The ipcs configuration file:
--------------------------
For information about the config file, see:
        IPCS_DISK:[IPCS.NETMAILER.EXE]nm_config.txt

The ipcs configuration file is named:
        on VMS:  IPCS_DISK:[IPCS.NETMAILER.EXE]netmailer.cfg
        on unix: nm_config.dat (probably in /opt/ipcs2/etc/)

        On VMS, the logical name IPCS$NETCFG *should* be an
        option to override the default config file, but does not
        work on Birch (Multinet).  I think that it *does* work on
        the other VMS nodes (UCX), but wouldn't bet on it.


Communication between ipcs cells:
--------------------------------

A cell is defined in the optional "this_host" line in config file:
        /cell_netmailer=nstx_pcs
in the config file for one host only

A cell acts as a single entity, through the cell_netmailer, when talking
to other cells.  Only the cell_netmailer node knows about other
cells to which it might communicate:

        ga_d3d <--> pppl_res
                    pppl_res <--> nstx_pcs

If EVTMGR_NSTX runs on a node in the pppl_res cell, then it can be
reached by nodes in both the ga_d3d and nstx_pcs cells.
However, if EVTMGR_NSTX runs on nstx_pcs, then it can be reached only
from pppl_res.


Example config files:
--------------------
Unix:  nm_config.dat
   epcisrv1 (cell_netmailer for "nstx_pcs"):
      this_host=epicsrv1 /cell_netmailer=nstx_pcs /protected
      nstx_pcs /CONNECTION=WOLLONGONG_TCP /INTERNET=198.35.10.21 /PORT=22375
      ppcc /CONNECTION=WOLLONGONG_TCP /INTERNET=198.35.10.102 /PORT=22375
      pppl_res  /connection=tcp /internet=192.55.106.24 /port=1000

   ppcc (member of nstx_pcs cell)
      this_host=*/protected
      nstx_pcs /CONNECTION=TCP /INTERNET=198.35.10.21 /PORT=22375
      ppcc /CONNECTION=TCP /INTERNET=198.35.10.102 /PORT=22375


VMS:  netmailer.cfg
   birch (cell_netmailer for "pppl_res")
      this_host=* /cell_netmailer=pppl_res /protected
      RAX    /connection=tcp /internet=192.55.106.12 /port=1000
      BEANIE /connection=tcp /internet=192.55.106.22 /port=1000
      BIRCH  /connection=tcp /internet=192.55.106.24 /port=1000
      ga_d3d /CONNECTION=TCP /INTERNET=192.73.62.46 /PORT=22375
      ga_usc /CONNECTION=TCP /INTERNET=192.5.166.24 /PORT=22375
      llnl_res /CONNECTION=TCP /INTERNET=128.115.15.25 /PORT=22375
      nstx_pcs /CONNECTION=TCP /INTERNET=198.35.10.21 /PORT=22375

   beanie, rax (member of "pppl_res" cell)
      this_host=* /protected
      RAX    /connection=tcp /internet=192.55.106.12 /port=1000
      BEANIE /connection=tcp /internet=192.55.106.22 /port=1000
      pppl_res  /connection=tcp /internet=192.55.106.24 /port=1000


An explanatory (?) e-mail:
-------------------------

From:   BEANIE::GIBNEY
To:     PSICHTA@PPPL.GOV,RONEY,BDAVIS
CC:     GIBNEY
Subj:   ipcs

Paul, Phyllis and Bill,

Oh happy day -- I think that I now have ipcs communications sorted out.
At this point, you should be able to communicate between the "nstx_pcs"
cell (epicsrv1, ppcc) and the "pppl_res" cell (birch, beanie, rax).

The "pppl_res" cell also knows about cells at GA and LLNL:

        nstx_pcs <==> pppl_res
                      pppl_res <==> ga_d3d
                      pppl_res <==> llnl_res
                           \
                            ipcs (EVT) messages from pppl_res can be seen
                             on both sides.

EVTMGR_NSTX should run within the nstx_pcs cell.  At the moment, it
is running on epicsrv1.  This way, events can be shared between the
nstx_pcs and pppl_res cells, but can not be seen offsite at
GA or LLNL.  If EVTMGR_NSTX was run on, say, Birch, then events
would also be visible at these remote sites:

Tom