[Apt-proxy-users] process characteristics / setting up xinetd

[mi <mic...@gm...>]

Hello,
Recently i moved to xinetd, and i'm now wondering how a-p handles its 
requests.
Does it multithreading ? Will there be new processes launched by inetd 
for every request of one client ?
How about multiple requests by different clients ?
And how to set up xinetd correctly...

This is the default inetd.conf entry for ap, set by debian debconf:
<<
9999    stream  tcp  nowait.400  aptproxy  /usr/sbin/tcpd 
/usr/sbin/apt-proxy -l /var/log/apt-proxy.log
 >>

The conversion script 'xconv.pl' didn't know how to 'translate' the 
inetd.conf  <nowait.max> entry.
In the documentation, i found the following xinetd.conf options:
<<
instances:  The maximum number of requests a particular service may 
handle at once.
per_source: The maximum number of connections a specific IP address 
can have to a specific service.
 >>

It's not matching exactly, but, anyway:
What would be the appropriate values for, say, 10 clients ?

This is what i set up so far in xinetd.conf:

<<
service 9999
{
         disable     = no
         port        = 9999
         socket_type = stream
         protocol    = tcp
         wait        = no
         instances   = 50
         per_source  = 10
         user        = aptproxy
         server      = /usr/sbin/apt-proxy
         server_args = -l /var/log/apt-proxy.log
}
 >>

And, serving a LAN with 100 clients, should a-p better run as 
standalone daemon there ?

Also, while there already are 82 services added for the debian 
distribution, i wonder why debconf didn't add
an entry to /etc/services ?

Another question is how important would you consider it to invoke ap 
though tcpd (say, on a production machine) ?
Though i didn't test it until now, I guess xinetd.conf then should read
<<
flags       = NAMEINARGS
server      = /usr/sbin/tcpd
server_args = /usr/sbin/apt-proxy -l /var/log/apt-proxy.log
 >>

What do you think  =)

-- 
			
			mi.