cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
4960
Views
0
Helpful
11
Replies

route stuck in OSPF - how to remove?

roger perkin
Level 2
Level 2

I have a network which runs OSPF everywhere, we have a small subnet that runs EIGRP as they are on an old switch.

I was trying to route this subnet over the ISP link and not our backup link which is currently the way it goes as it is a higher bandwidth.

I first tried by putting a static route to a single host 10.188.3.63 on SW2 - I then since removed this but now, I cannot get to this host from SW2 as for some reason on SW1 it says it is learning the route to 10.188.3.63 from the firewall to the right?

All other hosts on the subnet are being learnt from EIGRP and point the right way.

How can I flush this route from OSPF or is it just preferring it now because OSPF is better?

This is causing a minor issue as from sw2 the route to 10.188.3.63 is over the ISP routers to sw1 then to the firewall to sw4 to sw3 back to sw2 and onto the ISP again?

I think I have created a routing loop but have no idea how to fix this.

Any ideas welcomed!

Roger

1 Accepted Solution

Accepted Solutions

Hi,

    If possible,you should get rid of source of originating those 2 host routes.  You can use static routes to control the route for each device. You can use distribute-list to not install host routes in the routing table as well. In case of OSPF it will still advertise LSA to other upstream devices in the same ospf area.

HTH,

Toshi

View solution in original post

11 Replies 11

Hi,

   Seems you've learnt 2 things on SW1.

1. host route 10.183.3.63/32 advertised via OSPF

2. 10.183.3.0/24 network advertised via EIGRP.

    In this case it's a longest prefix. Eigrp is better than OSPF in regarding AD. Where is 10.183.3.63/32 assigned in OSPF area? 

    If I missed understanding your question,please clarify.

Toshi

Toshi,

Thanks for the reply, 10.183.3.63 should be up with all the other 10.183.3.0 addresses coming from EIGRP. This should be advertised into ospf and then all traffic was going over backup link,

For some reason which I can't explain when I put a static on SW2 to put a test route to 10.188.3.63 it seems to have got this into OSPF and now I can't get it out.

When I try to route to 10.183.3.63 from SW2 it goes over ISP - (it used to go over backup link - all other traffic to 10.188.3.0 goes over the backup link ) then it gets to SW1 and then goes onto the Firewall and back around the loop?

SW1#sh ip route 10.183.3.63

Routing entry for 10.183.0.0/16

  Known via "ospf 1", distance 110, metric 200

  Tag 4445, type extern 1

  Redistributing via eigrp 1

  Advertised by eigrp 1 metric 10000 100 255 1 1500

  Last update from 10.251.251.194 on GigabitEthernet1/0/24, 4d03h ago

  Routing Descriptor Blocks:

  * 10.251.251.194, from 213.12.53.108, 4d03h ago, via GigabitEthernet1/0/24

      Route metric is 200, traffic share count is 1

      Route tag 4445

It seems the host route to 10.188.3.63 is being advertised over the ISP link by the firewall?

Hi,

  As far as I can see from your original post is that 10.183.3.63/32 was learnt via 10.250.251.209 on Vlan997.  Please clarify.

HTH,

Toshi

Yes I know it is being learnt from 10.250.251.209 that is the firewall, my issue is I believe I now have a routing loop.

As working around the network 10.183.3.63/32 is being learnt from right hand firewall who is learning it from SW4 who is learning it from SW3 who is learning it from Firewall who is learning it from SW2 who is learning it from SP router who is learning it from Right hand firewall again.

I need to break this loop and get it to learn the correct path to 10.188.3.63 which is up the EIGRP link

This behaviour is only happening for 2 host routes I put in as statics on SW2 and then since removed?

Thanks

Roger

Hi,

    If possible,you should get rid of source of originating those 2 host routes.  You can use static routes to control the route for each device. You can use distribute-list to not install host routes in the routing table as well. In case of OSPF it will still advertise LSA to other upstream devices in the same ospf area.

HTH,

Toshi

Thanks I was looking into distribute lists. However I am not sure who is originating the host route?

Luckily it is only for a single host at the moment.

Roger,

The host route seems to be originated by the router with the OSPF Router ID 213.12.53.108. Can you identify the router with this OSPF RID and verify its configuration? Obviously it redistributes the route into OSPF.

Best regards,

Peter

Hi,

You can post the configuration here.

Toshi

Sent from Cisco Technical Support iPhone App

Peter,

That router is the SP router on left of the diagam

It does redist BGP into OSPF

router bgp 64800

no synchronization

bgp log-neighbor-changes

network 0.0.0.0 route-map SET_MED_20K

network 10.250.255.224 mask 255.255.255.224

network 172.31.144.252 mask 255.255.255.252

aggregate-address 10.250.0.0 255.255.0.0 summary-only

redistribute ospf 1 route-map SET_MED_20K

route-map SET_MED_20K deny 5

match ip address 1

!

route-map SET_MED_20K permit 10

match ip address prefix-list permit-subnets

set metric 20000

!

route-map SET_MED_20K permit 30

match ip address 2

set metric 20000

ip prefix-list permit-subnets seq 10 permit 10.250.0.0/16 le 32

ip prefix-list permit-subnets seq 20 permit 10.251.0.0/16 le 32

The issue came when we added a prefix list in to the SP routers for the 10.188.3 subnet. So I could get it into OSPF

My question is:

Why when I put a static onto SW2 to route to 10.188.3.63 go via 10.251.251.194 did it get stuck in a loop and OSPF override the EIGRP route coming in from top right?

I think it has something to do with this on SW2

router ospf 1

log-adjacency-changes detail

auto-cost reference-bandwidth 20000

redistribute static subnets

Hello Roger,

My first comment is not directly to the issue you're experiencing but more towards the general understandability of your configuration: on the SP, you are using the same route-map both to set the attributes of the default route injected into BGP and to filter the routes injected from OSPF into BGP. I would personally use two different route maps to avoid possible complications and unwanted side effects.

Second, you are telling me that this router performs BGP-to-OSPF redistribution. From the output, though, I can see only the opposite direction, OSPF-to-BGP.

To your question about OSPF possibly overriding the EIGRP route - I guess that if the route is also redistributed into EIGRP, then obviously the AD of the EIGRP-discovered route (170) is greater than the AD of the OSPF-discovered route (110), hence the OSPF wins. But I am not quite sure where exactly is the host route injected into your routing protocol - is it injected into multiple routing protocols in several places?

Best regards,

Peter

Hi Peter,

There is also these commands on the router

router ospf 1

log-adjacency-changes

redistribute bgp 64800 metric-type 1 subnets route-map BGP_to_OSPF

route-map BGP_to_OSPF deny 20

match ip address prefix-list deny-subnets

!

route-map BGP_to_OSPF permit 30

ip prefix-list deny-subnets seq 10 permit 10.251.0.0/16

The static route was added onto SW1 and there is a redistribute static command in ospf

I now think I need to be looking to tie this with a route map to control which static routes get redistributed.

The config has since been reverted but I am trying to understand why OSPF host route was preferred over the EIGRP subnet and incorectly routed this host around in a circle.

I have the setup now in GNS3 so should have a better understanding later today

Roger

Review Cisco Networking for a $25 gift card