The Production and Ntupling daemons

How to submit a job to the ntupling daemon

1. You need to create a definition file for each ntupling job you wish to run. The file will be executed as a bash shell script; this means you can set environment variables to be passed to your tcl file, run commands, etc. The file will be executed under the peonyisi username, so please be careful with what commands you use (in particular be careful about assuming what directory you're in). It should look something like this:

   USERNAME=peonyisi
   STNCODEVER=4.0.0
   STNTCL=/cdf/data2e/stream_daemons/stnfilter_inclusive.tcl
   LOGFILE=/cdf/data2h/ntupled_318int3/log_3.18.0int1_125994_1.txt
   SPLITMODE=1
   
   STN_INPUT=/cdf/data9c/produced_stream_a/prod_3.18.0int1_125994_1.root
   STN_OUTPUT=/cdf/data2h/ntupled_318int3/ntupled_3.18.0int7_125994_1.root
   export STN_INPUT STN_OUTPUT
   

   The only required fields are USERNAME and STNTCL. A listing of commands that have meaning to the daemon follows:
   • USERNAME: if email necessary, will email to this username
   • STNTCL: sets tcl file that is passed to stnmaker
   • STNCODEVER: sets version of code to use (if not set, defaults to whatever version is "current")
   • LOGFILE: sets log file name is to be placed. If not set, output of stnmaker will be emailed to USERNAME.
   • SPLITMODE: if set to anything non-empty, will cause files to be written in split mode. Non-split mode is default.
   STNINPUT and STNOUTPUT are environment variables that I set for the benefit of my tcl file, and so need to be exported; they're an example of what else you can do in these files.
2. Place the description file in the ntuple spool directory: /cdf/data2e/stream_daemons/nt_spool. Please give it group-write permissions, as the spooler needs to delete it when done.
3. Wait.
4. Unfortunately, the output will be under my username; at least it will be group-writable, so you can do as you wish with the files.

How to use the daemon to make ntuples from a list of post-production files:

An example shell script, using the ntupling daemon, follows. Cut and paste into a .sh file, adapt as desired [though if you want to remake the ntuples in /cdf/data2h/ntupled_318int3, please change only STNCODEVER], and call with absolute path names of the production files you want to ntuple.

#!/bin/bash
 
EXECDIRECT=/cdf/data2e/stream_daemons
STNDIR=/cdf/data2h/ntupled_318int3
STNCODEVER=4.0.0
 
for i in $@ ; do {
SPOOLFILE=$EXECDIRECT/nt_spool/`basename $i`.dsc
BASENAME=`basename $i`

RUNNUM=`echo $BASENAME | awk -F _ '{print $3}'`
FILENUM=`echo $BASENAME | awk -F _ '{print $4}' | awk -F . '{print $1}'`
 
ERROR=log_${STNCODEVER}_${RUNNUM}_${FILENUM}.txt
 
echo "USERNAME=$USER" > $SPOOLFILE
echo "STNCODEVER=$STNCODEVER" >> $SPOOLFILE
echo "STNTCL=$EXECDIRECT/stnfilter_inclusive.tcl" >> $SPOOLFILE
echo "LOGFILE=$STNDIR/error_logs/$ERROR" >> $SPOOLFILE
echo "STN_INPUT=$i" >> $SPOOLFILE
echo "STN_OUTPUT=$STNDIR/ntupled_${STNCODEVER}_${RUNNUM}_${FILENUM}.root" >> $SPOOLFILE
echo "export STN_INPUT STN_OUTPUT" >> $SPOOLFILE
 
} ; done