I've tried hunting all over the net to find some information about this,
but haven't had much luck. I figure there are some basics that I just
don't grasp, and the reading I've done hasn't helped. All the stuff I've
read basically just says you run a few IOS commands and all is right with
I have two independant connections up into the network at our facility,
each one hooks into a different switch and core router in their network.
On our local side, I have a bunch of subnets which my two routers act has a
redundant routing pair for. Ideally I'd like to have the routers talk to
the upstream router and the other router via OSPF, so that they can hand
traffic between them if necessary.
As long as I bring up zebra with only the interfaces up for the external
network and the network I want the two routers exchanging packets on, it
seems to work as I'd like. However, if I have the interfaces for all the
local subnets (aliased interfaces) up, ospfd will start trying to exchange
routing information on all those virtual interfaces, and none of them come
up. I haven't seen a way to block announcements on all but the few
interfaces I want them to talk on.
Many of the subnets are congituous, and I'd like to just announce the
Also, we are getting a whole pile of little routes from our upstream
routers, which I'd be just as happy to ignore and let the default take care
of it. Is there any way to filter incoming OSPF route announcements? I've
seen mention of it, but haven't actually seen a way to do it.
My config is basically as follows. Say my routers talk to the facility in
/30 blocks in 10.254.0.0, and my local blocks are 10.0.0.0 and 10.0.2.0:
ip ospf cost 10
ip ospf cost 1
ip ospf priority 1
ip ospf authentication-key xyzzy
ospf router-id 10.254.0.2
network 10.254.0.0/30 area 0
network 10.0.0.0/24 area 100
network 10.0.2.0/24 area 100
area 100 authentication
The other router would have a router-id of 10.254.0.6/30. I'd have, for
example, router IPs of 10.0.0.252-254/24 for one block, and then might have
10.0.2.5-7/29, 10.0.2.12-14/29, ...
Any thoughts on this?
I'm tempted to go back to BGP because it seemed like there was much more
control of the things I'd like to control that way. However, it also took
a hell of a long time for the routing table to update when a primary became
unavailable. Even on a 2.6GHz P4 with Linux it took around a minute to
revoke the old routes and instate new ones. I guess maybe I could filter
incoming announcements so that I could get something between a default
route and all the individual routes. On the other hand, I'm probably just
doing premature optimization...
"If all you have is a hammer, every problem tends to look like a nail."
Sean Reifschneider, Member of Technical Staff <[email protected]>
tummy.com, ltd. - Linux Consulting since 1995: Ask me about High Availability
Back off man. I'm a scientist. http:// Society.org/
Zebra mailing list