After router relaod - eBGP prefix is vaild but not best - next hope is reachable

[g114112118s]

My topology is very simple using platform 4500x, i have few eBGP peers via SVIs. All the peers were fine and all of SVIs were receiving prefixes. However when we reloaded the router (no config, no IOS change), noticed that after the router reload, one of the BGP peer has received 0 prefix. On investigation found that BGP peer received prefixes (sh ip bgp nei x.x.x.x received-route). But those prefixes are not getting installed in RIB (sh ip bgp nei x.x.x.x route). Noticed that there is no rib failure. Next hope is still reachable. Tried to clear BGP soft sessions. But no luck... it seems as the received prefix is not best, it is not installing in routing table. My understanding is that as next hope is reachable, prefix should be the valid & best but for this prefix it's not a case. just wondering where could be the issue? And what suppose to be my next step.?

[willwetherman]

Hi,

Can you post the 'show ip bgp x.x.x.x' output replacing x.x.x.x with one of the prefixes that you are experiencing issues with?

Thanks,

[g114112118s]

I will post the output later but just compared show ip bgp x.x.x.x for other peers as well and output is just identical.

[willwetherman]

Hi,

Thanks for posting the show ip bgp output.

From the output it looks like network 10.128.0.0/9 is being locally advertised by BGP on both routers using the network command. If this is the case, then route 10.128.0.0/9 is mostly likely to already be in the RIB via either connected or static which will prevent you seeing this in the RIB as BGP.

Can you please explain exactly what you are expecting to see?

[g114112118s]

Basically while performing post checks after the router reload, come to know that one of the bgp neighbor has received prefix count 0.  There is no changes in environment so this is strange to me. Comparing neighbor received prefix logs,only a minors change is from the logs, now prefix received but it is not best.

Yes, network 10.128.0.0/9 is being locally advertised by BGP. Have static route with ad 254. However received prefix network is * 10.128.0.0/25. 

[willwetherman]

Is it the iBGP or eBGP neighbour that is showing a received prefix of 0 following the reload?

Can you post the output from 'show bgp 10.128.0.0/25' from router A and router B as well please?

[g114112118s]

It's both ebgp peer has issue.

From RouterA/B

show bgp 10.128.0.0/25

% Network not in table 

Thanks 

[willwetherman]

Ok I was expecting to see an entry as you mentioned that this prefix was being received.

Can you post the output where you are seeing received prefix 10.128.0.0/25 and highlight where its not being selected as the best entry

[g114112118s]

Hi, willwetherman,

sorry, here is correct output from both the router.

RouterA#sh ip bgp neighbors 10.198.254.2 received -routes
BGP table version is 11, local router ID is 10.0.192.7
Status codes:
s suppressed, d damped, h history, * valid, > best, i - internal,
r RIB -failure, S Stale, m multipath, b backup -path, f RI -Filter,
x best -external, a additional -path, c RIB -compressed,
Origin codes: i - IGP, e - EGP, ? - incomplete
RPKI validation codes: V valid, I invalid, N Not found
Network Next Hop
* 10.198.255.0/25  10.198.254.2
Total number of prefixes 1
Metric LocPrf Weight Path
100 0 65291 ?

RouterA#sh bgp 10.198.255.0/25
BGP routing table entry for 10.198.255.0/25, version 0
BGP Bestpath: compare-routerid
Paths: (1 available, no best path)I
Not advertised to any peer
Refresh Epoch 1
65291, (received -only)
10.198.254.2 from 10.198.254.2 (172.18.254.200)
Origin incomplete, metric 100, localpref 100, valid, external
rx pathid: 0, tx pathid: 0

RouterB#sh ip bgp nei 10.198.254.42 received -routes
BGP table version is 11, local router ID is 10.0.192.8
Status codes: a suppressed, d damped, h history, * valid, > best, i - internal,
r RIB -failure, S Stale, m multipath, b backup -path, f RI -Filter,
x best -external, a additional -path, c RIB -compressed,
Origin codes: i - IGP, e - EGP, ? - incomplete
RPKI validation codes: V valid, I invalid, N Not found
Network Next Hop
*10.198.255.0/25 10.198.254.42
Total number of prefixes 1
Metric LocPrf Weight Path
200 0 65291 ?

RouterB#sh bgp 10.198.255.0/25
BGP routing table entry for 10.198.255.0/25, version 0
BGP Bestpath: compare-routerid
Paths: (1 available, no best path)
Flag: 0x820
Not advertised to any peer
Refresh Epoch 1
65291, (received -only)
10.198.254.42 from 10.198.254.42 (172.18.254.201)
Origin incomplete, metric 200, localpref 100, valid, external
rx pathid: 0, tx pathid: 0

[willwetherman]

Hi,

I can see the following for the 10.198.255.0/25 route entries

65291, (received -only)

Both prefixes are in 'received-only' mode which means that they have been received but not entered into the regular BGP table. This occurs when a route has been filtered using an inbound policy such a route-map but is still viewable because soft-reconfiguration is configured.

Can you check if this prefix is being filtered inbound from your eBGP neighbors?

Please see below thread which explains this behaviour:

https://supportforums.cisco.com/discussion/11746236/difference-between-received-only-and-receivedused-about-bgp

[g114112118s]

here is what i found from the pre reboot log:

Network Next Hop
*>10.198.255.0/25 10.198.254.42

There is no change in route-map so if my understanding is correct, route-map is in effect when prefix is valid and best. Do route-map makes prefix not best ? just wondering what could be the reason that my prefix was valid and best earlier.

>Can you check if this prefix is being filtered inbound from your eBGP neighbors?

i will check on this detail but, as i mentioned there is no configuration change is been involved in this case.

Thank you,

[milan.kulik · Accepted solution]

Hi,

it is possible the prefix had been received in the past originally and the incoming route-map modified after that without BGP session cleared?

In that case the prefix would stay in your BGP table as the best one.

But after the router reload the modified route-map was applied when the prefix was received from the neighbor and filtered the prefix out?

So I'd recommend checking the inbound route-map again.

Best regards,

Milan

[willwetherman]

Yeah this is the only logical explanation that I can think of that explains this behaviour as well.

Please

let us know what you find

[g114112118s]

Hi, milan, willwetherman, yes, you seems correct. it was the issue related to prefix list.

[paul driver]

Hello

one of the BGP peer has received 0 prefix. On investigation found that BGP peer received prefixes (sh ip bgp nei x.x.x.x received-route). But those prefixes are not getting installed in RIB

Do you only have ebgp peerings or do have also ibgp ones?
Do have have synchronization enabled
Are these prefixes being advertised from anywhere else?
Do you have any IGP running?

Have you tried a manual shutdown of the peering ?


res

Paul