gibney$:[handbook]IPCS_NSTX_RESTART.MEM
                        Evtmgr and IPCS
                        ---------------
Event Manager (evtmgr) on epicsrv:
---------------------------------

  Evtmgr can be restarted if you think there's a problem with it:

  Use epics control screen CH01, or :
        cd $PPPL_RES/bin
        su cicopr             #if not already there
        kill {evtmgr pid}
        ./evtmgr &


IPCSMAILER: (unix and vms nodes)
----------
  Each node has an IPCSMAILER program which handles ipcs messages
  between different computers (nodes).  These are low-level connections
  upon which the ipcs messaging environment is based.

  Several cooperating nodes can be grouped in a "cell".  A single node
  in the cell is designated the "cell netmailer".  This is the
  only node which communicates with nodes or cells outside
  the local cell.

  IPCSMAILER should *not* be restarted at whim because connections
  between active processes will be lost.

  Status of the IPCSMAILER on the local node can be checked by
  nm_status program:
    KEES$ nm_status
               pppl_res                KEES PROTECTED CELL NETMAILER (KEES)
               pppl_res            pppl_res     0 TCP  NULL
                  BIRCH               BIRCH     6 TCP  CONN
                 EUROPA              EUROPA     4 TCP  CONN
                 taurus              taurus   267 TCP  CONN
                epicsrv             epicsrv    10 TCP  CONN CELL ()

             IPCSMAILER   pppl_res            21a004e5 3f798fca    REG
            EVTMGR_TFTR     BEANIE            3320155f 3f572873    REG
               TFTR_QCS     BEANIE            33201561 3f572874    REG
              ACQ_PARMS   pppl_res            21a00433 3f79865c    REG
              ACQ_EPICS   pppl_res            21a00432 3f798663    REG
         ACQ_CREATETREE   pppl_res            21a00444 3f798665    REG
                RUNNSTX   pppl_res            21a021ad 3f7aacfb    REG
            EVTMGR_NSTX    epicsrv    epicsrv 275c0000 9c277c3f   pppl_res  epicsrv

  o If you must restart IPCSMAILER, be sure to restart
    related processes as mentioned below.
  o If IPCSMAILER is restarted on epicsrv, then *ALL* evt-driven
    processes must be restarted (both unix and vms) because connection
    with evtmgr will have been lost ...

  Unix:
  ----

    To restart ipcs netmailer on epicsrv (for example)
     - Note: shouldn't need to do this too often!  See above.

        cd $LLNL_RES/etc        #typically /opt/ipcs2/etc
        su cicopr
        istop IPCSMAILER
        istart IPCSMAILER		#ask grandpa to start mailer

    Processes to restart:
        - evtmgr (see above)


  VMS:
  ---
    To restart ipcs mailer on VMS (e.g., on Kees)

        $ @pppl_res:[evt]defs
        $ istop "IPCSMAILER"        !note: the quotes are essential
         -- wait a bit (maybe 60 sec?) --
         -- istart will fail if you try too soon (no problem) --
        $ istart "IPCSMAILER"

    VMS processes to stop/restart:
    o KEES:
        runnstx:         @nstx$:[nstx.source.runnstx]START_RUNNSTX.COM
        clocksync_nstx:  @pppl_res:[shotclock]START_CLOCKSYNC_NSTX.COM
        nstx_redeclare:  @nstx$:[nstx.source.events]START_NSTX_REDECLARE.COM
        clock/* : "shotclock" program for individual users

    o other VMS nodes:
        clock/* : "shotclock" program for individual users

    Further notes on KEES processes:
      o runnstx:  Watches for "cycle" events, issues TCL commands
          to initialize camac, acquire data, etc.
      o clocksync_nstx:  receives "shot cycle" messages from epics,
          declares both EVT and MDSplus events for NSTX_SOC,
          NSTX_SOS, NSTX_TMINUS60, NSTX_SOD, etc.
      o nstx_redeclare:  redeclares selected MDSplus events as EVT events.