Research Cluster Q/A

This page includes commonly asked questions using the Research Cluster.

Question Topics:

Installing and Uninstalling Software/Packages

1. When installing packages, do you need to launch a container first?

   1. Yes. You must first launch a container then use pip install. The package will be saved in your home directory and available to use each time you log into your account and for any container.

2. Can you run GPU containers in the terminal?

   1. Yes. Note: You cannot launch duplicates of the same container at the same time – meaning one in the terminal and one datahub.

3. How do I view packages installed in my home directory?

   1. After ssh and launching a container, enter ‘pip list --user’.

4. How do I install packages in my home directory?

   1. After ssh and launching a container, then install package using ‘pip install <package>’.

5. How do I uninstall packages from my home directory?

   1. After ssh and launching a container, then install package using ‘pip uninstall <package>’

Launching containers in the terminal/command prompt

1. How do you launch containers in JupyterHub from the terminal?

   1. After launching the container in terminal, a link will be provided. Copy and paste the link in the browser. Make sure you are not already logged into the datahub.

Uploading files in the terminal

1. How do you get to your home directory?

   1. For researchers, enter ‘ls’ after logging in.

   2. For admin), enter ‘cd /datasets/home/72/072/kkt008/’ then ‘ls’

2. How do you upload files from OneDrive files to your home directory?

   1. For researchers:

      • No just navigate to the OneDrive folder via terminal and follow instructions here.

      • Afterwards, ssh into Research cluster and enter ‘ls’ to view files.

   2. For me (admin):

      • At ‘sftp>’ step, enter ‘cd /datasets/home/72/072/kkt008/’. Then enter ‘put filename.txt’ to upload files.

3. How do you upload files to shared storage (e.g. lab, group, or USS storage)?

   1. For researchers:

      • For saving to team shared storage enter ‘cd /uss/sharename’ (after sftp>) as a the step prior to uploading files. To view files after logging into the RC, enter ‘cd /dsmlp/teams’.

         To upload files via the terminal to the group storage, proceed with the following steps:

        • Via the terminal navigate to the folder that stores the file(s).

        

        • Enter ‘ls’ to list the files.
        • Enter ‘sftp bah@dsmlp-login.ucsd.edu’ and enter your AD password.
        • Enter ‘cd /uss/labname’.
          1. Entering this code will direct your uploads to the team storage. This command is not shown in the image below.
        • Enter ‘put filename.txt’. Repeat this code for every file to be uploaded.

        

        • Enter ‘exit’ or ‘quit’ when done uploading files.
        • Next, ssh in the Research Cluster ‘ssh dsmlp-login.ucsd.edu’ and enter AD password
        • Enter ‘cd /dsmlp/teams’ to view files in the shared storage.

   2. For administrators:

      • At ‘sftp>’ step, enter ‘cd /uss’. Then 'cd /sharename'. Next, enter ‘put filename.txt’ to upload files.

Jupyter Notebook: File uploads

1. How to upload files in Jupyter Notebook?

   1. Log in at datahub.ucsd.edu.

   2. Launch your preferred environment.

   3. Navigate to your teams folder.

   4. Select ‘Upload’ to transfer new files.

Jupyter Notebook: Installing and Uninstalling Software/Packages

1. How do I view packages installed?

   1. From your homepage, select ‘New’ > ‘Terminal’. In the Jupyter Notebook terminal, enter ‘pip list --user’.

2. How do I install packages in my home directory?Jupyter Notebook terminal:

   1. From your homepage, select ‘New’ > ‘Terminal’. In the Jupyter Notebook terminal, enter ‘pip install <package>’

3. How do I uninstall packages from my home directory?

   1. From your homepage, select ‘New’ > ‘Terminal’. In the Jupyter Notebook terminal, enter ‘pip uninstall <package>’