Request for comment: a new software to manage linux networking features

[martin f krafft]

[Ccing netconf-devel with permission]

also sprach namnd <namnd at fpt.com.vn> [2008.02.18.1414 +0100]:
> I have read all available materials on your web site and on
> mailing  lists, and I have not found the advantage in netconf to
> shift my interest to. Can you convince me about it.

I think the strongest argument for netconf is that it tries to be as
stateless as possible and that it is pretty much distro-independent.
Moreover, it does not touch the kernel interfaces but leaves it up
to scripts to do so, for ultimate flexibility.

It's also aimed at replacing ifupdown, not overriding it.

> Daemon based approach doesn't allow the system to boot in single mode.
> It's a disadvantage, isn't it?

Single-mode does not mean that all daemons have to stop.

> By the way, please explain your definition of "stateless".

I don't store state information about interfaces but ask the kernel
if I need any data. That way, my virtual "model" of the interfaces
cannot get out of sync from the way things really are.

> Please tell me the problems you are interested to address.

Provide comprehensive network configuration services in an
extensible fashion with maximum flexibility for the admin.

> netupdown, when installed, completely overrides ifupdown (ifup and
> ifdown will be symlinks to netupdown). The usage syntax is
> completely compatible so other scripts can continue calling
> ifup/ifdown as is. If I want to remove ifupdown, it'll be an easy
> job for netupdown.

Interesting. Do netupdown has full compatibility of all
/etc/network/interfaces options, including hooks and mappings?

> In your opinion, what's the problem that netupdown doesn't help?

I don't know netupdown enough to identify problems. I said it looks
good. It's just not what I have in mind.

That said, I am the netconf author, so I am biased. Maybe other
people can chime in and compare our two approaches?

>> As a Debian developer with good connections to Fedora, I think
>> I am in a good position to push this, and netconf has attracted
>> a lot of interest already, even though noone has joined the
>> development team yet (the code isn't ready enough yet).
>>
> I believe that after the first prototype, you will need to rewrite
> the  code at least once. It'll take you several work months before
> people  start joining your project.

Yes, I am aware of that. Unless I can get people like you interested
earlier so that they help. Because I don't have very much time for
netconf...

> Currently, netconf is too young, mostly an idea instead of
> a product. I still recommend you to use the netupdown code base at
> the starting point to start coding in C++, if you do so, I can
> join your project.

So you want me to throw away everything I've done so far and adopt
your code under the netconf project name?

-- 
 .''`.   martin f. krafft <madduck at debian.org>
: :'  :  proud Debian developer, author, administrator, and user
`. `'`   http://people.debian.org/~madduck - http://debiansystem.info
  `-  Debian - when you have better things to do than fixing systems
 
"was aus liebe getan wird,
 geschieht immer jenseits von gut und böse."
                                                 - friedrich nietzsche
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature (see http://martin-krafft.net/gpg/)
Url : http://lists.alioth.debian.org/pipermail/netconf-devel/attachments/20080220/b3eb6323/attachment.pgp