Hi all,
I recently ran an apt-get upgrade and things seemed to have worked fine, until this morning (about 10 days later) when all windows clients failed to acquire an IP address. if i read the conf file correctly the max lease time is set to 2 days. so the upgrade may not have caused the problem, but nothing else was changed recently. i assume there is a problem with the dhcp3-server (which i didnt configure - i inherited this network).
we run debian etch kernel 2.6.8.
i checked /etc/default/dhcp3-server which seems to have the correct entry (eth1). init.d script also points at that file.
any idea where to start? at this point i dont even know what to google for.
funnily enough, when i did an apt-get update it couldnt resolve the proxy server name to its address. i changed apt.conf from computer name to ip address. apt-get update ran fine. i checked hosts entries and it should have resolved. i could even ping the proxy, under its name.
well, these are the symptoms. any idea? cheers, rotezecke

If the DHCP daemon is running on the server the first thing I would do (after verifying all the network settings on the server) is fire up Wireshark on a client and capture the DHCP traffic from a client to the server and analyze what is happening during the exchange.
http://broadcastengineering.com/infr...tutorial_part/

Also sounds like your servers DNS resolution isn't working properly.

Is there anything showing up in the server logs ?

Well as mentioned check for your dhcp dameon.
if that is not running for any reason just issue
as far as name resolution I would check /etc/resolv.conf and see what your nameservers are and if you can ping them.

Hi,
relevant entires in ps aux:
/usr/sbin/named -u bind
/usr/sbin/dhcpd3 -q eth1 (which seems to be correct).
restarting these services doesnt help. the configuration of these settings didnt change in over a year. but as i said - i did an upgrade a couple of weeks ago.
and yes, i can ping both my nameservers.
/var/log/syslog:
dhcp & dns related errors look like something like this:
time servername dhcpd: comutername.domain.org: temporary name server failure
time servername dhcpd: DHCPDISCOVER from (mac address) via eth1: network xxx.xxx.xxx/24 no free leases

i shall have a go at wireshark and post the results (if any)
Cheers,

Well just from the errors that I see it looks like the problem is that you have used all leases. is comutername.domain.org actually defined in your dhcp.conf file? If so try commenting that then restart the server.

Hiya
not sure how to define comutername.domain.org dhcp3.conf
i cant copy paste the entry, but it goes something like this:

ddns-update-style interim;
option domain-name "domain.org";
default lease time ...;
max lease time ...;
authorative;
update static leases: off;
log-facility local7;
subnet 192.xxx.xxx netmask 255.255.255.0
option netbios-name-servers computername.domain.org
...
option domain-name-servers computername1.domain.org computername2.domain.org
(please ignore syntax)
i changed the netbios and domain name server entries to ip-addresses (didnt help) - as i assume all this goes back to DNS problems.
there is something with bind9 and refreshing keys when changes are made. maybe that also applies to debian upgrades. this is far beyond my linux skillz though.
if this makes any sense to anyone, please let me know. i may have to move this thread and or change the title.
cheers,

What is the possibility of posting your full dhcpd.conf file. Also what editor are you using?

dhcp3.conf:

# $Id: dhcpd.conf,v 1.1.1.1 2002/05/21 00:07:44 peloy Exp $
#

# The ddns-updates-style parameter controls whether or not the server will
# attempt to do a DNS update when a lease is confirmed. We default to the
# behavior of the version 2 packages ('none', since DHCP v2 didn't
# have support for DDNS.)

ddns-update-style interim;

# option definitions common to all supported networks...
#option domain-name-servers ns1.example.org, ns2.example.org;
option domain-name "nimfm.org";

default-lease-time 86400;
max-lease-time 172800;

# If this DHCP server is the official DHCP server for the local
# network, the authoritative directive should be uncommented.
authoritative;

# deny client-updates;
update-static-leases off;

# Use this to send dhcp log messages to a different log file (you also
# have to hack syslog.conf to complete the redirection).
log-facility local7;

# No service will be given on this subnet, but declaring it helps the
# DHCP server to understand the network topology.

subnet 10.0.0.0 netmask 255.255.255.248 {
option routers 10.0.0.6;
}

# A slightly different configuration for an internal subnet.
subnet 192.168.9.0 netmask 255.255.255.0 {
option netbios-name-servers nim-fm.nimfm.org;
option netbios-node-type 2;
option subnet-mask 255.255.255.0;
option routers 192.168.9.10;
default-lease-time 86400;
max-lease-time 172800;
option domain-name-servers nim-fm.nimfm.org, commune.nimfm.org;
}

host nimfm-studio-1n {
hardware ethernet 00:a0:c9:32:45:22;
fixed-address nimfm-studio-1n.nimfm.org;
}
host nimfm-studio-2s {
hardware ethernet 00:11:d8:b5:a4:f2;
fixed-address nimfm-studio-2s.nimfm.org;
}
host nimfm-library {
hardware ethernet 00:17:31:a4:8c:6d;
fixed-address nimfm-library.nimfm.org;
}
host Nimfm-library {
hardware ethernet 00:1a:92:4c:7d:c1;
fixed-address nimfm-library.nimfm.org;
}
host officer {
hardware ethernet 00:11:d8:b5:a7:6d;
fixed-address officer.nimfm.org;
}
host hippies {
hardware ethernet 00:16:17:39:36:be;
fixed-address hippies.nimfm.org;
}

host printer1 {
hardware ethernet 00:80:77:0a:15:d8;
# ddns-hostname "printer1";
fixed-address printer1.nimfm.org;
}

host pre-production {
hardware ethernet 00:11:d8:b5:a4:eb;
fixed-address pre-production.nimfm.org;
}

# host moped {
# hardware ethernet 00:11:50:11:5a:0d;
# fixed-address moped.nimfm.org;
# }

## In case wirelass lappies need wire
# host Moped {
# hardware ethernet 00:0f:1f:2b:4a:e2;
# fixed-address moped.nimfm.org;
# }
host ethnic {
hardware ethernet 00:11:50:db:d2:d4;
fixed-address ethnic.nimfm.org;
}
host Ethnic {
hardware ethernet 00:0f:1f:0b:bc:53;
fixed-address ethnic.nimfm.org;
}

i usually use vim, as i am too lazy to properly learn vi.

try adding a range to the subnet of 192.16.9.0. Add it for say 15 addresses. this can be done by adding in the following line
lets see if that helps.

That is what I was thinking...

Actually on some systems vim actually point to vim. From my experience when you learn vim you know most of vi.

oh, thats what it is! i got along just fine in vi until i met debian where vi does not seem to point at vim. it drove me nuts!

I think in this setup DHCP requires DNS to be working - but it's broken. in our setup a range wasnt needed. the IPs are defined in bind's db.files. dhcp cant find the domain.

This thread is looking at the problem from the wrong end (well, I guess that's my fault ain't it?) and I herewith close it. I shall post a conclusion once DNS is fixed (if ever). Thank you all for your input (and watch out for a new thread on DNS and bind - possibly coming soon)
Cheers,

Yeah and this is why I should not post as soon as I get out of bed. but yeah vim will point to vi but on debian they are different.

problem solved.
DNS was broken - DHCP relied on DNS. I used # instead of ; to comment in the zone files. also, i update serial number when i made changes. at some point bind must have decided it had enough of my behavior and stopped working. i made the changes, ran rndc relaod and bobs our uncle.
thanks for all your input.
by the way, we have no range setting as we do not allow unauthorised computers on the network.
cheers, rotezecke