From: Godswill Oletu (oletu@inbox.lv)
Date: Wed Dec 07 2005 - 10:37:50 GMT-3
RE: Controling mutual redistribution loopsDave,
If I understand you correctly......Do you mean something like this?
!
router ospf 1
redistribute rip sub tag 120
!
router rip
redistribute ospf route-map O2R
!
route-map O2R deny 10
match tag 120
match tag 110
!
route-map O2R permit 20
set tag 110
If that is what you meant, I do not think it will work. After awhile you would have undone all the benefits of your redistribution. When both RIP&OSPF converged, all your OSPF&RIP routes will be tagged with either 120 or 110 (except new routes), and if you prevent both tags from getting into a particular routing protocol, then the purpose of the redistribution would have been defeated.
In the above example, the first set of updates will be fine and redistribution will occur but subsequent redistributions of OSPF into RIP will not occur.
OSPF by default have an AD of 110 for all its routes internal, external, etc, within OSPF it has a mechanism to always prefer internal routes to external routes. But the router make its routing decision base on ADs and once there is an external OSPF route with an AD of 110, the route will prefer OSPF to RIP's AD of 120 despite the fact that, that route originated from the RIP domain and RIP is the closest routing protocol to that route.
So, along with the route tagging/filtering, manipulating the AD will come very handy. However, this is not in all cases. If there is a single point of mutual redistribution, I will not care about it. Also, in a single point of mutual redistribution, tagging routes in one direction and dropping them in the other direction is effective, so double tagging and double filtering in both directions will not be necessary.
HTH
Godswill Oletu
----- Original Message -----
From: Schulz, Dave
To: Godswill Oletu ; Venkataramanaiah.R
Cc: Serge N'GBESSO ; nobody@groupstudy.com ; ccielab@groupstudy.com
Sent: Wednesday, December 07, 2005 5:34 AM
Subject: RE: Controling mutual redistribution loops
Godswill -
Great explanation on the AD and the redistribution. Thanks! Let's take this one step further....understanding that we have different AD's with different protocols....And, if we we have a practice to always tag a route when it is redistributed, as well as, denying any pre-tagged routes....would I always prevent routing loops, no matter what administrative distance existed? So, if I have Eigrp somewhere else, and I am redistributing between ospf and Ripv2, for discussion purposes....I could do something like this, right? (and the reverse, elsewhere)....note that I changed and expanded this tagging idea....
route-map O2R deny 10
match tag 110 (drop any tagged routes coming from EIGRP)
match tag 120 (drop any tagged routes coming from RIP)
redistributed
!
route-map E2O permit 20
set tag 90
(all untagged traffic is redistributed and tagged with the admin distance from where it came from)
And, of course, if you would want to allow a previously redistributed route to be passed into another protocol, you could do that by permitting this. Does this work?
No copyrights....I noticed someone else doing a similar thing as well....it just helps to keep things straight and remember what series of tags you are using.
Dave
-----Original Message-----
From: Godswill Oletu
To: Schulz, Dave; Venkataramanaiah.R
Cc: Serge N'GBESSO; nobody@groupstudy.com; ccielab@groupstudy.com
Sent: 12/7/2005 12:18 AM
Subject: Re: Controling mutual redistribution loops
Dave,
I think there are some routing loop issues that tagging will not
eliminate.
e.g. redistributing between RIP&OSPF in two routers say R1&R2.
The loop will occur because, e.g. in R1, all native RIP routes have an
AD of
120, when those routes are redistributed into OSPF, they took on AD of
110
but as E2 or E1 depending on how they were redistributed. Same thing
will
happen when RIP is redistributed into OSPF in R2. The "better"
redistributed
OSPF routes with AD of 110, will eventually replace the original RIP
routes
having AD of 120 where they were sourced from and this will lead to some
interesting loops.
EIGRP have kind of taken care of its self, in that it automatically
assign
an AD of 170 to all external routes, but for the other routing protocols
that leaves the distinction between an external route and an internal
route
within their routing processes, manual AD setting might be needed.
The distance command either under router RIP to reduce all native RIP
routes' AD to a value lesser than OSPF routes or under router OSPF to
increase the AD of all external OSPF routes to a value higher than
native
RIP routes will go a long way in addition to the tag/filtering to
checkmate
routing loops.
Dave, it is kool using the distance as the tag values, I think I will
adopt
that, hope you have not copyrighted it :). My method before now is to
using
1111, 2222, 3333, 6666, etc but using the AD values help to reinforce
one's
prior knowledge.
HTH
Godswill
----- Original Message -----
From: "Schulz, Dave" <DSchulz@dpsciences.com>
To: "Venkataramanaiah.R " <vramanaiah@gmail.com>
Cc: "Serge N'GBESSO " <sergeng@yahoo.fr>; <nobody@groupstudy.com>;
<ccielab@groupstudy.com>
Sent: Tuesday, December 06, 2005 7:41 AM
Subject: RE: Controling mutual redistribution loops
> You are correct, Venkat. Thanks for pointing that out. Here is the
> correction. Good advice on using the debug ip routing to find the
loops.
>
> router ospf 1
> redistribute eigrp 100 route-map E2O subnets
>
> router eigrp 100
> redistribute ospf 1 route-map O2E metric 100000 1000 255 1 1500
> !
> !
> route-map E2O deny 10
> match tag 110 (I use the tag that has the same admin distance #)
> !
> route-map E2O permit 20
> set tag 90
> !
> route-map O2E deny 10
> match tag 90
> !
> route-map O2E permit 20
> set tag 110
> !
>
>
> -----Original Message-----
> From: Venkataramanaiah.R
> To: Schulz, Dave
> Cc: Serge N'GBESSO; nobody@groupstudy.com; ccielab@groupstudy.com
> Sent: 12/6/2005 1:21 AM
> Subject: Re: Controling mutual redistribution loops
>
> May be you meant to set the tags and not match them in your permit 20
> statements...
>
> If you only match them, where would you set them exactly..
>
> A very important tip to spot the routing loops is to enable debug ip
> routing. If you see too many debug messages even after the routing
> protocol is supposed to have converged, then you certainly have some
> routing loop somewhere..
>
> -Venkat
>
>
>
>
> On 12/4/05, Schulz, Dave <DSchulz@dpsciences.com
> <mailto:DSchulz@dpsciences.com> > wrote:
>
> Serge -
>
> The rule of the thumb (I think, at least one of them) on mutual
> distribution
> ...is not redistribute routes back into the area that they originally
> came
> from (causing a loop). IMHO, the best way to do this is through tags,
> though
> you can use specific route....but these are little less
> administration-friendly, since if the routes change, so do your
> access-lists/route-maps. Also, watch out for your metrics (not as big
> in ospf
> as in eigrp). Here is an example that I have....hope this
> helps....(mutual
> redistribution from ospf to eigrp). Also, I am doing this for memory,
> so it
> may look different on your router. Please correct me if I missed
> something.
>
> ------------------------
>
> router ospf 1
> redistribute eigrp 100 route-map E2O subnets
>
> router eigrp 100
> redistribute ospf 1 route-map O2E metric 100000 1000 255 1 1500
> !
> !
> route-map E2O deny 10
> match tag 110 (I use the tag that has the same admin distance #)
> !
> route-map E2O permit 20
> match tag 90
> !
> route-map O2E deny 10
> match tag 90
> !
> route-map O2E permit 20
> match tag 110
> !
> Note: I like to use the tag that has the same admin distance where
the
> route
> originally came from, just to keep it straight....but this your
choice.
>
>
> -----Original Message-----
> From: nobody@groupstudy.com <mailto:nobody@groupstudy.com>
> To: ccielab@groupstudy.com <mailto:ccielab@groupstudy.com>
> Sent: 12/4/2005 9:23 AM
> Subject: Controling mutual redistribution loops
>
> Hi all,
>
> I'm always afraid when it comes to spoting mutual redistribution
loops
> !!
> What are the rules of tumbs for visualising the potential loops ?
> What is the logic behind this ?
> Does packet tagging always block the redistribution loops ?
> what is the implementation logic ?
>
> Thx for your answers !
>
>
>
> Serge R. N'GBESSO
> serge.ngbesso@bnpparibas.com <mailto:serge.ngbesso@bnpparibas.com>
> Security Consultant
> CCDA, CCSP, CCIE:Security,
> Certified Etichal Hacker.
> www.1000club.com/club/japon/ <http://www.1000club.com/club/japon/>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> ---------------------------------
> Appel audio GRATUIT partout dans le monde avec le nouveau Yahoo!
> Messenger
> Tilichargez le ici !
>
>
_______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
> <http://www.groupstudy.com/list/CCIELab.html>
>
>
_______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
> <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 : Mon Jan 09 2006 - 07:07:50 GMT-3