On May 10, 2011, at 7:25 AM, DJ Lucas wrote: > * The three items above are the best I think we can do with it and > should cover > 99% of all cases, the known exceptions being starting the > dhcp client or ppp client manually, and possibly manual configuration of > wireless interfaces (I've never configured wireless in LFS - also what > about VPN tools started manually?). I believe Bryan is already on board > with these changes, Bruce, Jeremy, Zach?
It sounds like a good plan to me.
The one change I'd suggest is to put the transient files (the copied config
files) into /var (or even someplace in /tmp) though, and not in /etc. Otherwise
/etc must be writable for the network scripts to work as expected, and
boot-time cleanup will need to clear files from /etc, both of which seem
undesirable to me.
As for the less-common cases:
General manual configuration: If the ifup/ifdown system is only going to
support a single address per family per named interface, it should be safe to
always `ip -4/-6 addr flush` in the down script, rather than manually removing
individual addresses. This will ensure things are in a known state when/if you
go back up, even if there were changes outside the on-disk configuration. The
downside is you can't manually add a temporary address without having the
ifdown script destroy your changes. I don't know which behavior is desired.
Manual DHCP: You could call the DHCP-down behavior in all ifdown calls and just
ignore the result if DHCP wasn't configured. I can't think of a case where it
would be inappropriate to stop DHCP on an interface that's going down,
regardless of the on-disk or manual configuration.
VPNs: I wouldn't try to handle them. IPSec doesn't necessarily fit into the
"kernel network interface" model that ifup/ifdown are built around and
commercial VPN clients muck with far to many things to handle in a generic
network script. There might be some subset that can be handled but I don't
think there's a general solution.
WiFi: Should fit the model, if someone writes a service script for it. There's
some work around how to manage SSID switches and the like, but that's outside
the scope of this system -- at some point there's still going to be an
ifdown-reconfigure-ifup set of calls.
PPPoE: So long as the scripts are call with the Ethernet interface name and not
the PPP interface name it should fit into this model. And that seems reasonable
to me, even if you're moving a single Ethernet port between PPPoE and normal
networks.
Raw PPP: Has it's own up/down scripts and people using raw PPP are probably
expecting to use those anyway.
Zach
smime.p7s
Description: S/MIME cryptographic signature
-- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
