How to change the calibration constants used by recon?
(by Dave Smith, for the CAL case. ) There are 5 types of CAL calib.xml files, called
- asym = light asymmnetry
- ci_intnonlin = integral non linearity
- peds = pedestals
- mpd = MeV per DAC
- thold. = thresholds
These files are in $LATCalibRoot/CAL/CU06.
If you have new .xml files then the first thing to do is to copy them to that directory.
Don't have write privileges there? Then contact me or Johan Bregeon or Anders. In a hurry and none of them are available? Then try Johan Cohen-Tanugi, Joanne Bogaert, Richard Dubois.
Your next step is to enter the new files names into the RDB metadata database. If you're not in a flaming hurry, it's best to have one of us "experts" do it.
But if we're unavailable, then log on to noric (or whatever machine your rdb MySQL system is) and type rdbGUI-new.
There is documentation, see
http://www.slac.stanford.edu/exp/glast/ground/software/notes/rdbGui/rdbGui-use.shtml
Older but still of some merit is the following.
http://www.slac.stanford.edu/exp/glast/ground/software/calibration/docs/mysqlDirect.shtml;
Once you've mastered rdbGUI and you've added your file names to the metadata database, triple check your entries. It is very important that vstart be consistent with the data you want (re-)processed. If your calibrations are to supersede older calibrations, that is, render the old ones obsolete, then YOU MUST CHANGE THE PROC_LEVEL FROM PROD TO SUPSED for the old file names.
If you have spaces, typos, invisible characters in your file names, it will CRASH THE PIPELINE so before you update the rdb database, send e-mail to francesco.longo@ts.infn.it (and copies to focke@slac.stanford.edu borgland@slac.stanford.edu). They may ask you to wait for a less critical time(it's the same pipeline for TVAC, you know). Remember, you can use PROC_LEVEL DEV or TEST to run your own reconstruction if you don't want to wait for the pipeline.
Okay, it won't actually crash the whole pipeline, just those jobs that look for non-existing calibration files. Coordinating with the pipeline caretakers before you make a change will help them be alert to why many jobs are suddenly failing...
How to find out which calibration constants were used ?
Example : CAL calib constants
Here's one way (by Dave Smith). From the eLog, click on "list root files" to find the name of the directory on noric where the recon files are. In that directory, you can do a grep. For example,
dsmith@noric06 $ cd /nfs/farm/g/glast/u37/BeamTest06/rootData/700000698/v4r0909p2/recon
dsmith@noric06 $ grep asym *.log
recon-v1r030603p4_700000698_00000-07469.log:XmlBaseCnv INFO successfully parsed document $(LATCalibRoot)/CAL/CU06/mc_asym.jun8th2006.CU_pass1_Pisa.xml aka /afs/slac.stanford.edu/g/glast/ground/releases/calibrations//CAL/CU06/mc_asym.jun8th2006.CU_pass1_Pisa.xml
recon-v1r030603p4_700000698_07470-14939.log:XmlBaseCnv INFO successfully parsed document $(LATCalibRoot)/CAL/CU06/mc_asym.jun8th2006.CU_pass1_Pisa.xml aka /afs/slac.stanford.edu/g/glast/ground/releases/calibrations//CAL/CU06/mc_asym.jun8th2006.CU_pass1_Pisa.xml
recon-v1r030603p4_700000698_14940-22409.log:XmlBaseCnv INFO successfully parsed document $(LATCalibRoot)/CAL/CU06/mc_asym.jun8th2006.CU_pass1_Pisa.xml aka /afs/slac.stanford.edu/g/glast/ground/releases/calibrations//CAL/CU06/mc_asym.jun8th2006.CU_pass1_Pisa.xml
dsmith@noric06 $
You can do the same for the other 4 types of CAL calib.xml files that are used in recon, e.g. ci_intnonlin, peds, mpd, thold.
CAL calibrations
Trending CAL CPTs
Here is the trendiing of CU CAL module performances from CPTs runs taken at Pisa (2 suites) and at CERN (July, 7) : T01-FM109, T02-FM119and T03-FM101 : no evolution was found
ACD calibrations
CU ACD have basically two calibration parameters for each channel (PMT), 6 in total - MIP peak position and pedestal (both - in ADC bins). Calibration parameters obtained in the 5 GeV pions runs on July 29, are given in the attachment