RIP explained: RIPv2 configuration, timers, loop prevention and troubleshooting

How RIP works

The Routing Information Protocol (RIP) is a distance-vector protocol. Each router sends its whole routing table to its neighbours every 30 seconds, and each neighbour adds one hop. The metric is simply the hop count; 15 is the maximum and 16 means unreachable, which limits RIP networks to small sizes. RIP uses UDP port 520 and its administrative distance on Cisco devices is 120.

RIP ignores bandwidth: a path over two 10 Gbit/s links loses to a single slow link. That, plus slow convergence, is why OSPF or EIGRP replaced it in almost every production network. You still meet it in labs, on small or very old equipment, and sometimes as a simple protocol between a customer and a provider.

Timers and loop prevention

RIP's Cisco timers are: update 30 seconds, invalid 180 seconds (no update heard, mark the route unreachable), hold-down 180 seconds (ignore worse information about a route that just failed) and flush 240 seconds (remove the route). You can see them with show ip protocols and change them with timers basic, but every router must use the same values.

Because routers only know what neighbours tell them, a failed network could bounce between two routers with the metric rising by one each time, the problem called counting to infinity. The rules in Figure 1 prevent it. Split horizon never advertises a route back out of the interface it came from; route poisoning advertises a failed route with metric 16 straight away; poison reverse sends a learned route back to its source as unreachable; hold-down stops a router believing a worse path too soon after a failure; and triggered updates send changes immediately rather than waiting for the next 30-second cycle.

Configuring RIPv2

router rip
 version 2
 no auto-summary
 network 10.0.0.0
 network 192.168.50.0
 passive-interface GigabitEthernet0/2
 default-information originate
 maximum-paths 4

• network takes a classful network and enables RIP on every interface inside it.
• no auto-summary stops RIP summarising at classful boundaries, which would break discontiguous subnets.
• A passive interface in RIP stops sending updates but still receives them, unlike OSPF and EIGRP. To ignore incoming updates too, filter them with a distribute-list.
• default-information originate advertises a default route to neighbours.
• maximum-paths sets how many equal-hop paths are used at once (4 by default).

RIPv2 can summarise on an interface with ip summary-address rip 10.10.0.0 255.255.0.0, but on Cisco IOS it cannot summarise beyond the classful boundary. An offset-list adds hops to selected routes, for example offset-list 10 in 3 GigabitEthernet0/1, which makes routes matched by ACL 10 arriving on that interface look three hops further away, so a backup link is used only when needed.

Troubleshooting and reliability

1. show ip protocols: version sent and received, networks, passive interfaces, timers and neighbours heard from.
2. show ip route rip: routes learned, their hop count and how many seconds since the last update.
3. debug ip rip in a lab or maintenance window: shows every update sent and received, and version or authentication mismatches.
4. Common faults: version mismatch (v1 and v2), a missing network statement, auto-summary on discontiguous networks, a passive interface, or an ACL blocking UDP 520.

To make a RIP-advertised internet exit reliable, track the ISP with an IP SLA probe, tie the static default route to the tracker, and use redistribute static in RIP: if the provider stops answering, the static route is withdrawn and RIP stops advertising it. On a DMVPN hub running RIP, split horizon must be disabled on the tunnel interface so spokes learn each other's networks.

Real-world example: a small charity still had RIPv2 between its office router and an old Layer 3 switch. After a new VLAN was added on the switch, the office router never learned it. show ip protocols on the switch showed the new subnet, 172.20.5.0/24, was in a classful network with no network 172.20.0.0 statement. Adding the statement fixed it within one update cycle. The charity later replaced RIP with OSPF when it added a second site.