expired leases

[Greg.Rabil at ins.com]

Hello folks,
I'm trying to fully understand the state transitions for leases with respect to failover, and I'm hoping that someone can help me understand the situation I'm seeing.  This behavior appears consistent between version 3.0.6 and 3.1.

After the primary server has issued the "real" lease (i.e. not the MCLT lease) to the client, and the binding update has been acknowledged by the failover server, then everything is happy.  Now, say the client goes offline and does not renew the lease.  Also suppose that the failover server becomes unavailable, either not reachable from the primary or else it just fell over.

When the lease expires, you see the state change in the dhcpd.leases file on the primary:

lease 192.168.0.108 {
  starts 3 2008/11/05 19:10:54;
  ends 3 2008/11/05 19:40:54;
  tstp 3 2008/11/05 19:55:54;
  tsfp 3 2008/11/05 19:55:54;
  cltt 3 2008/11/05 19:10:54;
  binding state expired;
  next binding state free;
  hardware ethernet c0:a8:c0:7c:00:09;
  client-hostname "host-c0-a8-c0-7c-00-09";
}

If the failover is not available, as mentioned above, then this lease will remain in this state (expired, next state free) and not be available for new client requests.

What is the expected behavior here?  Shouldn't these leases eventually be freed?  If not, what is the proper procedure for making them available, assuming that the failover box is out-of-commission for a long time?  Note my previous post about attempting to set "partner-down" state on the primary using omshell, which doesn't appear to accomplish the stated goal.

Thanks,
Greg
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.isc.org/pipermail/dhcp-users/attachments/20081105/8e4d59e1/attachment.html>