Jonathan Stone wrote:

>I suppose this isn't a good time to comment that I have broadly
>similar code -- vrrp-style packet format, IP only, no layer-2
>frobbing-- that's been working for about a year now?
>  
>
No, it's a fine time to comment. 

This little project has helped me a lot.  Regardless of the outcome it 
has already been a great learning experience.  I've learned a good bit 
about NetBSD's networking subsystem, been introduced to C, solved a few 
problems, and I am having fun.  I've already begun thinking of ways to 
build upon my experiences with this and write an even better IP 
failover/LB scheme.


By "no layer-2 frobbing", I suppose you mean your code doesn't use a 
floating MAC address.

>I really, really would prefer not to see this in the tree (which is
>why I never submitted my version).
>
>The Right Way to do this is to go through the IETF: write up a spec,
>get a new IP-multicast or a new protocol assigned.  (While you're at
>it, increase the id space to 16 bits, if not 32, and bump the version
>number.)
>  
>
If OpenBSD were to push CARP through the IETF and do all those things, 
would you care to see it in the tree then?

More id space... good point.  We need to keep up with IPv6 address space.

The rest sounds like a plan, to me.  If I can come up with an original 
idea I may very well try it.  Of course, only after I patent it ;-)


~Mike.