Hi,
I'm trying to figure out something here and not sure which one is correct. I am currently doing MPLS VPN with 3 PEs. One of them is the hub and the other two are just stub centres connected to MPLS-VPN backbone. I'm using OSPF as the routing protocol on both spoke sites between PE and CE then the main hub just have a loopback interface being advertised to the VPN for now for testing purposes. However, I'm wonderin why there's a different behaviour on each spoke sites although they have the same configuration (different IPs of course). When I advertised 1.1.1.1/32 network from the hub PE using a network statement under MP-BGP, the result in spoke centres are not the same.
Spoke1-PE1:
show ip ospf database external 1.1.1.1
OSPF Router with ID (10.200.21.33) (Process ID 65020)
Type-5 AS External Link States
LS age: 1307
Options: (No TOS-capability, DC, Downward)
LS Type: AS External Link
Link State ID: 1.1.1.1 (External Network Number )
Advertising Router: 10.200.21.33
LS Seq Number: 80000002
Checksum: 0x5486
Length: 36
Network Mask: /32
Metric Type: 1 (Comparable directly to link state metric)
MTID: 0
Metric: 15
Forward Address: 0.0.0.0
External Route Tag: 3489725933
Spoke2-PE1:
sh ip ospf database external 1.1.1.1
OSPF Router with ID (10.200.21.18) (Process ID 65020)
Type-5 AS External Link States
LS age: 1254
Options: (No TOS-capability, DC)
LS Type: AS External Link
Link State ID: 1.1.1.1 (External Network Number )
Advertising Router: 10.200.21.18
LS Seq Number: 80000002
Checksum: 0x3634
Length: 36
Network Mask: /32
Metric Type: 1 (Comparable directly to link state metric)
MTID: 0
Metric: 15
Forward Address: 0.0.0.0
External Route Tag: 3489725933
As you can see, Spoke1-PE1 has a downbit set on its Type5 LSA. I read on several articles that Type5 LSAs shouldn't have down bit as it uses domain-tag as the loop prevention mechanism but it isn't the case here. Spoke2-PE1 does not have downhard in the output so I suspect this type5 entry in the its database does not have downbit.
The other difference is the IOS, one is using 15.3(1)T and the other one is 15.2(1)T3a. The 15.3 is the one running type5 down bit and the 15.2 does not have.
Is this a bug or something or am I doing something wrong?
Thanks,
John
