cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
2538
Views
0
Helpful
13
Replies

OSPF Totally NSSA default route selction process

fallegretti
Level 1
Level 1

Hi all,

 

Dealing with a bit of a strange issue, hope you can help me understand it. The image below shows the topology.

NSSA DEF route topology.PNG

Please note, OSPF are is limited to area 313, I have some design constraints. R4 is redistributing a static default route into EIGRP

 

area 313 has been configured as follow:

R1#area 313 nssa default-information-originate no-summary

R2# area 313 nssa default-information-originate metric 100 no-summary

 

So I am getting Type3 and Type7 lsa:

R1#sh ip ospf 313 database summary 0.0.0.0

            OSPF Router with ID (192.168.103.1) (Process ID 313)

                Summary Net Link States (Area 313)

  LS age: 1117

  Options: (No TOS-capability, DC, Upward)

  LS Type: Summary Links(Network)

  Link State ID: 0.0.0.0 (summary Network Number)

  Advertising Router: 192.168.103.1

  LS Seq Number: 80000050

  Checksum: 0xB460

  Length: 28

  Network Mask: /0

        MTID: 0         Metric: 1

 

  LS age: 2012

  Options: (No TOS-capability, DC, Upward)

  LS Type: Summary Links(Network)

  Link State ID: 0.0.0.0 (summary Network Number)

  Advertising Router: 192.168.103.2

  LS Seq Number: 80000050

  Checksum: 0xAE65

  Length: 28

  Network Mask: /0

        MTID: 0         Metric: 1

 

R1#sh ip ospf 313 database nssa-external 0.0.0.0

            OSPF Router with ID (192.168.103.1) (Process ID 313)

                Type-7 AS External Link States (Area 313)

  LS age: 1210

  Options: (No TOS-capability, Type 7/5 translation, DC, Upward)

  LS Type: AS External Link

  Link State ID: 0.0.0.0 (External Network Number )

  Advertising Router: 192.168.103.1

  LS Seq Number: 80000659

  Checksum: 0xEBEC

  Length: 36

  Network Mask: /0

        Metric Type: 2 (Larger than any link state path)

        MTID: 0

        Metric: 1

        Forward Address: 192.168.53.2

        External Route Tag: 0

 

  LS age: 73

  Options: (No TOS-capability, Type 7/5 translation, DC, Upward)

  LS Type: AS External Link

  Link State ID: 0.0.0.0 (External Network Number )

  Advertising Router: 192.168.103.2

  LS Seq Number: 80000051

  Checksum: 0xF789

  Length: 36

  Network Mask: /0

        Metric Type: 2 (Larger than any link state path)

        MTID: 0

        Metric: 100

        Forward Address: 192.168.53.3

        External Route Tag: 0

 

Which default route would you expect to see on R1 and R2?

 

In this case on R1 and R2 I have the EIGRP route

R1#sh ip route

D*EX  0.0.0.0/0

           [170/3072] via 192.168.103.254, 1d20h, GigabitEthernet0/0/0.103

 

This actually what I need but I do not understand why and that's where  the problem is. In order to chose the path out for R3, I removed the no-summary keyword from the command  area 313 nssa default-information-originate no-summary, this implies to inject only Type 7 and manipulate the metrics.

As soon as I did that, R1 and R2 deciced to use OSPF N2 default route pointing at each other creating a nice routing loop.

 

Any idea why if the area is totally NSSA the routers choose the EIGRP route and if it is a plain NSSA they choose the OSPF N2?

 

Thanks

Fulvio

13 Replies 13

Hello,

 

post the full configs of all routers so we can lab this...

Hi Georg,

Can't put the entier configs, too many sensivitve info, I extracted what I believe are the relevant bits; it should give you enough information to put a lab together and hopefully replicate the issue.

BTW - I have been testing on GNS3 and have a different behaviour. One router learns the route from EIGRP, let's say R1 (it depends who learns it first) and passes it on to the other one (R2) as Type7 (note, in this case unless I put an interface in area 0 a Type 3 SLA is not generated) which installs the OSPF route in its routing table. Here comes the "fun" bit, if R4 loses its static default, it learns the route from R2 via Eigrp and of course passes it to R1, which passes it back to R2 as type 7, etc. you get the picture, a default route routing loop.

R3 btw, is not doing anything, just a router in the nssa area.

Many for looking at this,

Hello

 


@fallegretti wrote:

BTW - I have been testing on GNS3 and have a different behaviour. One router learns the route from EIGRP, let's say R1 (it depends who learns it first) and passes it on to the other one (R2) as Type7 (note, in this case unless I put an interface in area 0 a Type 3 SLA is not generated) which installs the OSPF route in its routing table.  - That seems correct behaviour as my understanding is the NSSA no-summary router(s) require to be ABR's and to be connected to the backbone for a default to be installed, This should then create a type 3/7 lsa on R1/R2/R3 even without the inclusion of any  redistributed default static from eigrp and the ospf  nssa default-originate being applied.

Now with the eigrp static being redistributed, nssa default-originate and the no-summary all  being applied then nssa-no summary I think takes precedence over the nssa default-originate, and its only when you remove the nssa summary does the type7 nssa lsa (default-originate) get used.


Here comes the "fun" bit, if R4 loses its static default, it learns the route from R2 via Eigrp and of course passes it to R1, which passes it back to R2 as type 7, etc. you get the picture, a default route routing loop.-  Humm ..This is something I don’t understand or not seeing from your config!

Aren’t both R1-R2 learning the eigrp default from R4 which has a default static route applied? So how does R4 firstly lose its default static and secondary then learn it from R2?

Because when R4 loses its default, That default shouldn’t get redistributed and advertised in eigrp so R1/R2 wont receive it , Also by the looks of it your not performing any mutual redistribution between the routing processes for R4 to receive a default from R1/R2?


Please rate and mark as an accepted solution if you have found any of the information provided useful.
This then could assist others on these forums to find a valuable answer and broadens the community’s global network.

Kind Regards
Paul

Hello

 

Humm ...Do you have both ospf interfaces of R1 and R2 advertised in eigrp passivley?

res

Paul


Please rate and mark as an accepted solution if you have found any of the information provided useful.
This then could assist others on these forums to find a valuable answer and broadens the community’s global network.

Kind Regards
Paul

no I don't. the network command under eigrp is 

network 192.168.103.0

on both routers, only the eigrp interfaces are included in that range.

 

Hello

Try adding them and test again...

Also if you wish for R3 to take just primary path via R1 then you could also just make R1 advertise the nssa default as a metric-type1.

 

 

res

Paul


Please rate and mark as an accepted solution if you have found any of the information provided useful.
This then could assist others on these forums to find a valuable answer and broadens the community’s global network.

Kind Regards
Paul

Hi Paul

What do you mean manipulating the OSPF metric? The metric is how it is seen from the router, in this case R3 which has one interface on the LAN so I can't make its path to R1 better then then its path to R2 and I can create subinterfaces (R3 is not actually a router but an appliance that communicates only OSPF nssa, don't ask).
This way I wanted to use area 313 nssa default-information-originate without the no-summary and use the metric on that. The idea of metric type would not work, that implies removeing the no-summary key word from the command which then causes the routers to prefer the nssa routes and the routing loop I was describing.
Which brings me back to the original question, why are the routers preferrring the EIGRP routes instead of the summary OSPF (type3) routes injected by the ASBRs? it is 110 vs 170 distance after all.
I have replicated the issue in the lab now which makes me partially happy.

Hello

Maybe i am missing something hear, so forgive me...however

 


@fallegretti wrote:

What do you mean manipulating the OSPF metric? The metric is how it is seen from the router, in this case R3 which has one interface on the LAN so I can't make its path to R1 better then then its path to R2  

WHY? -- You can apply a NSSA ospf metric-type 1 to R1  ospf and because R2 is advertsing the same NSSA default with a default metric-type 2 then R3 WILL prefer R1 as it nexthop



This way I wanted to use area 313 nssa default-information-originate without the no-summary and use the metric on that.  - No need to do that

The idea of metric type would not work, that implies removeing the no-summary key word from the command which then causes the routers to prefer the nssa routes and the routing loop I was describing.  - The metric-type sholud work and when you say NSSA routes, Isnt their only one the default?


Which brings me back to the original question, why are the routers preferrring the EIGRP routes instead of the summary OSPF (type3) routes injected by the ASBRs? it is 110 vs 170 distance after all. - How are you advertsing a defailt in eigrp, I assmne by the summary command on the eigrp interface?

 

can you post

 

sh ip vrf xx eigrp route
sh ip vrf xx ospf route
sh ip vrf xx eigrp  neighbors
sh ip vrf xx ospf  neighbor

res

Paul

 

 

 


Please rate and mark as an accepted solution if you have found any of the information provided useful.
This then could assist others on these forums to find a valuable answer and broadens the community’s global network.

Kind Regards
Paul

Hi Paul,

I think I am missing something

"WHY? -- You can apply a NSSA ospf metric-type 1 to R1  ospf and because R2 is advertsing the same NSSA default with a default metric-type 2 then R3 WILL prefer R1 as it nexthop"

 

can you post the actual command you mean?

 

attached is the output of the show commands you have requested

Hello

The command would be for R1  - area 313 nssa default-information-originate metric-type 1 (no-summary) if applicable

 

However I don't see any default route in the routing tables of the ospf  have removed this?


Please rate and mark as an accepted solution if you have found any of the information provided useful.
This then could assist others on these forums to find a valuable answer and broadens the community’s global network.

Kind Regards
Paul

Hi Paul,

 

As per configurations posted, the command is already applied with the no-summary keyword

 

That means that r1 and r2 inject a summary Type 3 lsa, which is preferred to external, so the metric type or indeed a metric I would apply to the Type 7, would not have any effect.

 

If I take the no summary keyword out, only inject a type 7, the default route changes from EIGRP to OSPF N2 with R1 pointing at R2 and R2 pointing at R1 creating a routing loop.

 

The default route is EIGRP, you need to look inot the show ip route eigrp section to see it. Again back to my original post, why are the routers choosing EIGRP (which again is what I want but I just don't understand how) and why if I only inject a type 7 lsa the router selects OSPF over EIGRP?

 

Predrag Jovic
Level 3
Level 3

By default  EIGRP  default route is marked as external D EX with AD 170, while OSPF has AD 110 for all route types which I would guess is root cause of your issue. You can try to change EIGRP external routes AD to 105 to keep preferring EIGRP default route over OSPF default route.

 

router eigrp X

distance eigrp 90 105

 

in named instance it is located under topology base

Thanks Predrag, I am aware of the default distance of routing protocols and the distance command.

 

The issue is that although I have the lsa in my OSPF database the router installs the EIGRP route which has a higher admin distance (as you correctly say 170).

 

There is something I am missing and I have to modify the design, I want to understand exactlly what is going on before moving forward. Hope this clarifies my issue.

Review Cisco Networking products for a $25 gift card