[afnog] route in rib, not in fib - why? SOLVED

[Frank Habicht]

Hi all,

Yes. Philip is spot on.
All your assumptions (including about my stupidity) are right.

But I have to admit thanks to earlier private mails from Nishal and 
Randy I got the right idea already. And it wasn't the first time....

Mark: they don't do BGP with upstream. They're $enduser with a /27 (and 
no IGP).
So more scary thing regarding stealing transit is that default route on 
their IXP-peering router....
acl..... what acl????

So: Thanks to all.
Prob resolved before I saw Mark & Philip's emails. (i'm late replying)
Take those emails to heart .

PS: i haven't seen their "show ip bgp" yet...

Thanks!
Frank


On 6/25/2008 4:14 PM, Philip Smith wrote:
> Hi Frank,
> 
> Ah yes, this one... And the next hop for these 60 routes is? ;-)
> 
> Make sure the iBGP on their IX peering router has "neighbor x.x.x.x 
> next-hop-self" configured. In fact, just do this on all iBGP speaking 
> routers, its industry best practice. (That's a note to all list readers. 
> ;-))
> 
> BGP will only put prefixes into the routing table (ie RIB) if they are 
> valid paths and there is no better path known via another routing 
> protocol. And the common cause of it not doing this is that the next-hop 
> is unreachable (ie not in the RIB). Next-hop for the IXP prefixes will 
> be the IX LAN, and the IX LAN is probably not in their IGP (and it 
> shouldn't be!), so you need to fix the next-hop so that it is the 
> address of their IX router.
> 
> If it is not the above, then as Mark said, a "show ip bgp" will tell all.
> 
> philip
> -- 
> 
> Frank Habicht said the following on 24/6/08 19:44:
>> Hi all,
>>
>> new peer at our IX has this prob:
>> at border router to uplink, they get the ~60 routes from iBGP from 
>> their other router (at IXP place) correctly.
>>
>> "show ip bgp" shows them (sorry for old syntax - dont have their "show 
>> version " (yet)).
>>
>> but (!):
>>  "show ip route" does only show the 4 connected and static routes
>> :-(
>> What could be wrong?
>>
>> and that (show ip route) is also how the packets behave.
>>
>> I hope the subject summarises correctly.
>>
>> Sorry to be very vendor C specific ;-)
>> let me know if you need more info.
>> (but the box isnt under my control)
>>
>> Frank
>>
>> snippets from config:
>>
>> version 12.4
>>
>> interface FastEthernet0/1
>>  description ...............
>>  bandwidth 512
>>  ip address 192.168.x.161 255.255.255.252 secondary
>>  ip address n.n.n.n m.m.m.m
>>  duplex auto
>>  speed auto
>> !
>> router bgp <priv-ASN>
>>  no synchronization
>>  no bgp log-neighbor-changes
>>  network s.o.m.e mask n.e.t.224
>>  neighbor 192.168.x.162 remote-as <priv-ASN>
>>  neighbor 192.168.x.162 soft-reconfiguration inbound
>>  no auto-summary
>> !
>> ip classless
>> ip route ................
>>
>> _______________________________________________
>> afnog mailing list
>> http://afnog.org/mailman/listinfo/afnog
>>
>