On Thu, May 22, 2008 at 03:05:13PM +0200, Frans Pop wrote: > On Thursday 22 May 2008 14:04:40 Josef Wolf wrote:
> > I am curious whether it is really a good idea to point > > $HOSTNAME.$DOMAINNAME to 127.0.1.1. While this works fine for services > > on the local host, it fails if the address is provided to other hosts. > > For example, if one installs dnsmasq, the $HOSTNAME.$DOMAINNAME will > > resolve to 127.0.1.1 for the whole network, which is obviously not the > > right thing to do. > > I have thought about this a little bit, but ppp-udeb now does *exactly* the > same thing as netcfg. It did not seem a good idea to deviate from that > without very good reasons and I could not think of any. > > > Maybe it would be better to set a hosts entry from ppp-udeb only if > > there are no other network interfaces available. When other interfaces > > exist (e.g. LAN on eth0 if the host in question is a DSL-router), this > > setting should be done by netcfg when it configures this interface. > > No. You currently use either ppp-udeb or netcfg, never both. If you use > ppp-udeb, the PPPoE interface is expected to be your primary (and in most > cases only) network interface. But wasn't ppp originally intended to connect multiple networks? In this (IMHO _very_ common) scenario you would always have at least two interfaces: ppp to the provider and a statically configured interface to the local network. Further, you would probably provide DHCP/DNS servers (e.g. via dnsmasq) for the local network. This gives the setup I mentioned above. I don't see the point in having ppp as your one and only interface. Is this scenario really _that_ common? Who would want to use such a setup? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
