Re: Probems with eBGP Neighbour

walu · ‎04-15-2003

We are using 2 BGP Routers to have Redundancy in a Multihomed BGP Network. Each router has an connection to an different ISP. The Routers are connected each other via fast ethernet and iBGP Peers using HSRP. Now we have the Problem that one router overwrites the routes from the eBGP (ISP) Peer with the Routes from the iBGP Peer although they have an worse metric. The neighbours are still up. If we shutdown the Fast Ethernet Interface than the router keep the route entrys. Does anybody has an idea what could be the reason for this behavior.

ives.dekoninck · ‎04-15-2003

This means that the route is more optimal through that other ISP. The decision BGP takes to determin the best path is based on the rules defined below :

Weight > Local Pref. > Locally originated > AS-Path > Origin protocol (IGP>EGP>Incomplete) > MED > EBGP > IBGP > IGP metric > Lowest IP address as BGP ID

So in your case it can be that eg. ISP 1 has a smaller AS-path to the destination then ISP2 has, and since that rule is before the EBGP>IBGP this means that that path will be chosen.

Different options exist to change this :

Eg. change the weight on routes that are received through EBGP. As such every packet arriving on that router will chose the EBGP route.

Hope this makes sence.

Good luck.

-Ives-

melkomy · ‎04-15-2003

Another solution to this problem is to set an inbound filter for IP prefixes on R1 recieved from the IBGP neighbor R2 to force it to learn those prefives via EBGP and use its directly connected ISP1.

walu · ‎04-15-2003

Thanks for all responses. In the meantime we recognized new things.

R2 is usually the active HSRP speaker. To reach a unequal loadsharing we configured as-path-prepend on R1 who is usually standby HSRP speaker. This scenario worked till last weekend. But now maybe something in the ISP network happend. Now It is only working if we disconnect the connection between R1 and R2 (the iBGP Peer). So without an iBGP Peer between them each of the router keep the BGP route entrys from the ISP with an distance of 20 in the routing table. As soon as we connect the router again, R2 overwrites the Routing entrys with a distance of 200 from the iBGP Peer. And also after 60 sec. both neighbours went to state idle.

R2#sh ip bgp summ

BGP router identifier 192.168.35.2, local AS number 21026

BGP table version is 127717, main routing table version 127717

121434 network entries and 274606 paths using 21664914 bytes of memory

30810 BGP path attribute entries using 1602120 bytes of memory

28185 BGP AS-PATH entries using 805624 bytes of memory

0 BGP route-map cache entries using 0 bytes of memory

0 BGP filter-list cache entries using 0 bytes of memory

Dampening enabled. 0 history paths, 0 dampened paths

121408 received paths for inbound soft reconfiguration

BGP activity 2190069/2075561 prefixes, 5038612/4763994 paths, scan interval 15 secs

Neighbor V AS MsgRcvd MsgSent TblVer InQ OutQ Up/Down State/PfxRcd

192.168.13.17 4 21026 238871 179527 22927 0 0 00:00:52 31788

217.9.96.9 4 15671 873631 3129 127536 0 0 00:00:39 121408

Ess-BGP-2#sh ip bgp summ

BGP router identifier 192.168.35.2, local AS number 21026

BGP table version is 1, main routing table version 1

Neighbor V AS MsgRcvd MsgSent TblVer InQ OutQ Up/Down State/PfxRcd

192.168.13.17 4 21026 238902 179748 0 0 0 00:00:05 Idle

217.9.96.9 4 15671 873631 3129 0 0 0 00:00:05 Idle

The only thing we changed is to put an second Loopback (lo1) address on the routers. This Loopback address is now used as BGP Router ID. But we thought that this can not be the problem because we use update source loopback0 for the iBGP Peer .

Here are extracts of the BGP Config of R2.

router bgp 21026

no synchronization

bgp log-neighbor-changes

bgp dampening route-map graded_dampening

network 194.145.88.0

aggregate-address 194.145.88.0 255.255.248.0 summary-only

neighbor ibgp-peer peer-group

neighbor ibgp-peer remote-as 21026

neighbor ibgp-peer password 7 011C0A015A5A5E1C24

neighbor ibgp-peer update-source Loopback0

neighbor EBGP-peer peer-group

neighbor EBGP-peer remote-as 3320

neighbor EBGP-peer prefix-list Net_Police_in_EBGP in

neighbor EBGP-peer prefix-list Net_Police_out_EBGP out

neighbor EBGP-peer route-map as-path-ISP1-in in

neighbor 62.156.133.253 peer-group EBGP-peer

neighbor 62.156.133.253 soft-reconfiguration inbound

neighbor 192.168.13.18 peer-group ibgp-peer

no auto-summary

Any help is welcome to solve this issue.

Thanks

t.baranski · ‎04-16-2003

This sounds like an AS-PATH issue, as mentioned. A "show ip bgp" on R1 should give an indication of why the routes from R2 are being preferred over the routes from the ISP.