03-10-2008 01:18 PM
Has anyone ever experienced an External Type 2 OSPF route to be preferred over an Inter Area route?
Here is our situation. We have two MPLS providers AT&T and Global Crossing. We are peering with AT&T through BGP and with Global Crossing through OSPF. On the AT&T router I am advertising a summary route 10.12.0.0/16 through the aggregate command. I am also sending this route via and area range command on an OSPF ABR to Global Crossing.
When I turn up a new site with both MPLS providers, I am receiving the summary route via BGP and OSPF into two separate routers. I have a core switch that is running OSPF as well. When I redistribute the BGP routes into OSPF, my core switch prefers the E2 route via the AT&T BGP router versus preferring the OSPF Inter Area route via Global Crossing. I would like to prefer all traffic to flow through the Global Crossing routers.
I think this may be an issue with the route being preferred over AT&T because the originating router of this route is closer to the destination (since Global Crossing does redistribute between MBGP and OSPF). See RFC 2328 section 16.4
http://www.faqs.org/rfcs/rfc2328.html
Any thoughts???
Solved! Go to Solution.
03-14-2008 03:32 AM
Hi,
In this case the 10.8/16 external LSA is the only one present, so it is selected. This is expected, standard compliant and nothing to wonder about.
Why are you presenting an output now with no LSA3 for 10.8/16 ?? Where did they go?
Could it be you are troubleshooting changing topologies? Then the picture will be inconsistent and conclusions very difficult, if not misleading.
General advice: First clean up your summary routes. make sure aggregation does not advertise the same aggregate from different locations with different subnets.
Second, make sure the topology and route presented are the same.
Regards, Martin
03-10-2008 08:17 PM
Please provide the following output to help us get a better understanding of this issue.
From the CORE SWITCH
show ip ospf database
show ip ospf interface
From the router with GLBX the same.
thanks,
Joe
03-11-2008 05:29 AM
Hi,
One more question and idea: is the inter-area route inserted into the routing table, if your external LSA is not present? Do you have a VRF configured on the L3 switch? If so the down bit could prevent the usage of the inter-area LSA. There is a command "capability vrf-lite" to ignore the down bit.
Regards, Martin
03-11-2008 07:35 AM
Attached are the outputs that you have requested. I have also attached the output that shows the E2 route being preferred and when I shut the interface where I am receiving the E2 route, the IA route then gets installed into the routing table.
Also to answer Martin's question vrf is not configured on the L3 switch.
Regards
03-11-2008 09:16 AM
Hi,
An even more helpful output maybe
"show ip ospf data summ 10.8.0.0"
"show ip ospf data external 10.8.0.0"
from the core switch
03-11-2008 09:42 AM
Please see the attached output.
Thanks
03-12-2008 04:13 AM
Hi,
What is your topology with respect to areas and MPLS VPN ABRs (PE)? Still some unclear aspects regarding your topology and configuration.
What I conclude from your outputs it looks like this:
Relevant routers by OSPF ID, area and function:
10.224.29.1, area 226, connects to redistribution router 10.12.60.4
10.12.60.1, area 0 and 12, core switch, ABR (aggregating 10.8/16 ?)
10.12.60.4, area 0, ASBR for 10.8/16
192.168.111.2, area 0, is creating an LSA3 for 10.8/16 with DOWN Bit NOT set.
The DOWN Bit (seen as "Upward" in your output, which means not set) will prevent routing loops or suboptimal routing in some MPLS VPN topologies with OSPF and will be set by a PE creating LSA3.
If 192.168.111.2 would be the Global Crossing PE, the down Bit should be set on LSA3 for 10.8/16. It is not so I conclude
1) that you use area range for 10.8/16 there and guess at the core switch as well?
OR 2) the Global Crossing PE does not set the DOWN Bit correctly - not that likely
From RFC 2328 Section 16.2
"If the router has active attachments to multiple areas, only backbone summary-LSAs are examined." Rules out area 12 LSA3 for 10.8/16
"(3) If it is a Type 3 summary-LSA, and the collection of destinations described by the summary-LSA equals one of the router's configured area address ranges (see Section 3.5), and the particular area address range is active, then the summary-LSA should be ignored. "Active" means that there are one or more reachable (by intra-area paths) networks contained in the area range."
So it could be that the area 0 LSA3 are ignored, because of 16.2 (3). This would mean, the router 10.12.60.1 has subnets of 10.8/16 and aggregates them.
Some assumptions, any comments, maybe post configurations to clarify?
Hope this helps!
Regards, Martin
03-12-2008 08:05 AM
Martin,
I think you may be on to something. What would cause the router to not set the DOWN bit for the summary address? BTW, the jp13 OSPF router ID is 10.224.29.3.
jp13vort01#show ip ospf data summ 10.8.0.0
OSPF Router with ID (10.224.29.3) (Process ID 100)
Summary Net Link States (Area 0)
Routing Bit Set on this LSA
LS age: 58
Options: (No TOS-capability, DC, Downward)
LS Type: Summary Links(Network)
Link State ID: 10.8.0.0 (summary Network Number)
Advertising Router: 162.97.235.249
LS Seq Number: 80000002
Checksum: 0x2993
Length: 28
Network Mask: /16
TOS: 0 Metric: 3
Summary Net Link States (Area 226)
LS age: 424
Options: (No TOS-capability, DC, Upward)
LS Type: Summary Links(Network)
Link State ID: 10.8.0.0 (summary Network Number)
Advertising Router: 10.224.29.3
LS Seq Number: 80000001
Checksum: 0xC636
Length: 28
Network Mask: /16
TOS: 0 Metric: 35
03-12-2008 08:46 AM
Hi,
actually I am lost as to where you redistribute from BGP and where the GC router is located. Can you please post the ABR configs related to OSPF and also describe the topology?
Where are you aggregating 10.8/16 (area range 10.8.0.0 ...)?
Thank you!
Regards, Martin
P.S.: seems your attachements are not available right now, further complicating the solution.
03-12-2008 08:59 AM
Attached is a simple visio drawing of our network. I have also reattached the relevant output. The core txt has the output that was requested as well. This is where we are advertising the 10.8.0.0 network with the area range command.
Could the issue be that the DOWN bit is not set on the GBX router and thus would prefer the E2 route since the DOWN bit is set??? Just a thought.
Thanks for your help.
03-12-2008 10:21 AM
Hi,
the down bit will not result in the observed behaviour. There is still something missing. According to your database output for 10.8/16 the router 192.168.111.2 creates a summary LSA for 10.8/16 in area 0. Why would that be?
03-12-2008 10:48 AM
Both core switches in our chi office have the area 8 range command applied to the OSPF process. (192.168.111.2 is Core1 and 192.168.111.7 is Core2).
Core1 and Core2 are both connected to areas 0 and 8. Now we do have some 10.8.x.x interfaces that reside in area 0 which from a design perspective is not good but I don't think this has any effect on this issue.
03-14-2008 03:01 AM
Hi,
In other words, as far as I understand: you have several aggregates for 10.8/16 ???
The area range command will have an effect, as I tried to point out in my previous post:
From RFC 2328 Section 16.2
"If the router has active attachments to multiple areas, only backbone summary-LSAs are examined." Rules out area 12 LSA3 for 10.8/16
"(3) If it is a Type 3 summary-LSA, and the collection of destinations described by the summary-LSA equals one of the router's configured area address ranges (see Section 3.5), and the particular area address range is active, then the summary-LSA should be ignored. "Active" means that there are one or more reachable (by intra-area paths) networks contained in the area range."
This means, afaik, that the 10.8/16 LSA3 is ignored. And thus the 10.8/16 external LSA is choosen to be the routing table entry.
Solution: aggregate into 10.8/17 for area 8 on the ABR.
Hope this helps! Please use the rating system.
Regards, Martin
03-12-2008 11:25 AM
Martin,
Maybe we can simplify this issue with another route. I have an external static route that I am redistributing into ospf as an E1 route.
chi16swcr01#
ip route 72.x.x.0 255.255.252.0 10.8.62.11
router ospf 100
redistribute static metric-type 1 subnets route-map ImportStatic
The preferred route in JPN is via the E2 route over the ATT router. When I shut down the interface to the ATT router on the GBX router in JPN, the E1 route then gets installed into the routing table. When I bring that interface back up, my E1 route is discarded again and removed from the database.
I have attached the output during this entire process.
Regards
03-12-2008 09:00 AM
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide