Hi. On Sun, Oct 28, 2018 at 05:59:52AM -0400, Gene Heskett wrote: > On Sunday 28 October 2018 02:55:09 Reco wrote: > > > Hi. > > > > On Sat, Oct 27, 2018 at 05:35:49PM -0400, Gene Heskett wrote: > > > On Saturday 27 October 2018 14:37:38 Reco wrote: > > > > Hi. > > > > > > > > On Sat, Oct 27, 2018 at 01:13:07PM -0400, Gene Heskett wrote: > > > > > Then give me an install that can be made to work in a hosts file > > > > > defined local network that can accept a gateway statement in its > > > > > e/n/i file. The default install does NOT accept it until the > > > > > network has been brought up. > > > > > > > > In another news, you cannot have a default gateway unless it's > > > > reachable according to existing routing table. > > > > That is the problem, a route -n is without a gateway assignment and has > been missing since my first attempt to install stretch.
Ok. > > > Its reachable, and listed in every hosts file here by both name and > > > address. > > > > Not before you bring up any non-local network interface. > > Then why, after giving the installer all that info, and it uses that > during the install, does it not have a gateway set after the initial > reboot or any subsequent reboot? Works for me every time I tried it, so I cannot answer that. > And nothing you can do will give it a gateway, so you wind up playing > 10,000 monkeys writing Shakespear, and a reboot for every session of > nano trying to find the magic twanger that makes it work? I have made > jessie work on an armhf but its been done after the initial reboot. An > armbian install that claims its debian 9, is the only install out of 5 > or 6 from various sources including the debian-arm iso twice. > > For the jessie install on an r-pi-3b, this /e/n/i works: > > auto lo > > # The loopback network interface > iface lo inet loopback > address 127.0.0.1 > netmask 255.255.255.0 > > auto eth0 > > # regular network for coyote.den > iface eth0 inet static > address 192.168.NN.12 > netmask 255.255.255.0 > gateway 192.168.NN.1 > > but it doesn't work if the interface address is given in address/24 > format so the netmask isn't needed. Ever consider you're doing it the wrong way? This will work: iface eth0 inet static address 192.168.NN.12 netmask 24 gateway 192.168.NN.1 > To me, thats all the evidence needed > to point the guilty finger at something in the ifup code. Accusing software of something may ease one's mind, but it won't make it work. Usually, that is. > > > > The reason is simple - a default gateway is not 'global', the > > > > kernel must decide with interface to attach to a default gateway > > > > route. So you bring a network interface up, add an address to it > > > > and only then configure a default gateway. > > > > > > Then how does one guarantee its done in that order? > > > > By using ifupdown, for instance. > > > > > And what was changed > > > to prevent its working in the newer way of doing things? They have this wonderful principle at movie industry - show but do not tell. Please provide 'ifup -v' output that shows ifupdown misbehaviour, or it did not happen. > > ifupdown works for me that way since etch was testing. > > If it does not - there's always troubleshooting in form of 'ifup -v'. > > As in "ifup -vvv eth0"? -vvv is not documented, and I'm too lazy to dig into the source to see if it does something at all. ifdown eth0 ifup -v eth0 > What log file, on stretch, would I find that trace data in? Stdout/stderr, as documented by ifup(8). > Theres not anything in /v/l/syslog w/o the -v. And there should not be anything about ifupdown, unless someone redirects stdout/stderr of /etc/init.d/networking to syslog. Reco