vnet without epair

[Nikos Vassiliadis]

On 2/10/2013 2:54 PM, Teske, Devin wrote:
> It's not in ports only because I first wanted to see where jail.conf would take us w/respect to vimages.

I see.

> However, this package not being in ports shouldn't prevented you from trying it -- it's extremely stable and as I mentioned, we've been using it heavily at $work for over 12 months now. When you download the package (*.tgz) and pkg_add it, it installs the following two files only:
>
> /etc/rc.d/vimage
> /etc/rc.conf.d/vimage
>
> NOTE: The rc.conf.d file is the "documentation" on usage
>
> If you haven't tried it, then I hope you will because I think the new jail.conf stuff falls short. Don't get me wrong, jail.conf is a great start, but simply adding the ability to manage the vnet aspect of a jail does not make a vimage (what's missing is the built-in support for generating bridges as vimages are brought up/down dynamically).
>
> I feel that before I add this to ports I need to reprogram it to use jail.conf (not directly). That will simplify its code and [should] make it smaller. I was somewhat waiting on /etc/rc.d/jail to blaze the trail for me.
>
> In short, the landscape has been changing fast enough that it's prevented me from adding this to ports, but in spite of that it's still very much real _and_ real stable.
>

Yes, of course.

I will try it and report back to you my findings.

What I - nikos - really need from a script like yours is the ability
to generate arbitrarily complex topologies with interconnected vnet
jails. Something like:
a----b----c---d
      |
      |
h----e----f---g
           |
           |
           i

Like a cut-down version of imunes[1] without the need of a graphical
user interface.

I understand that is not common case and that is why I was always using
ad hoc scripts.

But one can always hope(or write one himself/herself of course!).

1. 
http://web.archive.org/web/20120418053250/http://imunes.tel.fer.hr/imunes/

Thanks, Nikos