After configuring the dhcp server to provide DHCP services for 96 subnets, starting it will consumes 100% of a single CPU for an undetermined amount of time, without providing DHCP services. I have left it running for a maximum of 24 hours. Reproducible: Always Steps to Reproduce: 1. Create a dhcpd.conf file with a large number of subnets to be serviced. 2. Start up dhcpd. 3. Observe CPU usage and DHCP request responsiveness Actual Results: dhcpd will utilize 100% of a single processor, but will not respond to DHCP requests, even after a period of 24 hours. Expected Results: I would expect this to start relatively quickly and begin providing DHCP services. I've got a dual 3.2GHz Xeon system w/ 4GB of memory and a gigabit NIC acting as a router and DHCP server for 96 different subnets. The NIC is connected to a Cisco 5500 switch via an 802.1q trunk. The dhcpd.conf file is dynamically generated from a template, where ${id} is 100-196: lease-file-name "/var/db/dhcpd.leases"; ddns-update-style ad-hoc; authoritative; option time-offset -8; default-lease-time 21600; max-lease-time 43200; # vlan ${id} subnet 10.${id}.0.0 netmask 255.255.0.0 { range 10.${id}.1.0 10.${id}.255.253; option routers 10.${id}.255.254; option domain-name-servers 10.${id}.0.1,10.${id}.0.2; option time-servers 10.${id}.0.1,10.${id}.0.2; }
I can't imagine that this is a particularly common configuration. I've searched all the documentation I can find, but nothing has indicated a better method of providing DHCP services to such a large number of subnets.
Mass reassign wrt Bug 23718, maintainer being retired.
The only thing I can suggest is trying another dhcp server - this one is going to have to be solved upstream. I suggest you contact ISC about this to see if they can fix it for you