HPS DAQ Quick Reference

From hps run
Revision as of 16:01, 27 June 2014 by Baltzell (talk | contribs)
Jump to navigation Jump to search

Note, this is copied from here to allow those without jlab account.


DAQ Recovery Procedure

In most cases DAQ will crash because of ROC problem. Do following:

  • click 'End Run' or 'Abort', whatever button is active
  • click 'Cancel'
  • run 'roc_status' command from clon10 window; it will show troubled ROC(s) as 'UNDEFINED' or 'active'; most likely clastrig2 or/and polar
  • click 'Reset'
  • from clon10 window, reboot troubled ROC(s) using command 'roc_reboot <roc>'; if rebooting any ROC, reboot clastrig2 as well
  • wait 2 minutes and run 'roc_status' again; rebooted ROC(s) must be in 'booted' stage
  • go to 'EB_TERM' window, Ctrl-C and run previous command using arrow up (command is 'coda_eb ...')
  • start new run from 'Download'

NOTE: 'tage' and 'tage2' are Linux processes running on machines 'tage' and 'tage2' respectively. Instead of roc_reboot, goto appropriate window (or ssh to those machines if window is not opened), Ctrl-C and type command:

TAGE: coda_roc -s clasprod -o "tage ROC" -i
TAGE2: coda_roc -s clasprod -o "tage2 ROC" -i

If it does not help, page DAQ pager. If no reponce during 10 minutes, call Sergey (office 5795, for personal numbers see link at top).



FOLLOWING INFORMATION IS FOR DAQ EXPERTS ONLY (DO NOT DO IT IF YOU ARE NOT SURE !!!)

DAQ procedures

First make sure you can open X window from the terminal clon03:0.1: type 'xterm' and make sure xterminal is opened. If it is not, message 'X11 connection rejected because of wrong authentication.' will shows up. In this case kill xwindow, start new one (you will be on clon03) and 'ssh clon10'. Make sure 'xterm' command can open xterminal.

To start all DAQ processes goto clon10 as clasrun and type following command:

daq_start this

to open all windows on current workspace. If daq_start executed without parameter, it will open everything on clon03's bottom screen on currently active workspace.

To kill all DAQ processes goto clon10 as clasrun and type

daq_exit

To start/stop serial connections to ROCs and PMCs, type following on clon00:

roc_xterm_this
pmc_xterms_this
roc_xterms_kill
pmc_xterms_kill


Online procedures

To start online monitoring processes use following commands:

mon0_start (starts everything on clonmon0 - fast online reconstruction;
     NOTE: root presenter must be started separately: ssh to clonmon0 (usually from clon03:0.2) and type 'cmroot clasprod hist_clasprod')
mon1_start (starts everything on clonmon1 - event_monitor)
mon2_start (starts everything on clonmon2 - online recsis, ced; if ced does not started, run it manually by ssh to clonmon2 from clon04:0.2)


To kill online monitoring processes use following commands:

mon0_exit (kills everything on clonmon0)
mon1_exit (kills everything on clonmon1)
mon2_exit (kills everything on clonmon2)

Occupancy plots must be started on clonmon1. Login as clasrun and type

monb

Stadis to be started on clon05:0.1.

DCHV to be started on clon06:0.1.

Diman to be started on clon03:0.0.


Background information

  • CRON JOBS: on clon10, clon00, clon01 and clondb1 make sure cron jobs are running (they must start automatically after reboot); if have to start cron jobs, login as clascron to the mentioned machines, goto directory /home/clascron/backup and follow instructions in README file, normally it will be following:
login as 'clascron' to the clon10
goto /home/clascron/backup
type 'crontab -l' to see what is running
type 'crontab -r' to erase all existing cron jobs
type 'crontab clon10'
repeat everything for clon00 and clon01, using file 'clon00' and 'clon01' respectively

Process management

Sometimes you may want to turn some cron jobs off to avoid alarms during system reconfiguration: do not forget to turn them on again after you are done !

  • Process management: SmartSockets-based process management procedures must be started on clon10 as clasrun using following command:
control_process_manager start clasprod clasprod
control_procmgr_manager start clasprod clasprod
control_ipc_process_manager start clasprod clasprod

and killed by following:

control_process_manager stop clasprod clasprod
control_procmgr_manager stop clasprod clasprod
control_ipc_process_manager stop clasprod clasprod


Services on clondb1

Clondb1 is running several important servers which suppose to be started automatically on reboot:

SmartSockets
Mysql

Logbook and/or run database does not work

Call DAQ pager..

Useful commands

r_ts_prescale - reads prescales from TS2
s_ts_prescale 0 0 0 0 0 0 100 0 - sets prescales to TS2; 0 means prescale 1, 1 means prescale 2 etc
proconhost, netonhost - run everything on main VME CPU (execute from hosts's serial connection before Download)
proconpmc, netonpmc - run everything on co-processor (execute from hosts's serial connection before Download)


Sergey Boyarinov (from JLAB local phone):

office 5795 
For personal cell/home/skype contact info, see link at top of this page.