How to start DAQ
First of all
Now we have an updated DAQ using two VME crates with processors named c-eposly22
(for DC and FERA branches) and diracvme2 (for MSGCs). Both of them must
be switched on and booted. All DAQ processes start at boot time. That are
SC processes and read_master
processes:
c-eposly22 : sc_svcvme
and at least one read_master
diracvme2 : sc_svcvme2
and at least one read_master2
To get PS beam intensity you must start correspoding processes,
see PS data.
You must be logged in as a run user on main DAQ host computer
(pcdirac06) and set DAQ environments. You may do the last by commands:
cd ~/bin
source csh.daqvar
Two processes (primary receiver ddt-receive and central slow control
sc_central)
are commonly running on pcdirac06.
| To check that they are |
To start them |
ps ax | grep ddt ;
ps ax | grep sc_ |
receive &
sc start |
top
DAQ processes on VME hosts
Your actions depends on the prehistory of DAQ work. As it was mentioned
above, all DAQ processes starts at boot time.
If by some reason they are absent you can start them by hand:
Slow control :
sc start [restart]
Read-out process :
c-eposly22 : read_master
&
diracvme2: read_master2
&
If c-eposly22 was rebooted or CAMAC crate[s] were switched of,
it is necessary to perform
cd
~/bin
source
csh.daqvar
init_crate
0
and start process to get PS data .
Comments.
sc - script to work with slow control process. The argument
may be: start, restart, stop.
init_crate 0 - perform initialization (Z, C, Clear
I) of CAMAC crates in correspondence with a /home/run/param/RO/std.cfg
file.
init_crate Br Cr -
perform initialization of the CAMAC crate number Cr in the branch
number Br.
read_master - processes which performes data read-out and data
transfer to the main DAQ host.
top
DAQ processes on main DAQ
host
As you started DAQ processes on VME host, you can start processes on main
host which provide an event building and the control of run:
builder &
RC &
RC_gui &
RD &
automon >&/dev/null &
and display counting rates:
SM_dt
&
SM &
Comments.
builder - a process which starts childs to get data from
VME processors, build events and transfer data to main DAQ host.
RC - Run Control process. It provides control of run state by
a retransmission of commands from RC_gui (Graphical User Interface)
or slow control processes to read_masters. RC may be
killed and started again if read_masters is running.
RC_gui - graphical user interface to control run. It allows you
to control read-out electronics, select trigger file, change a run state.
RC_gui
may be killed and restarted again in any time without any influence on
the data taking work.
RD - Run Display, may be started in any time on main DAQ host
or on any computer receiving data on-line, but in the last case it will
not show size of data file.
automon - primitive automatic monitoring.
It writes message in file pcdirac02:run/log/monitoring.log in few cases:
run changed a state, a CAMAC bus or MSGC HV alarms or warnings were detected,
few empty bursts one by one were got. If you wish to see and hear these
messages (that is strongly recommended) you must start readinfo
process on pcdirac02.
top
Monitoring processes
pcdirac02
You must be logged in as user run on pcdirac02 and set DAQ
environments (see First of all).
-
To see and hear messages from automon running
on pcdirac06 you must type command:
readinfo start -
it
is a script which starts process looking in file ~run/log/monitoring.log
and displays new messages appeared in this file on pcdirac02 console. Messages
related to a change of run state are not displayed, only pronounced.
-
omon - common monitoring process
to show histograms
-
omonp - monitoring process dedicated to work with pedestals
of ADCs.
top
PS data
There are few processes running on VME host which get data from PS accelerator
and display them on a screen. One of processes writes a file which is used
in off-line and on-line data handling. Commonly these processes are started
on PS computer(). For that you must open connection from PS computer to
VME host by:
telnet c-eposly22
and logged in as user pscontrol. Than type command:
counts &
top
Central recording
Please, do not forget to start central recording process. In accordance
to O.Gorchakov :
-
You must be logged in on pcdirac06 as user DIRACXFR (password
will not appear here) and do
recorder &
-
You must be logged in on pcdirac06 as user run and do
run_move &
Please, look time to time how many free space is on /data disk!
What to do
In case of problems the DAQ writes a corresponding message in the VME or
main DAQ host terminal windows and in logfiles. These logfiles are in the
directory: pcdirac06:/data/log . One error detected by DAQ
may results in few messages. The most detailed messages are in pcdirac06:/data/log/daq.debug
file.
builder,
RC and RC_gui started, but run cannot be started
The most often reason is that an electronics module is not initialized
properly and you will see a corresponding message. Please, look on VME
host terminal window or/and in the pcdirac06:/data/log/daq.debug logfile.
Some messages appears only in logfiles! So, before next
attempts to start a run, it is necessary to look at
pcdirac06:/data/log/daq.debug
logfile for detailed information and if you cannot solve the problem, please,
call experts (trigger or DAQ ones, depending on trigger or read-out electronics
is not loaded properly).
top
RC died
It may happens in case of network problems or detected errors in electronics
operations. Look on the error message on the screen or/and in the pcdirac06:/data/log/daq.debug
logfile.
Possible solutions:
Try to restart RC, by the command RC & on
main DAQ host.
Check that builder
and read_master are running.
If it is not running, start it and then start RC.
If it did not help, than
-
Check that primary receiver and central
slow control processes are running on main DAQ host (pcdirac06).
-
Check that sc_scvme is running
on VME host.
If any of processes is absent, try to restart them (at first on the main
DAQ host - receiver and slow control, then slow control on VME hosts),
for that:
-
Kill builder (how to kill a process see Kill DAQ).
-
Kill RC (if it is running) on main DAQ host,
-
and start them again, see First of all
Extremely rare, you need to kill the primary receiver and central
slow control. Please, do not do that without good motivation, and
especially when builder and RC are running.
top
RC_gui does not
react on your action.
Possible reasons:
-
There is an opened window of lower level than the one in which you would
like to change something.
For inactive windows the background color is light gray, it may help
you to find out an active one.
-
You are trying to do something that is not allowed at current situation.
Look more carefully or ask experts.
-
RC_gui waits until previous command (issued by you or a slow control function)
will be completed.
Be patient and do not repeat commands too often.
-
RC_gui lost connection with the run control process and hanged up.
If you are absolutely sure that RC_gui hanged up you can
kill it (see Kill DAQ) and start it again by RC_gui
& command.
Please, remember that initialization of
electronics at start of run requires few
minutes.
top
Kill DAQ
You can use possibilities:
-
Kill only the graphical user interface RC_gui by button
Kill
GUI or
-
kill also RC and builder processes (by button Kill
DAQ) or
-
kill processes by standard UNIX commands.
It is the only way to kill builder when RC died
already. To do that:
-
you should know a process ID (pid)
For that you can use a command: ps ax | grep
process_name. For example, ps ax | grep
builder. You will see something like:
137 37 137 17 248 136 11800 57:05:22 ttyp1 run W /home/run/bin/builder
The most left number is pid of builder.
and then kill the process by command:
-
killall builder
-
kill pid
for any other process.
top
Please, send your comments to Serguei.Trousov@cern.ch