DHCPREQUEST of <null address>

[Alan Jenkins]

Alan Jenkins wrote:
> David W. Hankins wrote:
>> On Sun, May 04, 2008 at 01:31:45PM +0100, Alan Jenkins wrote:
>>  
>>> May  4 11:52:36 alan-eeepc dhclient: DHCPREQUEST of <null address> 
>>> on eth1     
>>
>> ISC DHCP logs of the form:
>>
>>         snprintf (msgbuf, sizeof msgbuf,
>>                  "DHCPREQUEST for %s%s from %s %s%s%svia %s",
>>
>> So I know not what sources you are running that nets 'DHCPREQUEST of '
>> in syslog.
>>
>> <null address> will appear whenever piaddr() is called with an
>> address of zero length.  In stock sources, I believe this to be
>> impossible on that log line ("DHCPREQUEST for ").
>>   
> Well caught.  For completeness, I'm using 
> dhcp3-client-3.0.6.dfsg-1ubuntu9 from Ubuntu Hardy.  I've downloaded 
> the source package corresponding to my installation.  It sounds like 
> you might be referring to the dhcp server and not the client, because 
> in the original source tarball the message starts "DHCPREQUST on", and 
> not "for".  But you're certainly right that the original sources don't 
> include a line "DHCPREQUEST of".
>
> The line "DHCPREQUEST of" comes from a particular Ubuntu-specific 
> patch called "dhclient-more-debug".  That probably explains why I 
> couldn't find any useful hits on Google.
>
> Here's the relevant hunk of the patch.  All it's supposed to do is add 
> the IP address dhclient requests to the log messages.  It's this 
> address which is reported as being null.
>
> @@ -1703,7 +1706,8 @@
>                        client -> packet.secs = htons (65535);
>        }
>
> -       log_info ("DHCPREQUEST on %s to %s port %d",
> +       log_info ("DHCPREQUEST of %s on %s to %s port %d",
> +               piaddr(client->requested_address),
>              client -> name ? client -> name : client -> interface -> 
> name,
>              inet_ntoa (destination.sin_addr),
>              ntohs (destination.sin_port));
>
> W.r.t to trying stock sources, I've just compiled ipv6 support into my 
> kernel to see if I can get dhcp-4.0.0 working.  Even when I included 
> the -4 command-line option, it complained
>
> "Error opening '/proc/net/if_inet6' to list IPv6 interfaces"
>
> Anyway, it works now so I just need to leave it long enough and see 
> what happens.
Nope, ISC dhcp version 4.0.0 doesn't do any better.

May  6 00:37:45 alan-eeepc dhclient: bound to 144.32.61.4 -- renewal in 
1632 seconds.
May  6 01:04:57 alan-eeepc dhclient: DHCPREQUEST on eth1 to 
144.32.129.66 port 67
May  6 01:05:39 alan-eeepc last message repeated 4 times
May  6 01:06:40 alan-eeepc last message repeated 4 times
<and so on...>
May  6 01:20:45 alan-eeepc last message repeated 4 times

Obviouslly it doesn't tell me that it's requesting a <null address>, 
because it lacks the debug patch, but I don't expect it's doing anything 
different.

So, do you have any idea why client->requested_address would be set to 
"<null address>" before the lease expires?

Thanks
Alan