I'm using Debian 8 and isc-dhcp-server.

my dhcpd.conf contains 2 subnets:

#wlan stuph
subnet 192.168.0.0 netmask 255.255.255.0 {
range 192.168.0.1 192.168.0.254;
}


# USB g_ether DHCP Service:
subnet 192.168.7.0 netmask 255.255.255.0 {
range 192.168.7.1 192.168.7.254;
}

DHCP is offered only on wlan and not USB when both are active, as above. But if I comment out wlan as below:

#wlan stuph
#subnet 192.168.0.0 netmask 255.255.255.0 {
# range 192.168.0.1 192.168.0.254;
#}


# USB g_ether DHCP Service:
subnet 192.168.7.0 netmask 255.255.255.0 {
range 192.168.7.1 192.168.7.254;
}

the USB port now offers DHCP (but not wlan of course).

the file /etc/default/isc-dhcp-server has:
INTERFACES=""
but I also tried:
INTERFACES="wlan1 usb0"

How do I get both ports to offer DHCP?

Have you double-checked that the interface names are really wlan1 and usb0? In Debian it is usually wlan0. Unless you had a previous WLAN interface on wlan0 and changed the hardware. What does /sbin/ifconfig show.

It seems that your configuration is correct, I can spot no errors. But that is no guarantee.

The "INTERFACES=" in /etc/ics-dhcp-server is mandatory. So it should contain the correct interface names.

Tail your /var/log/syslog to see DHCP requests and how they are answered. Try to increase logging verbosity: https://docs.menandmice.com/display/...+for+debugging

jlinkels

Have you looked at the logs for errors?

How are the network adapters configured and is Network Manager running?

By the way you should not use the entire subnet i.e x.x.x.1-x.x.x.254 because that would include the address for the adapter itself which could cause a conflict.

Thank you both for your replies.

ifconfig reports the following interfaces: br0, eth2, eth3, lo, usb0, and wlan1. Yes hardware changes do happen, I'm working with a SOM that gets moved to other carrier boards. That's why you're seeing the strange enumeration. I have a script that watches for these changes and the script modifies other scripts and configs that use those interfaces. Since I've started with this problem, the hardware has not changed, so the configs and scripts aren't changing at this point. And as I pointed out, the interfaces do indeed work - isc-dhcp-server only wants to talk to one interface at a time, which can't possibly be right.

syslog doesn't show any errors. I looked in dhcpd.conf and found the log file set at log-facility local7; and there was no /etc/syslog.conf, so I created one and put the line 'local7.debug /var/log/dhcpd' into it and rebooted. No /var/log/dhcpd ever gets created and still no entries in /var/log/syslog when I try pulling an address from a remote machine. And looking into "https://docs.menandmice.com/display/...+for+debugging" just confused the dickens out of me - I have no idea what to really put in there for my scenario.

As a test I had a few VirtualBox VMs that I could test a dhcp server to see if I could get it working. I could not test a USB network adapter.

The network adapters were configured using NetworkManager with a static IP address 192.168.1.1/24 and 192.168.2.1/24. The basic dhcp.conf subnet declaration was :
I did not test both networks at the same time but I configured a Mint guest for each network one at a time and the server/client worked as expected and the guest acquired the expected IP address. I used INTERFACES="eth0 eth1" in the /etc/default/isc-dhcp-server configuration file. Since debian 8 uses systemd the dhcp log messages can be viewed using journalctl -xe command.

The basics seem to work so it is a matter of how your system is configured unless I missed something...

Thanks again both of you, there was nothing wrong with this configuration. There *was* one of those self-healing scripts that was making a change in the configs after the daemon was loaded instead of before. So this one's totally my fault - as it usually is