From: Dan C (cdan2154@gmail.com)
Date: Sat Feb 02 2008 - 02:32:03 ARST
Very informative, just to add to the topic :
* Would start with a complete diagram taking note of multiple entry points
* Check for any existing redistribute connected making sure that I permit
the rest of the interfaces
* Enable debug ip routing on few edge routers and having them cascade on my
screen... this way as soon as I put a command in I should see some activity
at on the other devices
* Use sh ip route to check which routing protocol is advertising which route
* Use not only ping but traceroute to identify any problems
* Use tags via route-maps making easy to block them if any loops
Cheers,
Dan
On Feb 1, 2008 7:50 AM, vagish madhavan <vagish2003@indiatimes.com> wrote:
> Hi George/Navid,
>
> Below example has only one redistribution point (R2) R2 redistributes RIP
> routes into OSPF. These routes will be external and will only be generated
> by ASBR (R2 in this scenario).
>
> If we dig into OSPF databases of R3 and R4, we can find only prefixes
> advertised by R2. So R2 never gets the routes back which it had
> redistributed. I'm afraid below example doesn't cause any loop for the above
> reason.
>
> Sometimes I spend lot of time on redistribution. I've some ground rules
> which help me to some extent.
>
> 1. If there is more than one redistribution point, we are bound to get
> caught in the loop.
>
> 2. Two-way redistribute at only one point and check the routes. At this
> point I'd always have loop prevention configuration too.
>
> 3. It's always better to identify the loop in the diagram and have
> prevention techniques in place before we do redistribution at multiple
> points. The reason being it's difficult to deal with the loop after it
> starts.
>
> Will keep you posted if I find anything else.
>
> Thanks,
> Vageesh Madhavan.
>
>
>
>
>
>
>
> ----- Original Message -----
> From: George Goglidze <goglidze@gmail.com>
> To: Navid Daghighi <smart4D@free.fr>
> Cc: Cisco certification <ccielab@groupstudy.com>
> Sent: Wed, 30 Jan 2008 02:51:20 +0530 (IST)
> Subject: Re: strategy for redistribution
>
> Hi there Navid,
>
> you can't really say which is the best method for redistribution,
> every topology is different.
>
> but anyway first thing you have to look for is
> when you redistribute rip->eigrp on one router, not to redistribute same
> route eigrp->rip on another router.
> which you can fix by setting/checking tag's with route-map's.
>
> now another thing is finding out which routes need to belong to which
> protocol,
> and when redistributing making sure these routes still belong to correct
> protocol.
>
> for example,
> R1-----R2 (RIP)
> R2-----R3 (OSPF)
> R2-----R4 (OSPF)
> R3-----R4 (OSPF)
>
> all networks from R1 are announced by rip to R2.
> R2 redistributes this route to OSPF and announces this route to R3 and R4.
> R3 announces it to R4, R4 announces it back to R2 by OSPF.
> and as OSPF has better AD then RIP, better route will be now through R4
> which is wrong because the route is through R1.
>
>
> I hope my example was understandable.
>
> in this situations I would say it's acceptable to do following on R2 ->
> "distance ospf external 180"
> but what this out, as this may bring you other problems as well.
> because OSPF might have some routes that are coming from other place and
> are
> external, but must be OSPF routes.
> so you would need to filter that routes and set lower distance on them.
>
> I hope this helps clear it up a bit.
>
> Regards,
>
>
> On Jan 28, 2008 10:00 PM, Navid Daghighi <smart4D@free.fr> wrote:
>
> > Hi,
> >
> > I am trying to find out a good strategy for what is usually the last
> > question of the IGP section : REDISTRIBUTION.
> >
> > Some people may be able to find immediatly and exactly and exhaustively
> > where we are going to have troubles with loops etc. etc.. and where
> > redistribution should be done with tools preventing these loops
> > (route-map, distance etc..) : but this is not my case.
> >
> > So I came up with this "strategy" :
> >
> > 1/ do redistribution on all required routers and/ or switches
> >
> > 2/ write tcl scripts and / or macros
> >
> > 3/ run tcl / macros and see where pings are not functioning and "repair"
> >
> > 4/ run them on all routers annd switches until it all works without any
> > "repairing"
> >
> > Well, I agree there must be a "shorter" & more efficient way.
> >
> > If you have a constructive and rigurous way, it would be appreciated !
> >
> > thanks,
> > Navid
> >
> > _______________________________________________________________________
> > 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
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Sat Mar 01 2008 - 16:54:47 ARST