Charbonneau Research Group

How to start working at Patrick's Lab (For new group members)

   Edited by Kai Zhang 2012

This site is for new members who might want to know some primary information about how to start working in our lab. Typically, we code in C and run our jobs on DSCR clusters. Once you have your own computer, please follow the steps:

1. Have your net jack opened and NetID registered by the department IT supporting staff to get access to the internet from our lab in FFSC.

2. Have your NetID validated for accessing to Prof. Patrick Charbonneau's cluster resources by the department IT supporting staff. Once this is done, you can see your own folder under Patrick's directory on the clusters when you log in.

3. Login the cluster. Ways to login:

     •Linux Users:

          ssh NetID@dscr-login-05.oit.duke.edu (see DSCR ). After login to the cluster, you could login to any one of the nodes (e. g. node 01) by

          ssh NetID@pcharbon-n01 (gdb needs to be run under this environment)

    •Windows Users:

          If you don't want to install Linux simulator, try this tiny software: PuTTY (NetID needed)

          If you have enough free space and memory to run Linux simulator, try this free and memory saving software: Cygwin. You only need ~800MB space to download the installation package and ~2GB space to install the software.

          If you are an IT expert, do whatever you like.

4. Have your code compiled and ready to be executed.

5. You could run the program on the node directly if it is small. Otherwise, you need to submit the job to the cluster. The DSCR uses SGE Queueing System. You first need to write a *.q file (see example), make it executable by chmod +x *.q, then submit it by qsub *.q and let your job run on the cluster. Basic commands of SGE system:

           submit a job:                           qsub *.q

           delete all your jobs:                 qdel -u username (NetID)

           delete a job:                            qdel 1234567 (job ID)

           show all your jobs status:        qstat -u username

           show the detail of a job:          qstat -j 1234567 (job ID)

  Some scripts already existing on the cluster may facilitate you.

 

6. Backup your data at chemistry cluster

    login to ssh NetID@login.chem.duke.edu

    go to directory /data/PC/

    create your own folder, say Paddy, and transfer data there, e.g.

    scp -r NetID@login.chem.duke.edu:/home/run  /data/PC/Paddy