03-31-2021 11:49 AM - edited 03-31-2021 12:00 PM
Had an issue earlier where we had a local site that could not reach any resources at site XYZ. Just to give a concise details, This is a dmvpn hub and spoke. As shown below, the local site is advertising its local IP segment to hub as checked hub is receiving the prefix but other devices like R1 and R2 are not receiving the said routes.
Since this dmvpn connection is being used as a backup we just often use this connection(if primary fails). So in this case, Primary fails but creates an outage.
In order to resolve the issue, I removed the set community 65004 at router R4 after being able to see that connection work.
I have reviewed the setup and Existing configuration but unable to see that we are filtering anything related to the community value.
From R3 configuration we are match prefix with community value of 65004 and added another set of community value and do a as-prepending.
Note: from R1 to R2 we set it a route-reflector client as well R1 to R3 peering.
Sample configuration shared above:
R1:no related config to community value. (nexus device)
R1# sh run | i community
send-community
send-community
send-community
send-community
send-community
R2: no related config to community value. (4441 device)
R2#sh run | i community
neighbor 10.118.129.1 send-community
neighbor 10.118.129.2 send-community
.....
Current Status:
R1:
Advertised path-id 1
Path type: internal, path is valid, is best path, in rib
AS-Path: 191, path sourced external to AS
xxxx (metric 130816) from xxxxxx (R3)
Origin IGP, MED 3072, localpref 100, weight 0
Community: 22:1102 22:1300 65000:65003
R2:
191
10.118.129.10 (metric 131072) from 10.118.129.1 (10.118.129.1)
Origin IGP, metric 3072, localpref 100, valid, internal, best
Community: 22:1102 22:1300 65000:65003
Originator: 10.118.129.10, Cluster list: 10.118.129.1
rx pathid: 0, tx pathid: 0x0
what could go wrong?
What would be the issue preventing R1 to install the more specific route?
why removing the community value on spoke router r4 resolve the issue?
Thanks
03-31-2021 12:24 PM
Hello,
post the full configs (sh run) of all 4 routers...
03-31-2021 01:29 PM - edited 03-31-2021 03:19 PM
Hi,
I see you are running NX-OS on R1. BGP update is rejected because the AS path contains the local ASN (22). NX-OS behaves differently than IOS in that sense, whereas IOS does not have that validation and will accept the update even if it contains the local ASN. In other words, IOS only does perform that check when it receives the update from an eBGP peer. NX-OS does, even if the update is received from an iBGP peer.
To fix the issue you could do AS path prepend of the remote AS (191) instead of local AS (22).
Regards,
04-15-2021 09:15 PM
Thanks Harold,
Had a chance to run a debug and I'm able to see that it is due to the AS number.
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide