OSPF Router ID - No sign anywhere

[asidd]

Hi,

Was wondering if someone has seen anything like this before. The OSPF Router ID has not been selected according to the usual process. There is no id hard-coded via config, no loopbacks configured, no interface IP that matches the router ID showing up on the process.

RTR#show ip ospf
Routing Process "ospf 10" with ID 172.18.250.34

RTR#show running-config | include 172.18.250.34

RTR#show ip interface brief
Interface IP-Address OK? Method Status Protocol
GigabitEthernet0/0/0 172.18.5.1 YES NVRAM up up
GigabitEthernet0/0/1 172.18.250.42 YES manual up up
GigabitEthernet0/0/2 172.18.250.38 YES NVRAM up up
GigabitEthernet0 unassigned YES NVRAM administratively down down
Vlan1 unassigned YES unset administratively down down

How did it pick up the RID and is there a fix i need to do?

[Giuseppe Larosa · Accepted solution]

Hello @asidd ,

the OSPF RID is chosen at the moment that the router ospf process has started.

Have you changed the configuration of one of the interfaces ? Or Gi0 was used in the past ?

In any case if you use clear ip ospf process or clear ip ospf 10 and you answer yes the OSPF process is restarted and the new RID

should be 172.18.250.42 (the highest IP address of a normal interface because no loopback interface is configured)

Note: all OSPF neighbors are turned down a new Router LSA is created and then OSPF adjacencies are restored.

Hope to help

Giuseppe

[Reza Sharifi]

Hi,

In a maintenance window, can you clear the OSPF process and test again?

clear IP ospf process

HTH

[Martin L]

OSPF process will not start without ID, so at some point in time you had to have 172.18.250.34 as IP or as manually configured ID.  IP was overwritten or deleted recently but OSPF has not been restarted. 

[asidd]

Thank you all for quick feedback. I think the ospf process was never cleared. I have a maintenance coming up where we will be rebooting the router anyway. I can probably pick a new router-id or hard code it to 172.18.250.34. Which one is more advisable?

@Giuseppe Larosa Thanks for the note and the heads-up. OSPF is operated in Area 0 only with 8 routers in about 6 sites all over different point to point networks but not really configured as a particular ospf network type and just using pretty basic default configuration (something i would love to fix but at a later stage). I wouldnt imagine a risk to  other sites only the one site going offline for its neighbors? 

RTR#show ip ospf neighbor

Neighbor ID Pri State Dead Time Address Interface
172.16.117.201 1 FULL/BDR 00:00:32 172.18.250.41 GigabitEthernet0/0/1
172.16.117.200 1 FULL/BDR 00:00:38 172.18.250.37 GigabitEthernet0/0/2

router ospf 10
redistribute connected subnets
redistribute static subnets
network 172.18.250.38 0.0.0.0 area 0
network 172.18.250.42 0.0.0.0 area 0

[Reza Sharifi]

Hi,

Thank you all for quick feedback. I think the ospf process was never cleared. I have a maintenance coming up where we will be rebooting the router anyway. I can probably pick a new router-id or hard code it to 172.18.250.34. Which one is more advisable?

I would hard code an IP as the router-id. It is usually a good idea to use a loopback interface as it never goes down unless the whole switch/router reboots. If you assign any other interface IP, there is always the possibility of it going down for one reason or another.

HTH

[Giuseppe Larosa]

Hello @asidd ,

>> I can probably pick a new router-id or hard code it to 172.18.250.34. Which one is more advisable?

My personal recommendation is to start to use loopback interfaces in each router.

The current router-id 172.18.250.34 is too much similar to the p2p link addressing and it can create confusion.

If you have 8 routers you could deploy an addressing plan for loopback like the following

172.18.255.x     with x= 1 to 8

then in router ospf process you can advertise the loopback in area 0

router ospf 10

network 172.18.255.x 0.0.0.0 area 0

As noted by @Reza Sharifi a loopback interface is always on until the router is not isolated and this is a great advantage:

you can use the loopback interface to telnet/SSH to the device for SNMP polling and it can be the source of syslog messages and so on.(a specific command is needed for this and another one to be used as source for sending SNMP traps)

This provides consistency over time.

Final Note:

You have both redistribute connected and redistribute static.

This kind of configuration can create issues in some cases.

The LSA type 5 for external route has a field called Forwarding Address that provides a sort of next-hop for the prefix.

This field can contain :

0.0.0.0 meaning the router itself in this case there is no issue

or it can be the actual IP next-hop of the original static route for the prefix.

in this second case if the actual IP next-hop is not learned in an internal route the LSA type 5 will be ignored and not installed by other routers in OSPF domain.

So I would recommend to use

router ospf 10

network 172.18.0.0. 0.0.0.255.255 area 0

+

passive-inferface default

no passive-interface <interface to RX>

no passive-interface <interface to RY>

and to remove

no redist connected subnets

Hope to help

Giuseppe

[asidd]

@Giuseppe Larosa very helpful tips. I was thinking about removing the 'redistribute connected' config and after seeing your comment definitely going to do it this Sunday. Thank you.

As for:The LSA type 5 for external route has a field called Forwarding Address that provides a sort of next-hop for the prefix.

This looks similar to BGP next-hop self. In what cases will the FA wont be the local router. I meant what triggers an FA different to 0.0.0.0.

[asidd]

Successful planned event with loopback interface as RID, passive interface configuration and removing redistribute connected.

Thank you all @Giuseppe Larosa @Reza Sharifi @Martin L 

[Giuseppe Larosa]

Hello @asidd ,

thanks for your feedback

Best Regards

Giuseppe