Re: "network administrator" in GNU/Hurd

[Farid Hajji]

[Cc-trimmed to help-hurd]

> You probably want to look at
> http://www.tel.fer.hr/zec/papers/zec-03.pdf .

Hey, you're quite ahead of the pack ;-). I didn't mention Marko's
paper, because network stack cloning has not stabilized yet,
and because there's still a LOT of stuff to fix there.

I've played (a little) with the patch set last month, but didn't
have that much time to dig in deeper. However, this is a very
promising effort and I hope to see it appear in 5.3 or 5.4 (and
MFCed) if it proves efficient enough!

> Peter and I discussed exporting virtual interfaces to sub-hurds and
> bridging traffic (vlans anyone?) to them. The way BSD's jail works,
> ie. demuxing ip addresses to their respective jails, is somewhat
> defunct.

Ack, even JailNG doesn't fundamentally change this, and NSC will
probably be part of a bigger virtualization effort in BSD.

> The main argument for allowing sub-hurds to attach their
> own stacks is for protocol testing and development, in other words,
> you cant have the main hurd image care about the IPv4 and v6 address
> assigned to each sub-hurd, and the situation gets even worse when
> people start running ipx, atalk, etc.. in their own sub-hurds.

Sure, this is exactly the kind of problems we're dealing with
when running multiple network stacks.

If you have two interfaces (and don't want to route or bridge
across them), you can attach one network stack per sub-hurd
and interface. Of course, this is not very practical, yet it
is a start.

Exporting virtual interfaces looks like a much better choice.
Could you please write more about this?

How would you bridge traffic?

> Chaow,
> kotry

-- 
Farid Hajji. http://www.farid-hajji.net/address.html