 224.0.0.9 via Serial1/3 (10.1.1.1)00:47:15: RIP: build flash update entries00:47:15:       192.168.3.0/24 via 0.0.0.0, metric 16, tag 0
00:47:38: RIP: sending v2 update to 224.0.0.9 via Serial1/0 (192.168.1.1)
00:47:38: RIP: build update entries
00:47:38:       10.1.1.0/24 via 0.0.0.0, metric 1, tag 0
00:47:38:       192.168.2.0/24 via 0.0.0.0, metric 1, tag 0
00:47:38:       192.168.3.0/24 via 0.0.0.0, metric 16, tag 0
00:47:38: RIP: sending v2 update to 224.0.0.9 via Serial1/2 (192.168.2.1)
00:47:38: RIP: build update entries
00:47:38:       10.1.1.0/24 via 0.0.0.0, metric 1, tag 0
00:47:38:       192.168.1.0/24 via 0.0.0.0, metric 1, tag 0
00:47:38:       192.168.3.0/24 via 0.0.0.0, metric 16, tag 0
00:47:38:       192.168.20.0/24 via 0.0.0.0, metric 2, tag 0
00:47:38:       192.168.30.0/24 via 0.0.0.0, metric 3, tag 0
R4#
00:47:38: RIP: sending v2 update to 224.0.0.9 via Serial1/3 (10.1.1.1)
00:47:38: RIP: build update entries
00:47:38:       192.168.1.0/24 via 0.0.0.0, metric 1, tag 0
00:47:38:       192.168.2.0/24 via 0.0.0.0, metric 1, tag 0
00:47:38:       192.168.3.0/24 via 0.0.0.0, metric 16, tag 0
00:47:38:       192.168.20.0/24 via 0.0.0.0, metric 2, tag 0
00:47:38:       192.168.30.0/24 via 0.0.0.0, metric 3, tag 0
Now the rip route is triggering poison route and send them out, the poison route flow will get started. Finally, it's removed.
The rip database will always be consistent with routing table, not like other routing protocols such as eigrp, ospf etc.
Default route
=============
Except for 'ip default-network' command, we also can use the follow command to generate conditional default-route.
router rip
version 2
network 192.168.1.0
network 192.168.3.0
network 192.168.20.0
default-information originate route-map def-net
!
ip classless
ip route 192.168.3.0 255.255.255.0 Serial1/1
no ip http server
!
access-list 2 permit 2.2.2.2route-map def-net permit 10 match ip address 2 set interface Serial1/2
This means when network 2.2.2.2/32 is present in the routing table (no matter if this network is belonging to RIP domain), default network update will be advertised out thru interface s1/2.
               
               
               
               
               
               
               
               
               
               
               
               
               
               
               
               

本文来自ChinaUnix博客，如果查看原文请点：http://blog.chinaunix.net/u1/42903/showart_1077607.html