How does DHCPD determine what IP address to assign and...

Glenn Satchell Glenn.Satchell at
Fri Jan 4 14:34:46 UTC 2008

>Date: Thu, 03 Jan 2008 12:29:22 -0600
>From: "Ryan McCain" <Ryan.McCain at>
>To: <dhcp-users at>
>Subject: Re: RE: How does DHCPD determine what IP address to assign and...
>>>> On Thu, Jan 3, 2008 at  6:24 AM, in message
><200801031224.m03COxkP021770 at>, Glenn Satchell
><Glenn.Satchell at> wrote: 
>>>Date: Wed, 02 Jan 2008 14:01:03 -0600
>>>From: "Ryan McCain" <Ryan.McCain at>
>>>To: <dhcp-users at>
>>>Subject: Re: RE: How does DHCPD determine what IP address to assign and...
>>>class "DialUp" {
>>>    match if (substring(suffix(dhcp-client-identifier,7),0,5)="Async")
>>>                   or
>>>(substring(suffix(dhcp-client-identifier,6),0,5)="Async") ; 
>>>            log (info, " Matched Dialup Rule");
>>>                        }
>> ok Ryan, let's break it down. I assume you have looked at the dhcp-eval
>> man page to get the descriptions for substring() and suffix().
>> Let's look at the first evaluation and start with the assumption that
>> option dhcp-client-identifier is the string
>> suffix(option dhcp-client-identifier, 7) is the last 7 chars from out
>> string, in this case TUVWXYZ.
>> substring("TUVWXYZ", 0, 5) is the string starting at position 0 and is
>> 5 chars long, in this case TUVWX.
>> Now we are determining membership of the class if this final substring
>> is equal to "Async".
>> The second line uses a suffix of length 6. So it would evaluate the
>> suffix bit to be the last 6 chars, UVWXYZ. Doing a substring of that
>> gives UVWXY.
>> Now in your case to match the option dhcp-client-ientifier must be
>> something like
>> to log this, in the global section, ie outside the class and subnet put
>> something like
>> log(concat("dhcp-client-identifier is ", option
>> dhcp-client-identifier));
>> Looking at packet captures, global log entries, etc, does the string
>> look like that pattern? Often using a tool like wireshark will give
>> better results because there may be unprintable binary chars in the
>> string, especially sometimes a trailing null byte which you can't see
>> but gets counted in the suffix...
>> regards,
>> -glenn
>I haven't had a chance to look at the network today.  Hopefully tomorrow I'll 
have time to get wireshark up.  
>From what you typed it doesn't look like the syntax I'm using will
>always work. The string 'Async' could be anywhere in the UID.
>Is there an equivalent of "*Async*' using DHCP pattern matching?  

Regular expressions looks like it is in 4.0.0:

  match if option dhcp-client-identifier ~= "Async";

might do the trick. Is it feasible for you to upgrade to 4.0.0?

>Perhaps using the 'exists' condition?  

exists(dhcp-client-identifier) only returns true or false based on
whether that option was sent by the client.

Can the dial in servers be configured to pass the client identifier in
a more consistent way?


More information about the dhcp-users mailing list