Conserver chaining -- cyclades-ts to consolidating linux server

Hi all,

I'm looking for some assistance with regard to chaining conservers. The 
documentation which is provided with conserver on this particular issue 
seems be rather unclear -- and although average-distributed.cf seems to 
be an example of the chaining case, it is proving highly opaque to me, 
and I'm not at all clear whether it works in the way I'm hoping!

Here's the situation:

* I have conserver cross-compiled and running on a whole bunch (>10) of
  Cyclades TS-4000s (one of which is 10.1.2.20), which present their
  serial ports as ttyS* to the local embedded Linux system. I'm happy
  enough with this bit -- the local conservers need to configure
  consoles as type "device" and have a devicesubst entry in an
  appropriate default{} stanza and then be referred to as ports.

My current conserver config on these looks like this:

=========================================================

# CYCLADES TS-4000 config -- cyclades has ip 10.1.2.20

config * {
        autocomplete yes;
        defaultaccess rejected;
        daemonmode yes;
        passwdfile /etc/conserver.passwd;
        primaryport 782;
        secondaryport 2001;
}

default * {
        type device;
        device /dev/ttyS.;
        devicesubst .=Pd;
        host none;
        portbase 0;
        portinc 1;
        parity none;
        rw *;
	master 10.1.2.20;
        # master 10.1.1.1;
        baud 38400;
}

access * {
        allowed 127.0.0.1;
        trusted 10.1.1.1;
}

console cyc20port1 {
        port 1;
        # host 10.1.2.20;
}

=========================================================

* What I want to do is to *also* run conserver on a Linux server
  (10.1.1.1), which would allow me to use the `console` utility on that
  machine to reach consoles on the conservers on all of the Cyclades
  devices (essentially acting as a consolidator). I believe that this
  involves the use of the "master" keyword in the configs on the
  Cyclades, but when I have tried on the individual Cyclades to say that
  the master is the Linux server (master 10.1.1.1), the secondaryport
  for the console cyc20port1 isn't actually opened when running
  conserver on the Cyclades!  (Config as above opens the secondaryport
  when run on the cyclades -- switching the master in the default stanza
  and uncommenting the host entry in console cyc20port1 makes it
  vanish.)  Assuming I need to use a 'type host' connection from the
  Linux server to the individual Cyclades consoles (which is what all
  the examples and documentation suggest) this is an obvious
  showstopper.

I know there has been some wisdom here in using identical config files 
across all conserver instances (both "master" and "slave" here) -- but I 
can't tell whether this is the correct thing to do here. I have tried 
various versions of using the same config file on both cyclades and 
server, but there's also an (obvious) conflict between using 'type 
device' as is necessary on the Cyclades devices to get conserver lashed 
up to the raw serial devices, and the use of 'type host' with 'host 
<cyclades ip>' and 'port 2001' et al in the conserver.cf on the Linux 
server.

Essentially:

* will doing what I am trying to do work? :)
* do I need separate conserver.cf files for each conserver instance in
  this case?
* what am I doing wrong in the above?

I'd be very happy to provide updates to the documentation and a worked 
example for inclusion in future releases to make this clearer in future!

Thanks

Matt

--
----------------------------------------------------------------------
Matt Johnson <mwj@doc.ic.ac.uk>               (020) 7594 8440 / x48440
Systems Programmer, Computing Support Group         Office: Huxley 225
Department of Computing, Imperial College London
----------------------------------------------------------------------