Exponential Lease Time

[Nicholas F Miller]

Here are some observations from my testing. The results are not consistent enough to be used in production but it is still pretty interesting.

1) I haven't seen one lease that had a renewal = 1. They all seem to start with a renewal = 2 (Windows and Mac clients, wired and wireless). This may be a result of using helper addresses on the routers and WISM2s.
2) iPhones seem to jump straight to the longest lease time (if renewal = 3).
3) Windows phones jump straight to the second longest lease time (if renewal = 2).
4) Android phones behave the same as Windows and Mac clients.

Here are the values I used for my testing which were defined in the pool statement:

        if renewal = 3 {
                default-lease-time 86400;
                max-lease-time 86400;

        } elsif renewal = 2 {
                default-lease-time 14400;
                max-lease-time 14400;
                set renewal = 3;

        } elsif renewal = 1 {
                default-lease-time 3600;
                max-lease-time 3600;
                set renewal = 2;

        } else {
                default-lease-time 900;
                max-lease-time 900;
                set renewal = 1;
        }

_________________________________________________________
Nicholas Miller, OIT, University of Colorado at Boulder




On Aug 31, 2012, at 2:42 PM, Norman Elton wrote:

>> Out of interest, does processing that sequence of statements create more or
>> less load than just processing renewals without dynamic lease lengths ?
> 
> Definitely deserves further testing. We've done some scripting before
> and it didn't contribute to a noticeable load.
> 
>> Also, when the lease expires, does "renewal" stay set or does it get removed
>> from the lease record ?
> 
> After the lease expires, it appears that the "renewal" variable gets
> wiped. Again, this could use some more testing.
> 
> I pretty much hacked this up. I'd love some feedback from folks who
> are more familiar with how the variables work and are designed to be
> used.
> 
> Norman
> _______________________________________________
> dhcp-users mailing list
> dhcp-users at lists.isc.org
> https://lists.isc.org/mailman/listinfo/dhcp-users