From: Chua, Parry (Parry.Chua@xxxxxxxxxx)
Date: Thu Feb 21 2002 - 00:18:42 GMT-3
Hi,
You are not alone, it is not only happen on bgp, it happen on other
routing protocols as well. Ie RIP->OSPF->"other routing protocols" in
the same router will not see the RIP routes that in "other routing
protcol" domains but if it is redistribution at ospf of the same domain
at other router to other routing protocols will work. Some answer from
the group said that it is to prevent loop, or this just another version
of split-horizon ? May be someone
can give a better reasons...
Parry Chua
-----Original Message-----
From: Przemyslaw Karwasiecki [mailto:karwas@ifxcorp.com]
Sent: Thursday, February 21, 2002 10:40 AM
To: ccielab@groupstudy.com
Subject: beating a dead horse (redistribution again)
All,
I am desperatly trying to find some deeper logic
in the way how redistribution works.
There are 2 routers. They exchange prefixes via iBGP.
There is no static injection into BGP on any of them.
Step 1
------
1. router 1 runs rip on its loopback
2. router 1 runs ospf on its ethernet
3. router 1 redistributes rip into ospf
-- verifications that this redistribution works:
a) sh ip ospf database external - shows loopback ip there
b) sh ip route on router 2 - shows loopback ip as O E2
4. router 1 redistributes ospf (internal external 1 external 2)
into BGP.
Problem:
The only route in BGP is IP of the ethernet.
Route redistributed from rip is missing in BGP.
Step 2
------
1. router 2 runs ospf and is adjacent with r1
-- both routes originated on r1 are shown in RIB on roter 2:
a) ethernet route as 0
b) loopback route as O E2 (redistributed as such from rip on r1)
2. redistribution from ospf into bgp is removed from router 1
3. redistribution from ospf into bgp is configured on router 2
(internal external 1 external 2)
4. All ospf routes (O and O E2) are redistributed into BGP.
Why -- ????
Why the same redistribution, from the same OSPF database
performed on 2 routers have different results????
And, please note that in step 1 redistribution was partially
sucsesfull, DESPITE of the fack that no prefixes has been in RIB.
Any ideas?
And, please forgive me if I have overlooked some recent posts
about how redistribution REALLY work. I was following them,
but maybe I overlooked some post addressing my question.
Thanks,
Przemek
PS. Similar problem is in well known lab number 2, but what I presented
here is totally generic.....
This archive was generated by hypermail 2.1.4 : Thu Jun 20 2002 - 13:46:29 GMT-3