Instructions for making a tarball to be used on the farm to run simulation on 
HPEG files:

I start from this webpage:

http://www-cdf.fnal.gov/internal/mcProduction/howtoRequestMcv4_5_3.html

This is Frank's prescription for how to generate MC events with Pythia and run 
them through simulation, all on the farm. What I've done is to take this, 
remove the Pythia stuff, change the input in DHInput from the output of Pythia 
to a file Miro gets from the DB, and change the output in DHOutput to write to 
the database...

Here's how:

First, start in your scratch on fcdflnx2 or fcdflnx3 (I use fcdflnx3). 
Following the instructions from the webpage above, type 

      source ~cdfsoft/cdf2.cshrc
      setup cdfsoft2 4.5.3
      newrel -t 4.5.3 myRelease
      cd myRelease
      cvs co -r mcv4_5_3 mcProduction
      cd ..
      myRelease/mcProduction/scripts/compileExeTarball 4.5.3 >& exe4.5.3.log &
      
This takes a bit...

This creates two things, a directory CDF4.5.3 and a tarball of the directory in
 your scratch area (i.e. in the directory above myRelease). We're going to 
ignore the tarball, edit stuff in the directory and make a new tarball by 
hand...

So, in the directory CDF4.5.3 type:

      mv runMC4.5.3 runMC

This is because we want to create a wrapper script called runMC4.5.3 which is 
what Miro will run from the farm. This script will do two things. If given only
 one input, it will run a monte carlo job generating events from Pythia through
 simulation. This will be done through a .tcl file which we don't concern 
ourselves with here. If given two inputs, it will simulate the second input 
(the HEPG file Miro gives it) with the first input (which is the appropriate 
.tcl file which we will write shortly. 

Here is the script I wrote:

fcdflnx3:~/scratch/CDF4.5.3> more runMC4.5.3
#!/bin/tcsh -f 
 
if ( $2 == "" ) then  
  source source_me   
  ./runMC $1 >& $1.log  
else  
  setenv INPUT_FILE $2 
  setenv SECNUM `echo $INPUT_FILE | awk -F . '{print $2}' | awk -F g '{print $1}' | hextodec.exe` 
  @ GRNDM3 =( {154710} + (${SECNUM} * {28293}) ) 
  @ GRNDM4 =( {78951} + (${SECNUM} * {6482}) ) 
  setenv GRNDM1 $GRNDM3 
  setenv GRNDM2 $GRNDM4 
  source source_me  
  ./runMC $1 >& $2.log 
endif 

BTW, this is a good time to tell you that the directory I created with all the 
modified files I have mentioned and will mention here is 

fcdflnx3:~cwolfe/scratch/CDF4.5.3

The second part of the script takes the input file, parses out the section 
number of the file name (which takes the form mc014fb4.0001gene), sets two 
random number seeds from the section number which will be needed for the 
simulation, and then executes runMC. The program hextodec.exe, which can be 
found in the directory above, is a simple program which converts hex numbers 
to decimal. The code for this program can be found in 
fcdflnx3:~cwolfe/scratch/. 

So, that's easy enough. The next thing we need to is create the .tcl file. 
Again, this was taken from what Frank has, and modified slightly.

I don't want to list the entire .tcl file here, so I urge you to just copy the 
one I made which is fcdflnx3:~cwolfe/scratch/CDF4.5.3/farm.tcl. To actually do 
this yourself, create a .tcl file from Frank's instructions in the web page 
above. e.g.:

myRelease/mcProduction/tcl>../scripts/makeTcl.csh 85889 5000 template_WW.tcl 20

will make some .tcl files and place them in myRelease/mcProduction/tcl/85889/. 
Take one of them (any one really) and compare with my .tcl file noted above. 
Delete everything you don't see in the .tcl above -- I've tried to indicate 
which sections I've deleted in farm.tcl. (These files are just quite large, so 
I don't want to print them out here.)

Other changes to make which are important are:

set fRandomSeedGRNDM1  $env(GRNDM1)
set fRandomSeedGRNDM2  $env(GRNDM2)

#instead, specify the datasetID
set dataset_id mtestc

# This sets up the Random Number seed:
talk RandomGenManager
  RandomNumberMenu
    RandomSeedGRNDM1 set $fRandomSeedGRNDM1
    RandomSeedGRNDM2 set $fRandomSeedGRNDM2
  exit
exit

# Input the file...

module input DHInput

# for the database, DHInput should look something like this:
talk DHInput
   input file $env(INPUT_FILE)
exit

module output DHOutput
talk DHOutput
   output create    stream $dataset_id
   output setstream stream pathname=.
   output setstream stream dfc=production_file_catalog_write anysize list
   output path     stream $outputPath

This file should be copied in the base directory of the tarball, i.e. 
CDF4.5.3/farm.tcl.

So, the files you should have added in CDF4.5.3/ are runMC4.5.3 (which should 
be made executable -- chmod 755 runMC4.5.3), farm.tcl, and hextodec.exe (which 
should also be executable). That's it. Now make the tarball, and we're done:

      tar -cvzf runMC4.5.3.tar.gz CDF4.5.3

If you have any questions, email me, Collin, at cwolfe@hep.uchicago.edu.