Over the summer we migrated from Netware 6.5 SP7 to OES2 SP1 on Linux. Ever since then, we run into problems where the dhcp pool for a few subnets run out of addresses. We havent added any extra computers. The errors we get are this (in /var/log/dhcpd.log):

Sep 1 09:04:36 khsfas01 dhcpd: DHCPDISCOVER from 00:1b:21:1e:d3:3f via 10.3.0.1: network 10.3.0/23: no free leases
Sep 1 09:04:36 khsfas01 dhcpd: DHCPDISCOVER from 00:1b:21:1e:cd:a5 via 10.3.0.1: network 10.3.0/23: no free leases
Sep 1 09:04:36 khsfas01 dhcpd: DHCPDISCOVER from 00:1b:21:1e:d2:a8 via 10.3.0.1: network 10.3.0/23: no free leases
Sep 1 09:04:37 khsfas01 dhcpd: DHCPDISCOVER from 00:1b:21:1e:d0:81 via 10.3.0.1: network 10.3.0/23: no free leases
Sep 1 09:04:37 khsfas01 dhcpd: DHCPDISCOVER from 00:1b:21:1e:ce:b0 via 10.3.0.1: network 10.3.0/23: no free leases
Sep 1 09:04:37 khsfas01 dhcpd: DHCPDISCOVER from 00:1b:21:1e:cd:33 via 10.3.0.1: network 10.3.0/23: no free leases
I searched around and found some info where this happens when PXE clients dont present a GUID, so each client takes up two addresses - one for PXE, one for Windows. I dont know hot to confirm this, but this is what seems to be happening as we havent added any more workstations to the subnet nor reduced the size of the pool. Our DHCP and Proxy DHCP server are on two different servers so making a class to have the PXEClient addresses have a shorter lease wont work. Surely, someone must have run into this and a fix for it, besides setting the lease time for each subnet to 5 minutes.

Any help is appreciated!

Thanks!

-Rob