From: Bill Hill (bhill@xxxxxxxxxx)
Date: Mon Apr 01 2002 - 09:13:11 GMT-3
Off the top of my head, it's all I can think of at the moment. I do recall a s
cenario from ECP1 using a tunnel where the peers are connected using IPSEC and
IS-IS and OSPF is running so that when the tunnel would come up, the router wou
ld learn a better route to the tunnel endpoint via the tunnel and the tunnel wo
uld go up and down. Not really a routing loop but would drive you crazy just t
he same.
Basically though, all of this stuff comes down to common sense. Taking away an
y caveats, there are only so many things that can go wrong. Know your protocol
s, know your media types, and know how they interact and you won't have to worr
y about routing loops. Then if you do run into one you can take it apart prett
y quickly.
With regards to the test, I was always told by Bruce Caslow to nail it down....
make sure that when you redistribute, that you filter what you don't want or b
e very specific about what you want to redistribute. As long as you follow tha
t advice, you will rearely run into a problem, and if you do, you will know whe
re it is coming from.
HTH,
-Bill
#8882
-----Original Message-----
From: ying chang [mailto:ying_c@hotmail.com]
Sent: Sunday, March 31, 2002 10:59 PM
To: Bill Hill; ccielab@groupstudy.com
Subject: RE: routing loop in single redistribution point?
Bill,
This make sense when split-horizon is disabled and see it in action in the
lab. Can this happen when split-horizon is enabled? I normally would take
your advice and use the filter to prevent route feedback, but like to know
if this is the only situation that can cause this problem.
Thanks,
Chang
>From: "Bill Hill" <bhill@sgdata.com>
>Reply-To: "Bill Hill" <bhill@sgdata.com>
>To: "ying chang" <ying_c@hotmail.com>, <ccielab@groupstudy.com>
>Subject: RE: routing loop in single redistribution point?
>Date: Sun, 31 Mar 2002 20:52:27 -0500
>
>Sure, try redistributing IGRP into OSPF and vice versa on a medium where
>split horizon is not enabed by default. An OSPF route learned by IGRP will
>be redistributed back into OSPF meaning that the route will now be favored
>via the IGRP router, hence causing a routing loop. Actualy in this case
>you will see the route appear in the table then the router will state that
>the route is possibly invalid, then will disappear completely and reappear.
>
>In any case, proper use of distribute-lists, route-maps, enabling split
>horizon will fix these issues. Key to remember is to be specific with your
>redistribution so that you don't run into this issue. It does help to see
>it in action though so try mocking it up in the lab.
>
>HTH
>-Bill
>#8882
>
>-----Original Message-----
>From: ying chang [mailto:ying_c@hotmail.com]
>Sent: Sunday, March 31, 2002 7:58 PM
>To: ccielab@groupstudy.com
>Subject: routing loop in single redistribution point?
>
>
>Hi all,
>
>In Solie's CCIE Pratical Studies errata for page 623, he took out the
>following statement from the 'Caution' box: "If there is only one
>redistribution point, inherent loop prevention whithin the routing protocol
>will be sufficient enough to prevent loops". I can see multiple
>redistributing points can cause routing loop as demonstrate in Doyle I's
>Route Filtering chapter, but have troubles to see how a routing loop can be
>fromed in a single redistribution point. Can someone give me an example of
>such case?
>
>Thanks,
>Chang
>
>
>
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:57:51 GMT-3