Monitor data during the run

There are two types of processes for online monitoring:

1. The master process that fills histograms for experts, shift and has access to the event displays.

All functions are accessible through the master GUI. To start it do the following:

login as user 'shift' with the usual password on one of the online machines
type phat
type: g = new TPhOnGui(400,400)
hit the start button and enjoy the menu items & buttons

2. Background control processes (online clients).

These processes perform automatic checks on all sub-detector parts. There is about one process per sub-detector. To start them all at once do the following:

login as user shift with the usual password on phonl3 or (if you're already logged in) type cd
type ./online
hit all the start buttons of the windows that will appear

What is checked in the background processes:

1. Latchup client:

OnlLatchup1:

The latchup client checks whether the rms of the adc values within one chips is below a threshold (currently 10) or whether the VFP voltage is above -50 mV. If one or both conditions are met it will fire the alarm and print in the error window, where and when the problem occurred. The latchup client should run all the time and all alarms should be taken serious.

In order to avoid false alarms the latchup client doesn't check the rms during SICAL runs. It will show the status "okay"/"running" when the silicon is disabled and no silicon data are present.

org.author: Judith Katzy

responisble: TBA

OnlLatchup2:

This latchup client checks the data base if ports have a too low VFP voltage or a wrong current. The rocdb writes these data to the database in variable time intervals. Currently (August 01), the values are written every 30 sec at the start up of Silicon and every 5 minutes all the other times. This client checks the data base every 3 minutes. The time since the last check is displayed in the status window. If a check fails an error message will appear and the master alarmer will show an error as well. This client is basically meant to protect the silicon system even if something in the data stream from the daq to the event distributor to the online client goes wrong. Due to its high latency, it is not meant as a primary latchup client.

org.authors: Rachid Nouicer and Judith Katzy

responsible: Rachid Nouicer

2. Trigger Monitor:

The trigger monitor checks whether each of the paddle, zdc and cerenkov counters gets adc signal above pedestals together with the tdc signals. It does this by filling histograms with the adc values for all hits with tdcs. After a certain number of events(currently 200 for ALLBEAM runs, 400 for cosmics) the histograms are checked and an alarm occurs in case of failure. The histogram that caused the alarm will be displayed and the channels will be printed in the status window. The number of events since the last check are displayed in the status window.

org.author: Dave Hofman

responsible: Dave Hofman

3. Heartbeat:

The heartbeat is a very simple client to check whether the event distributor is still sending readable events. All it does is to read the events and display the event number. If some of the other background clients crash and the heart beat as well, you know that the events are very likely to be corrupted.

org.author: Judith Katzy

responsible: Burt Holzman

4. TOF Monitor

org.author: Frank Wolfs

responsible: Frank Wolfs

5. ROCDB client

The rocdb macro is part of the DAQ and should be running at all times. It does write scalers to the database every 5 minutes. This client checks in the database whether the scalers have been updated. The status window shows the time since the last db check. If there are no new values after an (adjustable) time limit (currently 11 minutes), an alarm will fire on the online machine and on the central alarmer.

org.author: Judith Katzy

responsible: Andrei Sukhanov

What to do in case of problems with online clients

1. If one of the windows seems to be frozen. just click the online button on the taskbar (for some reason it is a quadrant of somebody's face, centered on their eye).

2. If the computer seems to be badly broken -- the whole screen is frozen, not responding to keystrokes, etc, follow these steps to recover (phonl3 only; do not reboot phonl1 without contacting Burt):

Ctrl-Alt-Backspace (should kill the X server). Should let you log back in and click the button to restart online. If this doesn't work:
Ctrl-Alt-F1 should get you to a text window; then Ctrl-Alt-Delete to reboot. If nothing happens:
Ctrl-Alt-F1, Alt-SysRq-S, Alt-SysRq-U, Alt-SysRq-B (SysRq is the key to the right of F12). If nothing happens:
Big reset button on the front of the case. Last resort since linux will take a few minutes to fsck the disks.

Monitor data from file

get phat - if you're in the counting house you may use the phonl1, see above.
cd $PHATHOME/online/macros
phat online_setup.C this will create a file online_sm.root with the module initialization in the current directory. this step has to be done only once.
online.C this will start the gui.
go to the 'input' item in the menu bar and select 'file'
go to the 'file' item in the menu bar and select 'open'. a file selector box will appear.
hit the start run or the next button.