EIGRP Convergence when link fail its not happening

[Lucas Rene]

Hi everyone! Im doing some labs about EIGRP and convergence when a link fails. This is my topology and in router 4 f0/0 interface i have the 8.8.8.8 IP address

. 

In this topology the metric calculation is based on delay only, so all links by default would have the same metric value´s (delay 200 its in every interface) 

I modify the R2 f1/0 delay to 199 in order to be a FS to 8.8.8.8 from R3´s perspective, and here is the result of the ip eigrp topology from R3

P 8.8.8.0/24, 1 successors, FD is 102400
via 172.16.1.1 (102400/51200), FastEthernet2/0
via 10.1.1.1 (153344/102144), FastEthernet0/1

so based on this, when a link fails EIGRP will not QUERY its neighbors to look for an alternative route to 8.8.8.8 because the router already knows about how to reach that route and the exact calculation, so i will think it will fail over the f0/1link almost instantly, but this doesnt happen... I tried an extended ping and the result show me that the route its not taking in efect until the HOLD TIME from F2/0 realizes his neighbor is down, and when that happens i instantly get to the backup link through f0/1 and the ping goes on...

Why is this happening??? ( I figure this out by modifying the hold time value between both R1 and R3 f2/0 link to 180

Sending 500, 100-byte ICMP Echos to 8.8.8.8, timeout is 2 seconds:
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!............................................................
..........!
*Mar 1 00:19:57.595: %DUAL-5-NBRCHANGE: IP-EIGRP(0) 100: Neighbor 172.16.1.1 (FastEthernet2/0) is down: holding time expired!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!
Success rate is 86 percent (430/500), round-trip min/avg/max = 16/44/148 ms

[Giuseppe Larosa]

Hello Lucas,

what are using for your lab?

is an emulated lab in GNS3?

the message:

>> Mar 1 00:19:57.595: %DUAL-5-NBRCHANGE: IP-EIGRP(0) 100: Neighbor 172.16.1.1 (FastEthernet2/0) is down: holding time 

tells us that R3:fa2/0 didn't go down when you have shut down the interface on R1. But it is still up/up in a case like this R3 can only wait for EIGRP hold time to expire.

If you are using real devices are you using a LAN switch in the middle ?

In real world if directly connected just after shutting down R1:fas2/0 R3:fas2/0 would go down.

Hope to help

Giuseppe

[Lucas Rene]

Yes this is a emulated lab in GNS3, i still didnt try this at a real enviroment. I supposed that if i shutdown one of both links it will automatically back up to the other link. Are you saying this is just a "emulation" problem? And if it is, if this happens in real life it will backup immediately?  Thanks for your attention

[Giuseppe Larosa]

Hello Lucas,

yes in real world if directly connected after shutting one side the other side fails quickly to up/down an the feasible successor should be immediately installed.

If a LAN switch is in the middle of course this does not happen.

Hope to help

Giuseppe

[Lucas Rene]

Great, and what should i do in my emulated topology with gns3 in order to see this happening? I should put a switch in the middle? or there is some other way to test this in a emulated lab? You been very helpful, thanks.

[Giuseppe Larosa]

Hello Lucas,

no you shouldn't insert a switch in the middle or the observed behaviuor would be correct.

I know it is a dumb question, are you sure you have shutted down the correct interface on R1?

I apologize for the question but sometimes it can be the case

Hope to help

Giuseppe

[Lucas Rene]

What are you saying with the "observed behaviour would be correct" ? does that say that inmmediately when the link fails it will backup to the f0/1 route? because thats what i want it to do, wich it would be the correct behaviour, right?

And yes, when im in R1 i shutdown the f2/0 interface

[Lucas Rene]

NOTE: I placed a switch in the middle of every connection and this is still happening...

[Lucas Rene]

Here is a video i just upload it https://drive.google.com/open?id=1wICExzivyoh_DugmcTxZToY_X2mSCA3-

[Giuseppe Larosa]

Hello Lucas,

as far as I can see the network diagram on the left is the one you posted in the first post of this thread.

The R3 fas2/0 should fail just after shut down on R1 fas2/0.

I think something is not working in emulation, but actually I am not an expert of GNS3 so I cannot help.

What if you change the link between R1 and R3 in a serial link? You are using only delay in EIGRP you can configure the correct delay on serial interface as well.

This is just to see if something changes in behaviour.

Hope to help

Giuseppe

[Giuseppe Larosa]

Hello Lucas,

with a switch in a middle the R3 fas2/0 will stay up also in real world.

I apologize for misunderstanding.

You should remove the switch in the middle.

Now I give a look at the video you have posted.

Hope to help

Giuseppe

[paul driver]

Hello

Basically to summerize what Giuseppe has stated -  if the router doesn't see a physical connection loss ( which seems to be the case here) then the running IGP routing process (rip/eigrp/ospf) will use its timers for route validation prior to any convergence