Yes, it should take 3 minutes, but unfortunately it only takes a few seconds to be kicked out the routing table and replaced by the alternative route.

This is the test I did:

(config)# router rip

(config-router)# version 2

(config-router)# timers basic 10 30 60 180

So here update interval = 10, invalid 30, holddown 60 and flush 180 and here's what should happen:

1) After 30 secs the route will be marked 'possibly down', because no updates are received.

2) The holddown timer starts at this point and rejects all routes for 60 secs.

So problem is that step 2 doesn't work. The route is replaced by an alternative route a few seconds after the route is declared as 'possibly down'.

No one ever tried this?

Hello,

please send the topology you are using for testing. Note, the holddown timer doesn't mean that an alternative router can't be learn before the timeout, if it exists. Frome the manual:

Interval (in seconds) during which routing information regarding better paths is suppressed. The interval should be at least three times the value of update time. A route enters into a holddown state when an update packet is received that indicates the route is unreachable. The route is marked inaccessible and advertised as unreachable. However, the route continues to forward packets until an update is received with a better metric or until the holddown time expires. When the holddown expires, routes advertised by other sources are accepted and the route is no longer inaccessible. The default is 180 seconds.

Hope this helps, please rate post if it does!

My topology is just a few routers in a circle, nothing special really, but if you want I can post it of course.

This I'm not quite sure about:

"Interval (in seconds) during which routing information regarding better paths is suppressed."

- So they are suppressed only if they come from the same router, alternatives from other routers are accepted?

"However, the route continues to forward packets until an update is received with a better metric or until the holddown time expires."

- I noticed that the route continues to forward packets until the flush timer expires, not the holddown.

Hello, yes the topology is important, and the matter can be more complex of what it seems, there are other factors involved, like the "unreachable" advertisements, it is very right that if a suitable alternative exist is accepted immediately, would you think is reasonable to stay three minutes without a route so a subnet when viable alternative exist?

Hi, thanks for the quick reply.

I agree it's unreasonable to stay three minutes without a route when there's an alternative, so I don't really see the use of the holddown timer...

Topology:

LAN1

|

R1--R2--R3

|............|

R4------R5

R4 goes to LAN1 through R1 and after the link between R1 and R4 fails and it becomes invalid, the route through R5 will be immediately accepted. So no hold down.

Is there a debug command to see when the router enters hold down and an easy way to initiate this state?

Hi,

This topology won't trigger holddown. On fail of a link (interface down), all routes pointing to that link will be immediately removed. Suppose instead:

- R2 has interface to LAN

- all links are ethernet

- R1 is turned off

In this case, there is no link flap, but R4 not seeing updates anymore from R1, will place routes from R1 in holdown first, then flush. Eventually, route via R5-R3-R2 will be installed. To see how how RIP keeps routes, do "show ip rip database".

Result of the above hypothetical scenario, three minutes blackhole in presence of viable alternative. This is why link-state, or neighbor-forming protocols are superior.

Thanks for your suggestion!

I actually just tried the scenario you described. Unfortunately I didn't see any hold down as alternative routes were still immediately accepted...

Am I missing something here?

Well I'm not sure.

When a source of updates ceases, and there are no link status changes, other routers have no mean to invalidate the routes beside timers expiring.