Configuration problem

MickB mickb at free.fr
Sun May 3 14:05:58 UTC 2009


Thanks for your answer Randy.

I'm not sure to fully understand your answer, so I suspect my 
description was not clear enough ;)

To be sure, I'm going to be more precise.
My full IP addresses pool is: 192.168.0.3 to 192.168.0.14.
 From 192.168.0.3 to 192.168.0.12, it's the "normal" configuration: 
clients get the full configuration
 From 192.168.0.13 to 192.168.0.14, it's a "restricted" configuration: 
clients get the full configuration except for the router address.

Host "toto" is a machine which must be in "normal" configuration. But I 
would like this specific machine to get always the same IP address 
(192.168.0.3), that's why I added this host declaration with the 
"fixed-address" parameter.

What I've understand from the man pages is that the "range" parameter 
indicates to the DHCP server which IP from the subnet are usable for 
dynamic address allocation. That's why for my first pool, I've excluded 
192.168.0.3 from the range parameter. (If I don't do it, anyway, dhcpd 
returns me a warning about this when starting). But it doesn't change 
the fact a request from toto to the server doesn't match the class 
"test", so "option routers 192.168.0.1" from the first pool should be 
taken into account. It's not.
In the contrary, machines which must get a "normal" configuration, but 
having no host declaration, are working perfectly.

I'm currently supposing this: maybe a pool can't be matched by hosts 
having a fixed-address parameter declared. But it seems a bit strange: 
it's in contradiction with range and pool description in man pages...


Randall C Grimshaw a écrit :
> If I read your description correctly, your host declaration is an address that the host is not permitted to use. A sort of deny.
>
> In this case I suspect, if I read your description correctly, you will need a second host declaration without a fixed address that will allow it to be given a dynamic address from the allowed pool.
>
> We also see this where a fixed decalaration is made in one subnet and the machine is moved to another. It requires two host declarations - one fixed and one not fixed.
>
> An odd behavior, but one we have become accustomed to.
>
> Randy
>
> ________________________________________
> From: dhcp-users-bounces at lists.isc.org [dhcp-users-bounces at lists.isc.org] On Behalf Of MickB [mickb at free.fr]
> Sent: Sunday, May 03, 2009 7:41 AM
> To: dhcp-users at lists.isc.org
> Subject: Configuration problem
>
> Hi,
>
> I post on this mailing-list because I encounter a configuration problem
> of my DHCP server. After many tests and careful read of the man pages, I
> didn't succeed to solve it. I even suspect a bug.
>
> Configuration used:
>     * Linux 2.6.28.9
>     * Glibc 2.8
>     * DHCPd v4.1.0
>
> Here's some explanations.
> I have a single subnet: 192.168.0.0/28. I have declared two clients classes:
>     - one for clients which MAC address begins by 00:FF,
>     - one for clients which MAC address doesn't begin by 00:FF.
>
> Objective:
> - In the first class, clients get an IP in the pool 192.168.0.4 -
> 192.168.0.12, and get a router address
> - In the second class, clients get an IP in the pool 192.168.0.13 -
> 192.168.0.14, and no router address
>
> For this purpose, this DHCP server configuration is working correctly:
>
> ### Classes ###
> class "test" {
>         match if substring (hardware, 1, 2) = 00:FF;
> }
>
> ### Subnets ###
> subnet 192.168.0.0 netmask 255.255.255.240 {
>
>         pool {
>                 deny members of "test";
>                 range 192.168.0.4 192.168.0.12;
>                 option routers 192.168.0.1;
>         }
>
>         pool {
>                 allow members of "test";
>                 range 192.168.0.13 192.168.0.14;
>         }
>
>         option subnet-mask 255.255.255.240;
>         option broadcast-address 192.168.0.15;
>
> }
>
> However, it doesn't work anymore when I had a host declaration with a
> fixed-address like this:
>
> host toto {
>         hardware ethernet 00:12:A8:13:E1:40;
>         fixed-address 192.168.0.3;
> }
>
> In such case, the host "toto" doesn't get the router address. I believe
> it isn't coherent. Host "toto" doesn't belong to the "test" class, so
> the first pool in the subnet should match. The range is only for the
> case of dynamic address allocation and is no use in this case. But
> "option routers 192.168.0.1" should be taken into account, no ?
>
> Any help would be appreciated :)
>
>
> Mickaël
>
> _______________________________________________
> dhcp-users mailing list
> dhcp-users at lists.isc.org
> https://lists.isc.org/mailman/listinfo/dhcp-users
> _______________________________________________
> dhcp-users mailing list
> dhcp-users at lists.isc.org
> https://lists.isc.org/mailman/listinfo/dhcp-users
>
>
>   




More information about the dhcp-users mailing list