From: Gustavo Novais (gustavo.novais@novabase.pt)
Date: Thu Jun 30 2005 - 11:19:59 GMT-3
You are right!
I still have some more "OSPF playing" to do...
-----Original Message-----
From: Scott Morris [mailto:swm@emanon.com]
Sent: quinta-feira, 30 de Junho de 2005 14:42
To: Gustavo Novais; 'CCIE Group Study'; 'Schulz, Dave'; 'Larry Roberts'
Cc: ccielab@groupstudy.com
Subject: RE: OSPF Virtual Link and GRE tunnel.
That is very true. But keep in mind that your tunnel endpoints are not
referring to the ip address INSIDE interface tunnel0. The endpoints are
the iP addresses used in "tunnel source" and "tunnel destination"
commands. So what specific area those are a part of won't change, but
the increased cost (or not) to reach them may. The logic for that part
is that you can't go through yourself to get to yourself. So routing to
the "tunnel destination"
on both sides must be preferred OUTSIDE of the tunnel.
You may tweak metrics, you may tweak AD. That will depend on your lab
scenario. But in the end, you need to reach your tunnel destination the
same way that you did before the tunnel was built.
HTH,
Scott
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
Gustavo Novais
Sent: Thursday, June 30, 2005 8:20 AM
To: CCIE Group Study; Schulz, Dave; Larry Roberts
Cc: ccielab@groupstudy.com
Subject: RE: OSPF Virtual Link and GRE tunnel.
Hi all, your contributions are being great!
Dumb question (or not): Won't OSPF prefer always intra Area Paths over
extra area? Regardless of the metric? If I have both tunnel endpoints
(int tunnel
0) won't it prefer the way through the tunnel (inside area 0),
regardless of the metric?
Thanks
Gustavo
-----Original Message-----
From: CCIE Group Study [mailto:ccie@madisonsolutions.net]
Sent: quinta-feira, 30 de Junho de 2005 12:59
To: Schulz, Dave; Larry Roberts; Gustavo Novais
Cc: ccielab@groupstudy.com
Subject: Re: OSPF Virtual Link and GRE tunnel.
Good morning:
When you build the tunnel the virtual path metric over the tunnel in
OSPF is better than the physical OSPF metric path. So OSPF will use the
virtual path. When it loads the virtual path into the routing table and
removes the physical path; the tunnel goes down because the destination
end is no longer reachable.
Your tunnel will go up-down-up-down-up-down forever. The answer is to
make sure that the physical path has a better OSPF metric than the tunel
OSPF metric. You can do this by changing the bandwidth of the tunnel to
9k; for example.
Now the tunnel will stay up as a direct connection that is unique in the
routing table, but the physical path will remain the best path. Now if
I was a proctor I would have EIGRP and OSPF in the mesh to make the
tunnel and throw ISDN in the mix to mush your brain.
So create a lab with ISDN running EIGRP. OSPF needs a tunnel for Area
0, if the frame link goes down the tunnel must travers the EIGRP ISDN.
The Frame OSPF path must be the best path if the frame is up. 2 Points.
On the CCIE look for metric and distance problems. Now if the lab says
to build the tunnel and the OSPF metric needs to be 1 for your tunnel
you have a new problem. Then you would change the Administrative
Distance on each router for the tunnel path on each router to be 115.
Now the OSPF AD of 110 is better than the OSPF AD of 115, both stay up
and work the way you want them to.
George Morton, Ph. D.
Madison Solutions
----- Original Message -----
From: "Schulz, Dave" <DSchulz@dpsciences.com>
To: "Larry Roberts" <groupstudy@american-hero.com>; "Gustavo Novais"
<gustavo.novais@novabase.pt>
Cc: <ccielab@groupstudy.com>
Sent: Thursday, June 30, 2005 7:19 AM
Subject: RE: OSPF Virtual Link and GRE tunnel.
> If you are addressing the interfaces at both ends of your tunnel and
> putting them in area 0, then you are not violating OSPF. You have to
> remember to do it on both ends of the tunnels (R2 and R7, R2 and R7,
R2
> and R9). In this way, you are truly extending OSPF area 0. Hope this
> helps.
>
>
> Dave
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf
Of
> Larry Roberts
> Sent: Thursday, June 30, 2005 12:28 AM
> To: Gustavo Novais
> Cc: ccielab@groupstudy.com
> Subject: Re: OSPF Virtual Link and GRE tunnel.
>
> Somebody please correct me if Im wrong on my understanding of
> Virtual-links, but I believe the virtual link configuration mentioned
> below is invalid.
>
> I was under the understanding that Virtual links require that one
router
>
> be an ABR to area 0 (R2). This virtual link provides backbone
> connectivity to area 0 for the ABR (R5) but it doesn't *extend* the
> backbone to R5 ( R5 doesn't have an interface in Area 0). When you
> create your second virtual link from R5 to R7, neither of these
routers
> have an interface in Area 0.
>
> If you do a "show ip ospf interface" on R5, you will see that none of
> the interfaces are listed as in area 0. While this configuration seems
> to work fine, It appears to me that it is in violation of OSPF
> configuration guidelines.
>
> Can anyone correct my understanding on this?
>
>
> Gustavo Novais wrote:
>> Apparently I had a dumb config problem on R7... Duhhh... It's working
>> now.
>> Either way, any way how to solve recursive routing situation I
>> presented?
>> Thanks
>>
>> -----Original Message-----
>> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf
> Of
>> Gustavo Novais
>> Sent: quarta-feira, 29 de Junho de 2005 19:33
>> To: ccielab@groupstudy.com
>> Subject: OSPF Virtual Link and GRE tunnel.
>>
>> Hello group
>>
>> I have a topology like this.
>>
>> Area0--(R2)---Area25---(R5)---Area57----(R7)----Area78----(R9)
>>
>> It showed up on a exercise in order to test Virtual-links etc.
>> I did it using vlink between R5 and R2 for area 57 reachability and a
>> vlink between R7 and R5 for area 78 reachability.
>>
>> But I'd like to try the same topology using GRE tunnel between R7 and
>> R2.
>> If I extend Area 0 onto interface tunnel on R7, I get recursive
> routing
>> and the tunnel goes down. (OSPF will prefer intra area routes vs
extra
>> area routes, so the preferred path to the tunnel destination is
> through
>> the tunnel itself, which shuts it down.
>>
>> If I extend area 78 to R2 Tunnel, apparently all is well, but the
>> problem is that R5 starts complaining that Received invalid packet:
>> mismatch area ID, from backbone area must be virtual-link but not
> found
>> from 150.50.57.7, Ethernet0/0, even now that area 78 is connected
>> directly to area 0, and area 57 still has its vlink on R5 to Area 0.
>>
>> Any ideas why is the router R5 showing this behaviour? Any
suggestions
>> how to correct it?
>>
>> Thanks
>>
>>
>
This archive was generated by hypermail 2.1.4 : Wed Jul 06 2005 - 14:43:46 GMT-3