[nylug-talk] dhcp with multiple nics?

[jh]

Chris Knadle wrote:
> On Tuesday 06 November 2007, jh wrote:
>> Nov  6 16:45:30 dhcpdns dhcpd: DHCPDECLINE on 192.168.2.83 from
>> 00:0a:e4:f3:34:96 via eth1
> ...
>> subnet 192.168.2.0 netmask 255.255.255.0 {
>>     range 192.168.2.3 192.168.2.254;
>>     option routers 192.168.2.1;
>>     option broadcast-address 192.168.2.255;
>> }
> 
>    Is there any chance that you've got devices with statically assigned IP 
> addresses within the 192.168.2.0/24 netblock?  I wonder this because in your 
> second netblock the dynamic range starts low at 192.168.2.3, and DHCPDECLINE 
> supposedly happens due to "Address conflict detected".

While this is the case for a couple of hosts, dhcpd pings first, and 
skips of a host is already up. This is not the issue at all; in fact, on 
of the windows boxes is supposed to be getting a dhcp-served "static" 
address, and that doesn't work either.

This is what happens when it runs into a host that's already up on a 
given IP address:

Nov  6 17:43:27 dhcpdns dhcpd: DHCPDISCOVER from 00:0a:e4:f3:34:96 via eth1
Nov  6 17:43:27 dhcpdns dhcpd: ICMP Echo reply while lease 192.168.2.225 
valid.
Nov  6 17:43:27 dhcpdns dhcpd: Abandoning IP address 192.168.2.225: 
pinged before offer


I removed dhcpd, and put in dhcp3-server instead. New info:

Nov  6 17:44:23 dhcpdns dhcpd: DHCPDISCOVER from 00:0a:e4:f3:34:96 via eth1
Nov  6 17:44:24 dhcpdns dhcpd: DHCPOFFER on 192.168.2.220 to 
00:0a:e4:f3:34:96 (relev_portable) via eth1
Nov  6 17:44:24 dhcpdns dhcpd: DHCPREQUEST for 192.168.2.220 
(192.168.2.2) from 00:0a:e4:f3:34:96 (relev_portable) via eth1
Nov  6 17:44:24 dhcpdns dhcpd: DHCPACK on 192.168.2.220 to 
00:0a:e4:f3:34:96 (relev_portable) via eth1
Nov  6 17:44:24 dhcpdns dhcpd: Abandoning IP address 192.168.2.220: 
declined.
Nov  6 17:44:24 dhcpdns dhcpd: DHCPDECLINE of 192.168.2.220 from 
00:0a:e4:f3:34:96 (relev_portable) via eth1: not found

What does "not found" mean? WTF?

jh


> 
>    It's obviously overly coincidental that this only seems to happen to the 
> Windows boxes, so I'd be surprised if this were the actual problem.
> 
>    -- Chris
>