Startup and Shutdown of the S117 DAQ System

The following describes

Boot and login
SBS Startup
SBS Shutdown

See also notes on ESONE Server, tape handling and troubleshooting.

Boot and Login

Please look in the data acquisition layout which processors are currently working. Do all the commands in the following order !

Power up and reset all systems

  CAMAC  : Switch on crate and reset CVC's
  VME    : Switch on crate and reset CPU
  Fastbus: Switch on and start crate. Reset all AEB's

Connect to all the systems

  Command on VMS	  :  @inc_s117:daq_rterm_create

Boot all systems

  CAMAC  : Boot the CVC's with the following command:
               GO 20                         (expect)

  VME    : On the VSB-master (E7) give the following commands:
               RE    <- read setup from NVRAM
               GM    <- go to ULTRA monitor
               BO    <- Boot                 (expect)

  FASTBUS: Just wait 2 minutes

Login on the Lynx systems (CVC, E7).
There are 3 LAT windows for the DAQ control on E7 and CVC's.
In addition, there are 2 TELNET windows on the E7 for monitoring.
```
  username 'aladin' password 'mudex0' 
```
Change to the desired directory (test)
```
  Command on CVC and E7   :  cd test
```
Block process ID 2
Processes with a pid of 2 don't properly in Lynx. To block this pid with a dummy task m_pid2 enter:
```
  Command on CVC and E7   :  prod
                             lock2             takes some time ...
```

Start SBS

Start the slave-readout routines (on CVCs)

  Command on all CVC      :  sbs
  on sbs prompt in crate1 :  @slave1start      (expect)
  wait for completion!
  on sbs prompt in crate2 :  @slave2start      (expect)

From the VSB-master start the AEB slave-readout processes

  Command                 : aeb_start          (expect)

Start the collector and transport processes on the VSB-master

  Command                 :  sbs
  on SBS prompt           :  @masterstart      (expect)

The acquisition is started on stopped on the master-trigger processor

  start acquisition       | starts the acquisition
  stop acquisition        | stops the acquisition

Start SBS monitor and Rate display

  sbs_mon -p1 -t10        | starts sbs monitor (execute in SBS-Monitor window)
  rate -t1                | shows data rate    (execute in RATEMETER window)

All other commands (including tape handling) are only valid on the event builder processor (EB window).
Some important commands:

  show rate               | Number of events already processed
  show rate nnn           | Number of events averaged over nnn seconds

Notes:

In some cases you might see the message
```
Suspended (tty input)
```
In this case one process got started with pid 2 and mishandled by Lynx. Check with ps -a. If you see a SBS process with pid 2 you had bad luck. Kill all SBS processes on this node and try again...

Shutdown

Depending on whether the DAQ system is still reponding or not one can use more or less harsh methods to shutdown the SBS:

Stop the acquisition on the master-trigger processor :
```
  Command                 :  stop acquisition
```
If SBS was working correctly up tp now, try orderly shutdown:
- Stop the processes on the master-VSB processor :
```
  Command                 :  @masterstop
                             exit
```
- Stop the processes on the slave readout processors (CVC) :
```
  Command in Crate 1      :  @slave1stop
  Command in Crate 2      :  @slave2stop
  Command on any CVC      :  exit
```
- Stop the processes on the AEB's :
  The once advocated command aeb_kill is unreliable and should not longer be used. Instead:
  - Type ^E in all four AEB windows to abort readout processes
  - Check with aeb_ps whether they are all gone
  - Finish off the renegades with
```
    rsh nodename +"kill pid; procs"
```

If SBS was hung or components did abort, try it the rude way:

Type ^Z at the sbs prompt. Should abort all tasks.
Check with ps -a whether there are m_.... processes left. If yes, get rid of them with kill except for the pid 2 blocker m_pid2.
Abort AEB readout processes as described under orderly shutdown.
To reset message queues and the status segment on CVC's and E7:
```
    msg_new
    clear_status
```

Check on the event builder for hanging TCP connections with netstat.

Active Internet connections
Proto Recv-Q Send-Q  Local Address          Foreign Address        (state)
tcp        0      0  E7ELXA.gsi.de.6001     VSBZ.gsi.de.2667       FIN_WAIT_2
tcp        0      0  E7ELXA.gsi.de.telnet   AXP625.gsi.de.1907     ESTABLISHED
tcp        0      0  E7ELXA.gsi.de.telnet   AXP625.gsi.de.1906     ESTABLISHED

tcp        0      0  E7ELXA.gsi.de.6001     VSBZ.gsi.de.2667       TIME_WAIT

After this procedure all systems should be free of DAQ processes (check with ps -a and aeb_ps) and ready for a SBS startup.

S117 DAQ Home Page

Last updated: April 20th, 1995
Volker Serfling , Walter F.J. Müller