EIGRP routing table refreshed

[Leonlei]

i have a FR hub-and-spoke network running EIGRP. Point-to-multipoint interface configured at hub router and point-to-point interface configured on spoke router. Recently, the eigrp routing table on spoke routers refreshed multiple times everday, FR circuit and eigrp neighbor always up. The router platform is 2503 and ios is 12.0(26). I never seen this problem before.

[dathaide]

hi

On the spoke end are you seeing a loss of neighbor? Have you check with your FR service provider maybe the he may be having problems on that frame circuit

thanks

[ruwhite]

The first place to start is with a show ip eigrp events, and see what event occured that caused the routes to be changed. Note the timer in the routing table doesn't necessarily mean a lot. It's changed each time eigrp touches the route, whether it's to replace it, or just because it received an update that didn't change anything, so there's a lot of reasons that timer could change.

:-)

Russ.W

[Leonlei]

Thanks for the helpful direction. Do you think any network congestion on the pvc will cause the routing update is blocked?

[ruwhite]

EIGRP doesn't use periodic routing updates, so congestion shouldn't be an issue. As long as the neighbor is up, we consider all the routes reachable through that neighbor as valid.

:-)

Russ.W

[cdfowlie]

You may want to also take a look at "show ip eigrp neighbors" when you are seeing problems. PVC congestion can be an issue. If your Queue count (Q Cnt) is any number greater than zero you may have some PVC type problems. You can look for drops, particularly in your broadcast queue.

Also if you are running any traffic shaping/queuing it may have an effect. Do all the neighbors go down at the same time, or one at a time throughout the day?

-Colin

[Leonlei]

I have to correct myself first, EIGRP neighbor also will go down when the eigrp routing table is refreshed.

I checked the Q Cnt, it is 0. Currently three FR sites are having same problem, but not happen at exact same time. The latest refreshed was 2 days ago.

No any traffic shaping or queuing config.

[ruwhite]

Do you have log neighbor changes on? If so, why does it say the neighbors are dropping? There are four possible reasons:

-- Dead timer timeout, which means hellos and other eigrp packets are being dropped across the link. This probably indicates some sort of link failure.

-- The link is failing.

-- Retransmission timeout exceeded, which means the link is periodically failing to sucessfully transmit reliable packets, so EIGRP is retransmitting them enough times to kill the adjacency. This is normally a dirty link, or perhaps queues, or the like.

-- Stuck in Actives. These are generally caused by a wide query range combined with dirty or low speed links, possibly a router in trouble (fragmented memory, high cpu utilization, or lots of queries and not enough memory/cpu to handle the load).

If the neighbors are failing on a regular basis, you need to figure out which of these reasons is the culprit, and work from there.

:-)

Russ.W