[Dnsmasq-discuss] [PATCH] Change dhcp_release to use first address when no IP subnet matches

[Brian Haley]

On 8/22/19 6:50 PM, Simon Kelley wrote:
> On 26/04/2019 21:03, Brian Haley wrote:
>> Currently, dhcp_release will only send a 'fake' release
>> when the address given is in the same subnet as an IP
>> on the interface that was given.
>>
>> This doesn't work in an environment where dnsmasq is
>> managing leases for remote subnets via a DHCP relay, as
>> running dhcp_release locally will just cause it to
>> silently exit without doing anything, leaving the lease
>> n the database.
>>
>> Change it to save the first IP it finds on the interface,
>> and if no matching subnet IP is found, use it as a fall-back.
>> This fixes an issue we are seeing in certain Openstack
>> deployments where we are using dnsmasq to provision baremetal
>> systems in a datacenter.
>>
>> While using Dbus might have seemed like an obvious solution,
>> because of our extensive use of network namespaces (which
>> Dbus doesn't support), this seemed like a better solution
>> than creating system.d policy files for each dnsmasq we
>> might spawn and using --enable-dbus=$id in order to isolate
>> messages to specific dnsmasq instances.
> 
> 
> The address we're determining here is the address used for the "server
> identifier" for the DHCP lease. In the case of a client on the other
> side from a DHCP relay, dnsmasq uses one of the addresses of the
> interface through which the packet arrived. This patch also uses one of
> the addresses, but it doesn't choose it in the same way, so for an
> interface with multiple addresses, dhcp_release may pick a different
> address than the one dnsmasq did, the server_id in the release message
> will not match, and the operation will silently fail again.
> 
> If the interface has only one address, a common case, then that has to
> be chosen, and all works. For more robustness, it makes sense for this
> patch to pick one of the addresses in the same way that the dnsmasq code
> does, namely, by doing
> 
> ioctl(..., SIOCGIFADDR, ...)
> 
> in the case that none of the addresses match the address of the lease.
> 
> 
> A patch to do that instead of using the first one returned by netlink
> would be great.

Thanks Simon, I'll update it and send a v2.

-Brian