Hi,
On Fri, 08 Apr 2005 10:40:23 GMT, shufler <usenet@shufler.net> wrote:

>I haven't been able to find much information about this at all, be it
>from Google, IRC, any form of Slackware "manual" I can find, any other
>distro's manual, or even books (I've seemed to have discovered that most
>Linux books I've looked at are void of decent networking information in
>general, usually stopping at configuring a device).

Because you go into routing space, away from majority users? What you
trying to do? I'm thinking either low-routing or high-level load
sharing like a server or compiler farm.

I've had three NICs running, and that is scratching for interrupts on
some machines. Disable serial ports & printer to get some.
>
>Most people I have spoken with have indicated that this is probably a
>per-program setting, meaning if a program can support using a specific
>interface, then good news, otherwise I'm SOL. I honestly believe this is
>nonsense, and what I am proposing is quite possible (or at least, I am
>hoping this is the case). As far as I can tell, the processes I want to
>do this with are not equipped for me to specify the device.
>
>Conceptually, I see this as piping processes to interfaces, though I'm
>not claiming to be correct on this.
Well, my thought is normally a high-level program knows about an
endpoint, routing is different matter. Source routing went out
ages ago.
>
>Any thoughts on this matter are appreciated. Also, if there is somewhere
>else (another group) this question would be more appropriately targeted,
>please let me know.

Very much depends what you are doing, clustering, routing or some other.
clustering -> beowulf
server farm -> apache
compiler farm -> distcc

what's yours?

Cheers,
Grant.


>
>--JW