Hi all,

I am hoping to get confirmation of a problem I ran into recently.  Before I
post the configs I will say that I was very confident I knew what the
problem was the moment I saw the configs, but certain people need more
convincing...

Here's an idea of what the PE configs look like:

router ospf 100 vrf VPN_BROKEN
 router-id 1.1.1.1
 log-adjacency-changes
 capability vrf-lite
 area 0 sham-link 10.10.10.1 10.10.10.2 cost 10
 area 0 sham-link 10.10.10.1 10.10.10.2 cost 10
 redistribute bgp 1 subnets
 network 192.168.1.1 0.0.0.0 area 0
 network 192.168.1.3 0.0.0.0 area 0

router bgp 1
!
address-family ipv4 vrf B3933994
 redistribute connected route-map SHAMLINK
 redistribute ospf 100 vrf VPN_BROKEN match internal external 1 external 2
 no synchronization
 exit-address-family

And here's an idea of the CE config:

router ospf 99 vrf MPLS-DATA
 router-id 1.1.1.2
 log-adjacency-changes
 capability vrf-lite
 network 1.1.1.2 0.0.0.0 area 0
 network 192.168.1.1 0.0.0.0 area 0

And now for the problem description...

After a reload of the PE router whose config is presented above, neither of
the listed sham links came back up.  The person who found & corrected the
problem did so by creating distribute lists on the PE's to prevent the sham
link routes getting into the routing table from the OSPF database.

Now.. What I am pretty sure happened after reload:
1. The redistribution between OSPF-BGP allowed the sham-link routes into
OSPF on the CE's where they then transited the CE network.
2. The presence of capability vrf-lite on the PE's allowed the redistributed
sham link LSA's to get back into the remote PE's with the routing bit set -
allowing injection into the routing tables.
3. The sham links, now with routes via OSPF, failed to establish.

When the Sham links were brought up initially, they would have established
based on the BGP routes (it's even possible "capability vrf-lite" was only
added after the Sham links were up).  Because they function as demand
circuits, the routing could change from BGP-based to OSPF-based and not
actually cause the sham links to fail.  It was only when they were taken
down and then attempted to re-establish via OSPF-learned routes that they
fully broke.

If I'm right, I'm hoping someone has an even clearer explanation of why I'm
correct.  If I'm wrong, then perhaps someone can enlighten me!

Thanks,

Eric
_______________________________________________
For more information regarding industry leading CCIE Lab training, please visit 
www.ipexpert.com

Reply via email to