From: ccie2be (ccie2be@nyc.rr.com)
Date: Mon Mar 28 2005 - 17:36:53 GMT-3
Me neither. But, the lab makes avoiding redist impossible.
And, I'm trying to figure out the different ways of doing the same thing so
that if the lab has some constraint (and, it always does), I know all the
ways of accomplishing the same thing.
Unfortunately, at the moment, I don't access to a rack, so I can't test this
myself.
So, here's what you're saying:
ip route 0.0.0.0 0.0.0.0 e0 <--- rip/eigrp will auto redist
ip route 0.0.0.0 0.0.0.0 x.x.x.x <-- rip/eigrp will NOT auto redist
Do I have that right?
Thanks for getting back to me. Tim
-----Original Message-----
From: Pat Chui [mailto:cui666@gmail.com]
Sent: Monday, March 28, 2005 3:15 PM
To: ccie2be
Subject: Re: Potential Routing Loop
Hi, Tim
My understanding is because RIP/EIGRP have to have network statement
classful(eigrp can actually specify network n.n.n.0 0.0.0.255, which
make the statement classless), if you have a static route pointing to
an interface instead of next-hop ip, the router would treat the route
as "connected", therefore,assuming the network statement under
rip/eigrp is a class B(network n.n.0.0) for instance, which includes
the subnet of the static route, router will advertise it just as other
interfaces' falling in the big class B network.
if you only talk about default route, that's different story, in which
case, rip/ospf can use "default-information originate" regardless if
you have a default route in the router or not, for eigrp, I don't
think it has such an option, you will have to manually redistribute
the static default.
Personally, I am not a big fan of redistribution, unless I have to.
HTH,
Pat
On Mon, 28 Mar 2005 08:17:42 -0500, ccie2be <ccie2be@nyc.rr.com> wrote:
> Hi Pat,
>
> Thanks for getting back to me. I appreciate your confirming many of the
> things I thought but wasn't 100% sure about.
>
> If you don't mind, could you talk more about point 4 - when the static
route
> will or won't get automatically redistributed.
>
> Also, if I'm not sure or if I don't know, would there be any problem from
a
> routing point of view, of just manually entering redist static regardless
of
> the IGP?
>
> My thinking is that if I always enter redist static, then it won't matter
if
> the IGP does or doesn't automatically propagate the default route.
>
> Thanks again, Tim
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
Pat
> Chui
> Sent: Monday, March 28, 2005 2:29 AM
> To: ccie2be
> Cc: Group Study
> Subject: Re: Potential Routing Loop
>
> Hi, Tim
> Here is my $.02:
> 1,2. you will have a loop if you have default at each side pointing to
> each other, I don't know how you can prevent it if you leave the
> defaults there. the packet will eventually get dropped due to ip ttl
> expiration.
> 3. packet for networks outside of 172.16.0.0/16 will get dropped by
> the first router it reaches
> 4. depends on what network statement you have in rip/eigrp and how you
> put your static routes in the router(interface or next-hop ip?), the
> static routes may or may NOT get redistributed.
> 5. ip ospf demand-circuit is another method of doing isdn backup,
> normally you choose either ip ospf demand-circuit or floating route,
> of course some other ways also available like dialer-watch, backup
> interface, routing protocol specific methods like snapshot routing,
> triggered updates. but the basic rule is you use one and ONLY one of
> them unless specified otherwise.
> 6. you're right.
>
> On Sun, 27 Mar 2005 14:08:02 -0500, ccie2be <ccie2be@nyc.rr.com> wrote:
> > Hi Guys and Happy Easter,
> >
> > I'm wondering if this config will cause a routing loop? Even if it
> doesn't
> > this scenario raises some interesting issues which I'd like to hear your
> > opinions on.
> >
> > Lots of Left side networks R1 isdn R2 lots of Right side
> > networks
> >
> > Assume there's another path for packets to get back and forth between
the
> > networks on the Left and Right side.
> >
> > The isdn link is there to serve as a backup if the primary path fails.
> > Also, assume when the primary path is available
> >
> > the IGP's are fully converged and all routes are reachable.
> >
> > R1 and R2 are each configured with a default static floating route
> pointing
> > to the other side of the isdn link;
> >
> > For example,
> >
> > ip route 0.0.0.0 0.0.0.0 <ip addr of other end of isdn link> 250
> >
> > Now, there are a few things to think about when configuring this
scenario:
> >
> > 1 Assume the primary path is down and the isdn has come up. If R1 has
a
> > packet for which it doesn't have a path, it will send that packet
> >
> > to R2 based on the default static route. But, suppose R2 when it gets
> that
> > packet also doesn't have a path it. Will R2, based on its own static
> route,
> > send
> >
> > the packet back to R1? Or, will R2 drop the packet?
> >
> > 2 If the default behavior is for R2 to send the packet back to R1, is
> there
> > a way to prevent that? Enable split-horizon, maybe?
> >
> > 3 Assume that all networks on both sides are subnets of 172.16.0.0/16.
> > Would there be any benefit of configuring the default static route like
> > this?
> >
> > ip route 172.16.0.0 255.255.0.0 <ip addr of remote end of isdn link> 250
> >
> > If so, what is the advantage?
> >
> > 4 According to Doyle, if the IGP running on both sides of this network
> are
> > either rip or eigrp, I don't need to redist the static route into the
> > routing process. Both of those IGP's do that automatically, true? But,
> if
> > the IGP's are either OSPF or ISIS, then I have to use the command,
default
> > info originate. Agree?
> >
> > 5 Regarding interesting traffic, my thinking is that there's no point
to
> > having routing protocol traffic be considered interesting or even
allowed
> to
> > cross the isdn link (I would use passive interface bri 0) in this
> scenario.
> > Nor, if the IGP is ospf is there any reason to use ip ospf
demand-circuit.
> > Does everyone agree?
> >
> > 6 In the example above, I used an AD of 250 on the static route. Does
> > everyone agree that the actual value used for the AD doesn't matter as
> long
> > as it's higher than the AD of any IGP's running in the network but lower
> > than 255?
> >
> > I know I've thrown a lot of thoughts and issues in this post, but as you
> all
> > know, Cisco doesn't give partial credit on the lab. One tiny mistake
and
> > you lose all the points for that section.
> >
> > I appreciate any comments, feedback, or insights you might make.
> >
> > TIA, Tim
> >
> > _______________________________________________________________________
> > Subscription information may be found at:
> > http://www.groupstudy.com/list/CCIELab.html
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Sun Apr 03 2005 - 17:56:53 GMT-3