(RADIATOR) Threading support (again)

[Robert Blayzor]

Hugh Irvine wrote:
> I tend to find this easier to design, understand and operate (eash
> "service" is generally split into authentication and accounting
> instances as you describe).


I have something on this.  Hardware (cpu/memory) do not seem to be a
problem now, and running multiple 6-10) processors of Radiator seems to
work fine, especially if you can use a layer3 load balancer in front of
it to handle the requests.

My only questions is if we have say a server with four authentication
processes and four accounting processes with each pair of
auth/accounting bound to it's own virtual IP address listening on UDP
port 1812/1813, is it possible to tell those Radiator to source the
response packets from a specific address?  I know you can listen on
specific sockets, but last I knew, the UDP packets would only be sourced
from the main IP address of the box.  This is sometimes a problem for
the load balancing switch to track sessions, etc.

Any work around or do tests show that force binding Radiator to a
specific address fixes which IP the program sources packets from?

-- 
Robert Blayzor
INOC
rblayzor at inoc.net
http://www.inoc.net/~rblayzor/

telnet: Unable to connect to remote host: Connection refused

--
Archive at http://www.open.com.au/archives/radiator/
Announcements on radiator-announce at open.com.au
To unsubscribe, email 'majordomo at open.com.au' with
'unsubscribe radiator' in the body of the message.