README
1 The patch I have attached lets me get the behavior I wish out of
2 dnsmasq. I also include my version of dhclient-enter-hooks as
3 required for the switchover from pre-dnsmasq and dhclient.
4
5 On 8/16/05, Joseph Tate <dragonstrider (a] gmail.com> wrote:
6 > I'm trying to use dnsmasq on a laptop in order to facilitate openvpn
7 > connections. As such, the only configuration option I'm concerned
8 > about is a single server=3D/example.com/192.168.0.1 line.
9 >
10 > The way I currently have it set up is I modified dhclient to write its
11 > resolv.conf data to /etc/resolv.conf.dhclient and configured
12 > /etc/dnsmasq.conf to look there for its upstream dns servers.
13 > /etc/resolv.conf is set to nameserver 127.0.0.1
14 >
15 > All of this works great. When I start the openvpn service, it the
16 > routes, and queries to the domain in the server=3D line work just fine.
17 >
18 > The only problem is that the hostname for my system doesn't get set
19 > correctly. With the resolv.conf data written to something other than
20 > /etc/resolv.conf, the ifup scripts don't have a valid dns server to do
21 > the ipcalc call to set the laptop's hostname. If I start dnsmasq
22 > before the network comes up, something gets fubar'd. I'm not sure how
23 > to describe it exactly, but network services are slow to load, and
24 > restarting networking and dnsmasq doesn't solve the problem. Perhaps
25 > dnsmasq is answering the dhcp request when the network starts?
26 > Certainly not desired behavior.
27 >
28 > Anyway, my question: is there a way to have the best of both worlds?
29 > DHCP requests to another server, and DNS lookups that work at all
30 > times?
31 >
32 > My current best idea on how to solve this problem is modifying the
33 > dnsmasq initscript to tweak /etc/dhclient-enter-hooks to change where
34 > dhclient writes resolv.conf data, and fixing up /etc/resolv.conf on
35 > the fly to set 127.0.0.1 to the nameserver (and somehow keep the
36 > search domains intact), but I'm hoping that I'm just missing some key
37 > piece of the puzzle and that this problem has been solved before. Any
38 > insights?
39 >
40 > --
41 > Joseph Tate
42 > Personal e-mail: jtate AT dragonstrider DOT com
43 > Web: http://www.dragonstrider.com
44 >
45