Lease limiting via Option 82, failover and OMAPI

[Chris Poon]

After reading the code, I think these are all related. First off,
the easy question - it looks like OMAPI and dhcpd disagrees on how a lease
should be hashed by hardware address. dhcpd hash a lease by ONLY the hardware
address while OMAPI queries the hash with the hardware type tagged on in front.
Which should be the right way? I'm leaning towards keeping the hardware type
and mod dhcpd (don't know extensive that is yet, and I'm not certain if I'm
capable of doing so), but it's definitely easier to hack the OMAPI portion
to drop the hardware type before looking up the lease.

Then comes the bigger question - Lease limiting via Option 82 sub-classing
combined with failover. From my analysis, I see 2 things that need to happen:
1) failover protocol must have someway to communicate the lease limit, by
sending the circuit-id/agent-id/remote-id or the subclass that was created
on the server that gave the lease so that the other side can replicate the
classify action
2) Provision new sub-classes on the fly via OMAPI. This is very useful for
the except case where some clients get a different limit.

Being a newbie to the code, I can already see the complexity in the first one.
The 2nd one should be less of a problem - it's just a matter of digging all
the relevant code that creates a class, but it's still quite complicated with
respects to matching expression and stuff like.

Is there plans to do such thing? I'm certain there are enough interests out
there to have lease limiting via Option 82 work with failover, but it seems
too complicated to implement.