'condor_submit i' is probably the fastest way to get you running on the NAF as this will result in a ssh-login on a workernode with the batch environment in a default slot with 1 core, 2gb of memory and 3h runtime:
|[chbeyer@htc-it02]~/htcondor/testjobs% condor_submit -i|
1 job(s) submitted to cluster 1389.
Waiting for job to start...
Welcome to firstname.lastname@example.org!
You will be logged out after 7200 seconds of inactivity.
Even more sophisticated for debugging purpose you can run 'condor_submit i <your_submit_file>' which will run your job and give you an interactive ssh-login in the slot on the workernode where the job is running:
Another way to interact with a running batchjob is 'condor_ssh_to_job <JOBID>' which gives you an interactive login on the jobslot of the given job: