Solved: Regarding the redistribution

joneaton · ‎04-07-2016

I am under the impression that ospf path cost can be used to influence OSPF path selection, but am experiencing some difficulties.

I have a 4500-VSS and a 6500-VSS which have 2x10Gbps Ethernet links between them running as a port channel. In addition to this port channel they are also connected with a single 1Gbps link.

SVI's are used at both ends rather than configs on physical or portchannel interfaces.

The SVI (vlan 204) that is being trunked across the Port-Channel is only being used when the SVI (602) over the single 1Gbps link is shut down. I've tried statically configuring the OSPF cost on the interfaces with vlan 204 being the lowers (prefered) and vlan 602 being higher. However when both ports are up all routes are shown as going via the single 1Gbps link (vlan 602).

Sanitised config on the 6500-VSS:

!
interface Vlan204
 ip address 10.71.207.89 255.255.255.252
 ip ospf message-digest-key 1 md5 7 DELETED
 ip ospf cost 10
!
interface Vlan602
 ip address 192.168.86.254 255.255.255.248
 ip ospf message-digest-key 1 md5 7 DELETED
 ip ospf cost 100
!
router ospf 100
 auto-cost reference-bandwidth 40000
 nsf
 area 0 authentication message-digest
 area 89 authentication message-digest
 area 89 nssa no-summary
 redistribute connected subnets tag 100
 redistribute static subnets tag 100
 passive-interface default
 no passive-interface Vlan204
 no passive-interface Vlan602
 network 10.71.15.12 0.0.0.3 area 0
 network 10.71.15.16 0.0.0.3 area 0 
 network 10.71.207.88 0.0.0.3 area 89
 network 192.168.86.248 0.0.0.7 area 89
!

Sanitised config on the 4500-VSS

!
interface Vlan204
 ip address 10.71.207.90 255.255.255.252
 ip ospf message-digest-key 1 md5 7 DELETED
 ip ospf cost 10
!
interface Vlan602
 ip address 192.168.86.253 255.255.255.248
 ip ospf message-digest-key 1 md5 7 DELETED
 ip ospf cost 100
!
router ospf 100
 auto-cost reference-bandwidth 40000
 area 89 authentication message-digest
 area 89 nssa
 redistribute connected metric-type 1 subnets
 redistribute static metric-type 1 subnets
 passive-interface default
 no passive-interface Vlan204
 no passive-interface Vlan602
 network 10.71.207.88 0.0.0.3 area 89
 network 192.168.86.248 0.0.0.7 area 89
!

So on the core 6500-VSS I have the SVI 602 shutdown and checking route to 10.13.11.0 (a subnet directly connected on the 4500-VSS)


6500-VSS#sh ip ro 10.13.11.0
Routing entry for 10.13.11.0/24
 Known via "ospf 100", distance 110, metric 130, type NSSA extern 1
 Last update from 10.71.207.90 on Vlan204, 00:24:03 ago
 Routing Descriptor Blocks:
 * 10.71.207.90, from 192.168.89.254, 00:24:03 ago, via Vlan204
 Route metric is 130, traffic share count is 1
6500-VSS#
6500-VSS#
6500-VSS#conf t
Enter configuration commands, one per line. End with CNTL/Z.
6500-VSS(config)#int vl 602
6500-VSS(config-if)#no sh
6500-VSS(config-if)#^Z
6500-VSS#sh ip ro 10.13.11.0
Routing entry for 10.13.11.0/24
 Known via "ospf 100", distance 110, metric 130, type NSSA extern 1
 Last update from 10.71.207.90 on Vlan204, 00:24:31 ago
 Routing Descriptor Blocks:
 * 10.71.207.90, from 192.168.89.254, 00:24:31 ago, via Vlan204
 Route metric is 130, traffic share count is 1
6500-VSS#
6500-VSS#sh ip ro 10.13.11.0
Routing entry for 10.13.11.0/24
 Known via "ospf 100", distance 110, metric 120, type NSSA extern 1
 Last update from 192.168.86.253 on Vlan602, 00:00:02 ago
 Routing Descriptor Blocks:
 * 192.168.86.253, from 192.168.89.254, 00:00:02 ago, via Vlan602
 Route metric is 120, traffic share count is 1
6500-VSS#
6500-VSS#conf t
Enter configuration commands, one per line. End with CNTL/Z.
6500-VSS(config)#int vl 602
6500-VSS(config-if)#sh
6500-VSS(config-if)#^Z
6500-VSS#sh ip ro 10.13.11.0
Routing entry for 10.13.11.0/24
 Known via "ospf 100", distance 110, metric 120, type NSSA extern 1
 Last update from 192.168.86.253 on Vlan602, 00:00:15 ago
 Routing Descriptor Blocks:
 * 192.168.86.253, from 192.168.89.254, 00:00:15 ago, via Vlan602
 Route metric is 120, traffic share count is 1
6500-VSS#
6500-VSS#sh ip ro 10.13.11.0
Routing entry for 10.13.11.0/24
 Known via "ospf 100", distance 110, metric 130, type NSSA extern 1
 Last update from 10.71.207.90 on Vlan204, 00:00:01 ago
 Routing Descriptor Blocks:
 * 10.71.207.90, from 192.168.89.254, 00:00:01 ago, via Vlan204
 Route metric is 130, traffic share count is 1
6500-VSS#

So no matter what I do, the metric across the 204 SVI is always 10 higher than the 602 SVI.

What am I missing here ?

Rolf Fischer · ‎04-10-2016

Regarding the redistribution of the connected routes, tbh I am not sure why it is configured that way. I guess I could move them into the appropriate area. Especially if that helps the situation.

That will help for sure. Perhaps you're doing summarization or filtering at the area border (this is possible because the routes are NSSA-externals and (NSS)ABRs have to translate them and are therefore the originators of the Type-5 LSAs), but you can do this with internal routes as well.

Discussion: Disadvantages of redistributing connected networks into OSPF

Now to your original question:

the (seed-) metric of the Type-7 LSA is 20 (default value)
the Forwarding Address (FA) is 192.168.86.253 (4500's SVI VLAN 604; Cost: 100)
the route to the FA is learned via VLAN 204 (local Cost: 10)
the route to the FA has a metric of 110 (100 + 10)
total cost to external Network 10.13.11.0/24: 20 + 110 = 130

So what you see is expected behavior. In a NSSA an ASBR must set the FA to a non-zero IP-address and the path to this FA is one component in the routing calculation for external Networks. The second component is the LSA metric.

You could achieve some different results when the FA would be taken from a Loopback-interface but again, I'd recommend to propagate the connected networks as OSPF-internals.

HTH
Rolf

View solution in original post

joneaton · ‎04-07-2016

Investigating further:-

The subnet 10.13.11.0/24 is directly connected to Vlan 11 on the 4500-VSS

4500-VSS_089254#sh ip ro 10.13.11.0
Routing entry for 10.13.11.0/24
 Known via "connected", distance 0, metric 0 (connected, via interface)
 Redistributing via ospf 100
 Advertised by ospf 100 metric-type 1 subnets
 Routing Descriptor Blocks:
 * directly connected, via Vlan11
 Route metric is 0, traffic share count is 1
4500-VSS#

I am not doing anything special with the metrics from this 4500-VSS outbound, so looking on the 6500-VSS.

6500-VSS#sh ip ro 10.13.11.0
Routing entry for 10.13.11.0/24
 Known via "ospf 100", distance 110, metric 130, type NSSA extern 1
 Last update from 10.71.207.90 on Vlan204, 01:16:46 ago
 Routing Descriptor Blocks:
 * 10.71.207.90, from 192.168.89.254, 01:16:46 ago, via Vlan204
 Route metric is 130, traffic share count is 1
6500-VSS#

The metric is now 130, even though the output below indicates interface cost is configured as 10.

6500-VSS#sh ip ospf int vl 204
Vlan204 is up, line protocol is up
 Internet Address 10.71.207.89/30, Area 89, Attached via Network Statement
 Process ID 100, Router ID 10.71.207.31, Network Type BROADCAST, Cost: 10
 Topology-MTID Cost Disabled Shutdown Topology Name
 0 10 no no Base
 Transmit Delay is 1 sec, State DR, Priority 1
 Designated Router (ID) 10.71.207.31, Interface address 10.71.207.89
 Backup Designated router (ID) 192.168.89.254, Interface address 10.71.207.90
 Timer intervals configured, Hello 10, Dead 40, Wait 40, Retransmit 5
 oob-resync timeout 40
 Hello due in 00:00:00
 Supports Link-local Signaling (LLS)
 Cisco NSF helper support enabled
 IETF NSF helper support enabled
 Index 1/20, flood queue length 0
 Next 0x0(0)/0x0(0)
 Last flood scan length is 0, maximum is 40
 Last flood scan time is 0 msec, maximum is 4 msec
 Neighbor Count is 1, Adjacent neighbor count is 1
 Adjacent with neighbor 192.168.89.254 (Backup Designated Router)
 Suppress hello for 0 neighbor(s)
 Message digest authentication enabled
 Youngest key id is 1
6500-VSS#

So what I don't understand is where is the additional cost of 120 coming from ??

Mark Malone · ‎04-07-2016

just a quick thought is its your auto cost under the process causing it ? does the vss have 40gb links that you need this set that way

joneaton · ‎04-08-2016

The auto cost was one of my early thoughts, we do have

router ospf 100
 auto-cost reference-bandwidth 40000

under the router ospf process.

We do not have any 40Gbps links in the VSS, just 10Gbps.

Rolf Fischer · ‎04-07-2016

Hi,

it is a NSSA external type-1 route, that means the cost of the path to the Forwarding Address is one component of the total cost. The other component is the LSA Metric (should be 20 by default).

Could you share the output of 'show ip ospf database nssa-external 10.13.11.0' ?

There should be a line like this: Forward Address: <IP-Address>

Please share also the output of 'show ip route <IP-Address>'.

If this shows a connected route (e.g. "directly connected via Vlan X"), we'll also need a 'show ip ospf interface vlan X'.

HTH
Rolf

P.S.: It is generally not recommended to redistribute connected networks in OSPF. This is even more the case in NSSAs (as you can see). Is there a particular reason why you do that?

joneaton · ‎04-08-2016

Hi,

Outputs below:-

6500-VSS#
show ip ospf database nssa-external 10.13.11.0
OSPF Router with ID (10.71.207.31) (Process ID 100)
Type-7 AS External Link States (Area 89)
LS age: 1001
 Options: (No TOS-capability, Type 7/5 translation, DC, Upward)
 LS Type: AS External Link
 Link State ID: 10.13.11.0 (External Network Number )
 Advertising Router: 192.168.89.254
 LS Seq Number: 8000122B
 Checksum: 0xAB8D
 Length: 36
 Network Mask: /24
 Metric Type: 1 (Comparable directly to link state metric)
 MTID: 0
 Metric: 20
 Forward Address: 192.168.86.253
 External Route Tag: 0

6500-VSS#
6500-VSS#sh ip ro 192.168.86.253
Routing entry for 192.168.86.248/29
 Known via "ospf 100", distance 110, metric 110, type intra area
 Last update from 10.71.207.90 on Vlan204, 05:32:18 ago
 Routing Descriptor Blocks:
 * 10.71.207.90, from 192.168.89.254, 05:32:18 ago, via Vlan204
 Route metric is 110, traffic share count is 1
6500-VSS#

Regarding the redistribution of the connected routes, tbh I am not sure why it is configured that way. I guess I could move them into the appropriate area. Especially if that helps the situation.

Rolf Fischer · ‎04-10-2016

Regarding the redistribution of the connected routes, tbh I am not sure why it is configured that way. I guess I could move them into the appropriate area. Especially if that helps the situation.

That will help for sure. Perhaps you're doing summarization or filtering at the area border (this is possible because the routes are NSSA-externals and (NSS)ABRs have to translate them and are therefore the originators of the Type-5 LSAs), but you can do this with internal routes as well.

Discussion: Disadvantages of redistributing connected networks into OSPF

Now to your original question:

the (seed-) metric of the Type-7 LSA is 20 (default value)
the Forwarding Address (FA) is 192.168.86.253 (4500's SVI VLAN 604; Cost: 100)
the route to the FA is learned via VLAN 204 (local Cost: 10)
the route to the FA has a metric of 110 (100 + 10)
total cost to external Network 10.13.11.0/24: 20 + 110 = 130

So what you see is expected behavior. In a NSSA an ASBR must set the FA to a non-zero IP-address and the path to this FA is one component in the routing calculation for external Networks. The second component is the LSA metric.

You could achieve some different results when the FA would be taken from a Loopback-interface but again, I'd recommend to propagate the connected networks as OSPF-internals.

HTH
Rolf

joneaton · ‎04-18-2016

Thanks Rolf, that explains it perfectly.

Sorry for the delay I was away on vacation last week.

Regards

Jon

OSPF parallel path traffic engineering.