USING THE LINK FOR ONLINE MONITORING - MJL 9/6/96 ================================================= Looking at histograms, etc. --------------------------- 1) attach with paw and look at histograms paw> global_sect ANLY (attaches to histograms in shared memory) paw> greset (if you want to clear histograms) paw> exec dchits (or pthits, or hodhits, etc - see *.kumac files) 2) use finder (ask Beddo) Problems, first the link must be up. To get the link up: --------------------------------------------- 1) Check if it is already up: linkStats (at the VME console) ps -ef | grep ltodd (on fn866a) 2) If no linkServer or no ltodd: clean unix cd dd (on fn866a in e866 account) source setup (sets up dd environment) clean (repeatedly until it shows no actions) if linkServer in VME was still running it will now exit after it sees a few packets (wait until it does) 3) To start a link connection stop run on PDP and on VME clean unix (as above) dd_start (in dd) make sure ltodd up (ps -ef ...) linkSetPs 5 (set link prescale factor on VME) (number of events per spill / prescale ~= 4 is about right; this should give rougly 400 events across link per spill) 3) start run on VME and PDP only after link and analysis are started and ready to go Then start an analysis (run must be stopped on VME & PDP) --------------------------------------------------------- 1) cd analydd use looper as described below; using run directly requiers a different dimu version for separate analysis runs for each run nohup nice looper >& looper.log & (etmbor switch in cmzlogon must be on) (looper starts a new analysis job for each VME run and puts all the output files in files with run number in name, e.g. runxxx.log, runxxx.his, runxxx.nt) for one continuos run: nohup nice run >& run.log & (etmbor switch in cmzlogon must be off) 2) make sure analysis running ok tail -f run.log 3) If analysis goes astray kill dimu and run (and looper) restart as above (while run stopped on VME & PDP)