[Dnsmasq-discuss] Need help configuring dhcp stateful IPv6

[Simon Kelley]

On 19/05/12 10:19, Andre Coetzee wrote:
> --- On Fri, 18/5/12, Simon Kelley<simon at thekelleys.org.uk>  wrote:
>>
>> From: Simon Kelley<simon at thekelleys.org.uk>
>> Subject: Re: [Dnsmasq-discuss] Need help configuring dhcp stateful IPv6
>> To: dnsmasq-discuss at lists.thekelleys.org.uk
>> Date: Friday, 18 May, 2012, 12:10
>>
>> On 18/05/12 09:59, Andre Coetzee wrote:
>>> When I set my dhcp-range option in dnsmasq.conf with a prefix-length of
>>> 123, I get an "inconsistent DHCPv6 range" error. setting this value to
>>> 125 or greater seems to work, also setting it to 120 or less seems to be
>>> acceptable to dnsmasq.
>>>
>
> <snip>
>
>>> When I enable networking on a client machine (linux), I see
>>> "RTR-SOLICIT(eth0)" messages in the logs, but no responses or
>>> advertisements coming from the dnsmasq machine.
>>>
>>
>> OK, first thing, your post took me straight to an embarrassing bug: it's
>> all broken if the prefix length is NOT divisible by 8 - all bets are off
>> in that case.
>>
>> How is the relevant interface configured? The
>> prefix length there must  match that configured into dnsmasq.
>>
>>
>> Cheers,
>>
>> Simon.
>>
>
> Thank you Simon for this valuable information. Changing the prefix in the conf file
> to /120 and also changing the subnet mask on the interface to match (and restarting dnsmasq)
> now results in router advertisements and dhcp addresses being issued.
>
> May I suggest that the "prefix must be divisible by 8" caveat be documented in the man page
> either in the section where the dhcp-range IPv6 prefix length option is discussed,
> or under a new section "BUGS/KNOWN ISSUES"?
>

Better still is to fix the bug, now it;s know.

http://thekelleys.org.uk/dnsmasq/test-releases/dnsmasq-2.62test2.tar.gz

should do it.


There have been a couple of reasonably serious bugs found in 2.61 in the 
new DHCPv6 code, so I anticipate doing a 2.62 release just to address 
these soon.


Cheers,

Simon.


> Many thanks
> Andre
>
>