Introducing wicked

[Jon Masters]

Hi Olaf,

Thanks for posting about your project.

On Fri, 2010-11-26 at 10:24 +0100, Olaf Kirch wrote:

> 1. ifcfg files are dead

Ok. But I think we want to be very careful with this. Yes, it's nice to
use structured data formats but IMO we want to make sure sysadmins can
still edit files by hand without using any kind of Linux Registry to do
it (g/dconf/etc). So if you're just proposing making the format have
some kind of structured standard, ok, but please don't go too far :) 

I'll use something I can work around, but the main reason I turn off
NetworkManager on servers is that (historically) I just want a config
file that says "use these settings" that I know will work, is simple,
and well understood, and always behaves the same way on every boot. If
there's a lot more complexity, it'll be hard to sell to sysadmins.

> 3. Why not NetworkManager?
> 
> On the other hand, there's NetworkManager (and I'm getting to this point
> because Pete Zaitcev brought this up). Right now, NetworkManager doesn't
> handle bridges, bonds, infiniband, token ring - that's why I say it's a bit
> desktopish, the server environment simply has never been the focus of
> NetworkManager's development. Also, it links against a somewhat longish lists
> of fairly heavy-weight libraries (including nspr), and requires dbus - all of
> which make it pretty much impossible to use in initrd or an environment where
> space is a premium.

Good points, and I don't love NetworkManager on the server today (I do
use various bridging, etc.) but I think we're getting to the point soon
where NM might start to do some of these things nicely. So I think it's
worth being cautious not to have two solutions that half solve the
problem than one solution that is adequate enough for most folks.

Jon.