mbostwick

[Log in to get rid of this advertisement]

Hello I am trying to get fedora dhcp set up but for some reason the clients are not getting dns. If I statically set dns everything works as expected. Here is the dhcpd.conf:

deny client-updates;
ddns-update-style none;
ddns-updates off;
option T150 code 150 = string;
one-lease-per-client false;
allow bootp;
#
# DHCP Server Configuration file.
# see /usr/share/doc/dhcp*/dhcpd.conf.sample
# see 'man 5 dhcpd.conf'
#
allow client-updates;
allow unknown-clients;

subnet X.91.79.0 netmask 255.255.255.0 {
interface eth5;
range X.91.79.10 X.91.79.240;
default-lease-time 604800;
max-lease-time 604801;
option domain-name "cisconet.ucmo.edu";
option subnet-mask 255.255.255.0;
option broadcast-address X.91.79.255;
option routers X.91.79.252;
option domain-name-servers X.91.3.113;
option time-offset -3600;
option netbios-name-servers X.91.3.51;
}

..X is ip that is being used...Has anyone else had a similar issue with Fedora release 9 (Sulphur), dhcpd --version: isc-dhcpd-4.0.0?

mbostwick

Ok so in further looking at the issue, I found that windows was just using its old lease (which didnt have dns for some odd reason) and a full release(ipconfig /release) was required in order to show a network failure. All of dhcp has failed. It appears(from wireshark) that the problem is due to to the server not granting any more request, because I can see the clients asking for the dhcp address .

avijitp

Do you see anything on /var/db/dhcpclient/leases? How about the syslog messages.

mbostwick

I am getting closer to the problem, it appears in /var/log/messages. When a client tries to make a request this output comes out:
Oct 29 18:23:11 cisconet dhcpd: DHCPREQUEST for 0.0.0.0 from YY:YY:YY:YY:YY:YY via eth5: unknown lease 0.0.0.0.

-I have removed the mac address(YY)

avijitp

Make sure that he dhcp service running in the server? Or if you have made a recent change in the conf, do a restart and see how it goes.

mbostwick

I have not made any recent changes to the server(other then the power going out to cause all of this...so I am unable to track changes that may have been made by someone else). In /var/lib/dhcpd/dhcpd.leases there are some leases . Also dhcpcd is running:
ps awx | grep dhcpd
2277 ? Ss 0:00 /usr/sbin/dhcpd
Do you see any trouble in the config above ? Also looking at the leases, it dose not appear that I am anywhere near the max based on the range.

mbostwick

Hmmm, in looking around it appears the windows client is doing something strange:
"That makes no sense at all. Clients just shouldn't do that. The
client is clearly broadcasting this packet, as it's being picked up by
your relay.

What could be happening here is that the client is sending a
DHCPREQUEST with no requested-address option. In this case, piaddr(0)
would print "0.0.0.0". This is not valid according to RFC 2131.
"

mbostwick

Ok, so I have fixed it and the problem was clearly client side. The windows xp machines were requesting an improper address and were using cache before that. Doing an ipconfig /release on the clients then restarting the machine allows them to grab the proper ip.