OSPFv3 cost of the loopback is not added to the metric.

[Salah_IT_Nerd]

Hello everyone, I'm having some trouble understanding how OSPFv6 calculates costs in IPv6.

Attached are screenshots of some show commands on both routers.

My question is: why OSPF is not adding the cost of the interface loopback to the metric.

Thank you!

[M02@rt37]

Hello @Salah_IT_Nerd,

Could you please more elaborate with a rapid topology as example.

To be sure, if you modify the cost of an OSPF interface, that only affects the metric of routes 'going out' that specific interface.

[Salah_IT_Nerd]

Hi M02@rt37

I've added a screenshot of the topology which simply consist of 2 routers connected via Gigabit Ethernet link.

I did not change the cost on any interface, I've modified the reference bandwidth to 10000 Mbps.

[Blue_Bird]

It should add...that's how ospf/ospfv3 works...

it should add...2001:DB8:1::1/128[110/11]; It should display as cost 11.

I think...it is the limitation of the gns3....we can't expect 100%....it might be a bug gns3/ios image...

Sure...with real devices will be added...

Best regards
******* If This Helps, Please Rate *******

[Salah_IT_Nerd]

Hi Gopinah_Pigili

I've tried the configuration on Cisco IOSv 15.9(3)M6 and Cisco 3725 124-15.T14, both gave the same result.

I have some Cisco 2911 and Cisco ISR4331 routers at home, I will try that and see.

[Blue_Bird]

I tried....in gns3... ospf (ipv4) it is adding loopback cost...with ospfv3 (ipv6) it is not adding...

Probably when you do it on your real router...yes...will get it...

Thanks

[Salah_IT_Nerd]

Hi Gopinath_Pigili

There's no difference between GNS3 and real equipment in this scenario, one of the members proposed a solution that worked which consist of changing the network type on loopback interface into point-to-point.

[Blue_Bird]

ok...Nice to here that....

Thanks

[MHM Cisco World]

Redistrubte connect subet 

This will add metric of LO to total metric.

[Salah_IT_Nerd]

Hi MHM Cisco World,

Why do I have to redistribute the connected subnets since I've already enabled IPv6 OSPF on the loopback interface using the command: IPv6 ospf 1 area 0?

[M02@rt37]

@Salah_IT_Nerd,

You configure "ospf" command under interfaces to have an OSPF neighborship and start exchange LSA on a "shared-link".

Here, you want to have this loopback redistributed in your OSPF in order to be announce into this AREA... Then @MHM Cisco World  propose you to redistribute this loopback.

[Salah_IT_Nerd]

I've used the command: (ipv6 ospf 1 area 0) on both GigabitEthernet 0/0 and Loopback 0, if I didn't use the command on the loopback subcommand the show ipv6 route would not list any route to the loopback.

[Martin L · Accepted solution]

it is not IOS fault; it is IPv6 OPSF behavior; it is Host route vs Stub network (default) , see RFC 5340

1st, change /128 to something else (or add new IPv6 address) then make network point -to- point to see difference in metric.  I believe those 2 changes must be under loopback to see new metric;

Even with ipv6 ospf network point-to-point, metric is not taken into consideration if your prefix length is /128

Regards, ML
**Please Rate All Helpful Responses **

[Salah_IT_Nerd]

Hi Martin L,

Thank you so much for the clarification.

I have issued this command on L0 on both routers:

Router1#conf t
Enter configuration commands, one per line. End with CNTL/Z.
Router1(config)#interface loopback 0
Router1(config-if)#ipv6 ospf network point-to-point
Router1(config-if)#end
Router1#

Then, I checked the ipv6 route ospf on Router 2 and noticed that the metric has changed from 10 to 11.

This is without changing the prefix length, it's still /128, however, the show ipv6 route ospf on Router 1 indicates that it's /64.

Now, I'm still being a nerd, why does OSPFv3 behave that way which is not the case with IPv4?

[Martin L]

I don't have "why" in my notes; only RFC listed as reference;  I will check my sources later and will let u know if there is clear answer;

Note that Cisco implementation of OSFP does not have to follow RFC step by step; OSPF metric in RFC is not specified as bandwidth as Cisco does; RFC says just "metric" and "cost" which can be anything else.  So, this "issue" of loopbacks could be just Cisco "way" of doing things

Regards, ML
**Please Rate All Helpful Responses **