From: Albert Lu (albert_lu@optushome.com.au)
Date: Fri Feb 21 2003 - 05:45:43 GMT-3
Not too sure what you are looking for in this thread, but here's my 2cents.
1. Yes, that's a little rule that is defined by the RFC for which bgp and
ospf router-id must match for bgp sync, otherwise use 'no synch'. Since
router-id is put into an lsa by the router that generated the lsa, this is a
problem. As the router could be several hops away, and the bgp neighbor is 1
hop away. Other protocols don't suffer from this, as any advertisement uses
the router-id of the advertising router. Hardcoding router-id to a loopback
should be standard practise.
2. OSPF fundamentals. Read your RFCs
3. Not sure why you want to specify a p2p link as a multipoint link. Can you
elaborate?
Regards,
Albert
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
ray_gan74@hotmail.com
Sent: Friday, February 21, 2003 7:10 AM
To: ccielab@groupstudy.com
Subject: CCIE Gotchas: Watch out!
Wanted to start a thread on some gotchas to look out for when configuring
protocols/redistribution. PLEASE contribute. Ill go first.
1. BGP-->OSPF redistribution
What = Routes not advertised to ibgp peers.
Why? = Router ID's of ospf and bgp don't match.
Solution: Use BGP confederations which does not suffer from this issue.
Use another IGP instead of OSPF. Turn off BGP synchronization on each BGP
router within AS. Hardcode the OSPF and BGP router-id's to so
that they match.
2. You can not configure a virtual link across a stub area
3. If you set a p2p link as ip ospf point-to-multipoint then what you will
see
is /32's being advertised for both your end points of the link breaking the
split horizon rule. I am not sure why the /32 sets sent.
.
This archive was generated by hypermail 2.1.4 : Sat Mar 01 2003 - 11:06:31 GMT-3