[arch-releng] [PATCH 2/5] [configs/releng] Enable systemd services

[Dave Reisner]

On Sun, Sep 16, 2012 at 02:42:59PM +0200, Pierre Schmitz wrote:
> Am 16.09.2012 14:38, schrieb Dave Reisner:
> > On Sun, Sep 16, 2012 at 02:01:03PM +0200, Pierre Schmitz wrote:
> >> Am 16.09.2012 11:17, schrieb Thomas Bächler:
> >> > Am 16.09.2012 10:40, schrieb Pierre Schmitz:
> >> >> Am 12.09.2012 09:10, schrieb Gerardo Exequiel Pozzi:
> >> >>> dhcpcd at eth0.service ntpd.service || true' \
> >> >>
> >> >> This is not going to work. You only enable dhcp on eth0 here. It should
> >> >> be enabled on all devices. You cannot know if there is an eth0 device or
> >> >> if it is the one connected to the lan. We just need to implement this:
> >> >> https://bugs.archlinux.org/task/31543
> >> >
> >> > The service proposed there uses dhcpcd -w, which is a bad idea for us.
> >>
> >> Why? Don't we get a login when the network target is not reached?
> > 
> > The daemon would have 90s or so (whatever the default timeout is) to
> > pick up a network connection and fork. If it doesn't do that, it's going
> > to stall boot until the timeout, and then get brutally killed by systemd.
> 
> Well, then this unit file is kind of broken. What's the problem with
> just using -b as we do on the current media. I don't know of any issues
> with this. The network target could be reached even if there is no
> network, but for the live media I don't see a real problem here. (Or we
> use networkmanager, but that seems to do something similar). Either way,
> this is still better than just trying to run dhcp on eth0 only.

It's fine to do it that way. I'm not convinced that using -w is the
proper thing to do with the default dhcpcd.service anyways.