BGP not selecting higher Local preference

[hfakoor222]

So i'm doing a lab

bgp path selection is as thus

• Attributes
  • Weight
  • Local Preference
  • Originate
  • AS path length
  • Origin code
  • MED
  • eBGP path over iBGP path
  • Shortest IGP path to BGP next hop
  • Oldest Path
  • Router ID
  • Neighbor IP address

My question is about R4  and   R5  the 2 routers in the middle of this topology next to Switch1

Now R4, R5, R6, R7 are in a peer group. To get to router R1's loopback of 2.0.0.0 I've modified the LocalPref to 200 on R4, and so all routers in this BGP AS prefer R4,  except for R5 which is taking the eBGP route to R3, to R1.

All the other PA's seem equal to me.

Why is R5 taking R3 eBGP out over going through R4?  

The only difference I see in the above screenshot is R4 has a metric0, over R5's null metric. And so I remember reading in the OCG about how metrics are applied across different vendor brands with regards to BGP. I'm not sure if that's what is the cause here.

Any in sight would be helpful.

[M02@rt37 · Accepted solution]

Hello @hfakoor222,

Since the LP is the same, the tie breaker used is external versus internal BGP. Prefer eBGP over iBGP.

[hfakoor222]

I distinctly remember reading this somewhere it might've been the ENCOR ocg.  Okay thanks.

[MHM Cisco World · Accepted solution]

From metric I see you ise eigrp for ibgp neighbor.

Even if this metric high this step will not consider in best path selection' because before it there is step to prefer ebgp than ibgp.

That why you see external is best.

[hfakoor222]

Thank you

[David Ruess]

Hello,

I believe it’s hitting the MED requirement. In CISCO world a missing MED is better than 0 I believe. You can change this behavior by issuing the

med-missing-as-worst

command. What this does is any MED that is blank is treated as the worst path for that metric. Then it should choose R4.

-David

[hfakoor222]

Thanks