To start a job at the open science grid you need 3 things:
- an account
- a condor file
- something to do in the form of a Shell script and a tar.gz file

FIRST:
The account.

SECOND:
The condor script (example attached as lukas.cmd). It defines in which 
environment you want your job to take place.
Here is the content of lukas.cmd, lets go through it:

First there is a default entry in a condor file, if you use normal linux 
programs.
> universe=vanilla
Then you tell what you want to do in the form of a shell script. You 
will find it attached. What the attached SPhenoGridTest.sh does is very 
simple: It compiles SPheno and runs a LesHouches.in File
> executable=/home/lmitzka/SPhenoGridTest.sh
Then I always had the stream options as false (just because I don't 
really know what their benefits are)
> stream_output = False
> stream_error  = False
Then I specify that I want to get the results when the job is finished.
> should_transfer_files = YES
> WhenToTransferOutput = ON_EXIT
>
Then one has to spcify which tar.gz comes with the commands.
> transfer_input_files = /home/lmitzka/SPheno-3.2.3new.tar.gz
One can also tell condor to only copy specific files. If you provide 
nothing, it gives you back all the files that are present in the main 
folder on the distant computer
> transfer_Output_files =
One can give it a filename where it stores the condor output
> log    =  /home/lmitzka/logquestionmark.log
I don't really know anymore what the lines mean:
> Notification = Never
> +Owner = undefined
>
Then one has to customize the output names. In the .out files the output 
of the console is stored, the errors in the file ending in .err. The 
statement $Cluster is so to say denoting the job you are doing and 
$Process denotes which of the processes is meant if you do something in 
a queue.
> arguments= $(Cluster).$(Process)
> output = /home/lmitzka/1e.$(Cluster).$(Process).out
> error  = /home/lmitzka/1e.$(Cluster).$(Process).err
Here you tell that you want to run the above .sh file 5 times
> queue 5
>

The SPhenoGridTest should be self-explanatory. The only thing you may 
find surprising are the wildcards
'$1' in lines
> echo "I am Condor job $1"
or in
> mv SPheno.spc SPheno.spc.$1
If I remember correctly it means "the first statement in the .cmd file 
in the line with 'arguments' ", which in this case is
> arguments= $(Cluster).$(Process)


So all the output filenames will be
SPheno.spc.X.Y with X and Y being numbers

one executes via
condor_submit lukas.cmd

and checks the status via 
condor_q

to delete a job you have to do
condor_rm <your username>