At 04:52 PM 6/27/2001 -0400, Przemyslaw Karwasiecki wrote:
>Thank you!
>
>Is an idea of separating craft network from production interfaces
>totally wrong?
No, not really. You might even be able to do it now using the VRFs but I
wouldn't know how to configure it up. For most people the fxp0 routes are
really simple so they don't need a full routing table for it.
Note, you could have default point out fxp0 and all traffic from the RE
that did not find a more specific route would follow the default. But all
other traffic from the PFE that did not find a match so wanted to follow
default would black-hole. You can't have two separate default routes in
inet.0.
>Also, from totaly oposite side -- is it possible to forward packets
>between production interfaces, and fxp0: ?
Nope. The reason for this one is that to get to fxp0 you have to run over
the internal link to the RE and a routing flap could end up with NxGbE or
NxOCx trying to go over that 100Mb link. Trust me, this is BAD when it
happens.
>I was able to setup routes with nexthop pointing to fxp0: network,
>but I have never seen any packet forwarded via fxp0:
Yup, we intentionally block routing between the PFE and fxp0.
>I understand that second question implies totaly different approach
>to fxp0: usage, but I just wanted to be sure that Imy observations
>are correct.
They are.
GK
This archive was generated by hypermail 2b29 : Mon Aug 05 2002 - 10:42:39 EDT