EIGRP Metrics Clarification, Local vs Propagated

[corycandia]

Community,

I think I'm fundamentally misunderstanding how EIGRP works, so I need correction.  The more I learn, the less intell-a-ma-gent I feel...

I have an HQ HUB router with two equal cost/metric paths to a branch.  Primary path is DMVPN over public internet.  Secondary path is GRE running over Azure VPN transport.

I want to control the route to the single branch without effecting other branches on the DMVPN.  That means no changing values on hub DMVPN (delay or bandwidth).

Hub interfaces have no changes to default bandwidth/delay. 

I thought if I changed the bandwidth or delay on the branch side, it would propagate to the hub for that specific branch network, but it does not change anything on the hub.  One the branch table changes, and one path now wins over the other.  Hub shows two equal paths still.  

Question 1:  Does path metric calculation take place on each router independently?  Restated, does each router calculate it's own metrics for a route?

The above scenario shows me I either misconfigured the two routers, or I am not understanding that bandwidth and delay settings don't get propagated neighbors, only reachable networks, and each router calculates it's own metrics?

Question 2:  How should I affect path metrics between a hub and branch without affecting other branches on a multipoint link?

I've been reading about am offset list, is scenario the intended purpose?  Do I need to do an offset list on the hub or the branch?

HUB TOPOLOGY to branch

EIGRP-IPv4 Topology Entry for AS(1)/ID(172.20.19.246) for 172.20.32.0/24
State is Passive, Query origin flag is 1, 2 Successor(s), FD is 26882560
Descriptor Blocks:
172.20.253.2 (Tunnel50), from 172.20.253.2, Send flag is 0x0
Composite metric is (26882560/28160), route is Internal
Vector metric:
Minimum bandwidth is 100 Kbit
Total delay is 50100 microseconds
Reliability is 255/255
Load is 1/255
Minimum MTU is 1250
Hop count is 1
Originating router is 172.20.39.254
172.20.254.3 (Tunnel0), from 172.20.254.3, Send flag is 0x0
Composite metric is (26882560/28160), route is Internal
Vector metric:
Minimum bandwidth is 100 Kbit
Total delay is 50100 microseconds
Reliability is 255/255
Load is 255/255
Minimum MTU is 1400
Hop count is 1
Originating router is 172.20.39.254
ECMP Mode: Advertise out Tunnel0

BRANCH topology to HUB

EIGRP-IPv4 Topology Entry for AS(1)/ID(172.20.39.254) for 172.20.16.0/25
State is Passive, Query origin flag is 1, 1 Successor(s), FD is 1422592
Descriptor Blocks:
172.20.253.1 (Tunnel50), from 172.20.253.1, Send flag is 0x0
Composite metric is (1422592/3072), route is Internal
Vector metric:
Minimum bandwidth is 18000 Kbit
Total delay is 50020 microseconds
Reliability is 255/255
Load is 1/255
Minimum MTU is 1250
Hop count is 2
172.20.254.2 (Tunnel0), from 172.20.254.2, Send flag is 0x0
Composite metric is (3840512/3072), route is Internal
Vector metric:
Minimum bandwidth is 1000 Kbit
Total delay is 50020 microseconds
Reliability is 255/255
Load is 1/255
Minimum MTU is 1400
Hop count is 2

[Richard Burts · Accepted solution]

There are several parts to this question so let me attempt several explanations:

- for each entry that EIGRP will insert into the routing table it uses a reported metric for that prefix which it learned from its neighbor

- EIGRP then uses its local values for bandwidth and delay and calculates its own local metric for the prefix

- EIGRP then proposes the prefix and metric for insertion into the local routing table

So for Q1 yes EIGRP on each router does independently calculate its own metric

For Q2 an offset list can be used to alter the metric of entries. Offset list could be configured on the hub or on the spoke. It seems to me that it would be more simple to configure on the spoke in your case.

[MHM Cisco World]

https://www.cisco.com/c/en/us/support/docs/ip/enhanced-interior-gateway-routing-protocol-eigrp/13673-14.html

check this 

Use an offset-list on R4 to Modify the Composite Metric at R2