[libvirt] Plan A or Plan B?

[Gene Czarcinski]

On 11/19/2012 03:51 PM, Laine Stump wrote:
> On 11/19/2012 03:27 PM, Gene Czarcinski wrote:
>> On 11/19/2012 01:06 PM, Laine Stump wrote:
>>> Here is an open BZ that mentions the idea of adding similar
>>> functionality for dnsmasq (see comment 8, in particular):
>>>
>>>     https://bugzilla.redhat.com/show_bug.cgi?id=824573
>> I also found Comment 2 where Dave Allan says "The intent is that
>> people who want a basic setup can configure it thorough libvirt, but
>> people who want full control of DHCP can turn off the libvirt
>> configured DHCP server, and libvirt will play nicely with a manually
>> configured DHCP server."
>>
>> Really? I was unaware of the ability to disable libvirt's dnsmasq
>> instance except when no gateway addresses are specified.
> You can disable listening on the dhcp port by simply not having a <dhcp>
> section in the network definition. The DNS service is always present if
> you define an IP address for the network though.
>
> The confusion comes when someone tries to start up their own dnsmasq,
> which also wants to run its own DNS service, on the same bridge - the
> entire thing fails. If you run a pure dhcp server though, it should work
> with no problems.
>
> I have plans to disable even the dns server if someone specifies "<dns
> enable='no'/>" in their network config (we can't do it based on the
> absence of a <dns> element, since that has never before meant "no dns"),
> but that's not implemented yet.
>
This last part about completely disabling dnsmasq even for dns make some 
sense to me. You also need to disable radvd but with my patches and 
assuming >= 2.64, then radvd is not an issue.  This really would allow a 
user to roll his own specially configured dnsmasq for handling some of 
those very special options and be much closer to what Dave Allan said.

Of course, if this option was already available, I would not have gotten 
into this DHCPv6 stuff.  But, I do believe that libvirt should support 
both IPv4 and IPv6 with the same basic functionality ... if I can do 
something in IPv4, then I should be able to do the appropriate variation 
for IPv6.

Gene