VS Code

Overview

VS Code is a free editor from Microsoft, which is open source and support many useful extensions. 

It may be a good alternative to terminal based editors (like vim, nano and Emacs) for some of our users.

Benefits

• • Work with files on cluster, almost as if they would be on your own laptop

  • Great syntax highlighting, linting, etc. for many languages (using extensions). Tools for debuging

  • Extensions simplifying work with git (recommended: Git Graphs)

  • Graphs produced on server, can be viewed directly withing VS Code

  • When you are done editing - you can run sbatch job from terminal panel of VS Code

  • You can set up port forwarding

  • And more

Using VS Code 

IMPORTANT: don't run computations in VS Code!

Use VS Code only to edit scripts, use linting, etc.

Only run calculations using srun/sbatch

Download and Install

Download and install it on your local machine: download

Setting Up Remote Connections to HPC cluster

• Make sure you activated NYU VPN

  • VS Code Lock files shall NOT be saved in /home. Ways to achieve that

  • In extensions tab of VS Code find extension "Remote - SSH"

  • press settings icon -> Extension settings

  • Do one of the following:

    • Don’t use lock files (uncheck "Remote.SSH: Use Flock")

    • Store lock file in /tmp (check Remote.SSH: Lockfiles In Tmp)

  • Inside VS Code

  • F1

  • remote-ssh: Connect to Host

  • New host (or - Configure SSH host)

  • ssh <user>@greene.hpc.nyu.edu

  • specify the path to the file, where the configuration of this connection will be saved

More details here: https://code.visualstudio.com/docs/remote/ssh

Login using ssh key

• • Instead of typing your password every time you need to log in, you can also specify an ssh key.

  • Only do that on the computer you trust

  • Generate ssh key pair (terminal in Linux/Mac or cmd/WSL in Windows): https://www.ssh.com/ssh/keygen/

  • Note the path to ssh key files. Don't share key files with anybody - anybody with this key file can login to your account

  • Log into cluster using regular login/password and then add the content of generated public key file (the one with .pub) to $HOME/.ssh/authorized_keys on cluster

  • Next time you will log into cluster no password will be required.

How to Connect VS Code to a Greene Compute Node

Sometimes users want to connect their VS Code GUI to a compute node, such as to launch a Jupyter Notebook file on Greene. This requires several steps to properly route through Greene's login nodes to the compute nodes in use. Please follow these instructions to connect and run VSCode server on Greene compute node:

1. On Greene login node, submit a dummy Slurm job to sleep there, 1 CPU core and 1GB memory for 2 hours. This step will enable you to login to compute nodes you have jobs running there.
   
   sbatch --time=02:00:00 --mem=1GB --wrap "sleep infinity"

2. After the job is running, please check the compute node this job is running there with squeue
   
   [sw77@log-1 tmp]$ squeue -u sw77
   JOBID    PARTITION     NAME     USER ST       TIME  NODES NODELIST(REASON)
   23587213 cpu_gpu       wrap     sw77  R      18:29      1 gr070
   
   In this example, the job is running on gr070

3. On your local computer, add this block to ~/.ssh/config file
   
   Host greene-compute
   HostName <Compute Node your Jupyter Notebook is using e.g. gr070>
   User <NetID>
   ProxyJump <NetID>@greene.hpc.nyu.edu
   StrictHostKeyChecking no
   UserKnownHostsFile /dev/null
   LogLevel ERROR
   
   Here the ProxyJump is <NetID>@greene, if you are in campus, please change to YourNYUNetID@greene.hpc.nyu.edu. Please be sure to be connected to the NYU VPN (vpn.nyu.edu).

4. Set up ssh login to Greene with ssh key without password (see section above)

5. Now from VSCode, you can directly connect to  greene-compute nodes with:

ssh greene-compute

More settings

VS Code has many more settings and options. Explore!

Additional notes

High CPU load on login node from VS Code

You may notice (please check with 'top') that your VS Code connection causes 'node' process running from your user to use a lot of CPU resources. One of the reason leading to that - large number of files within your home directory. Try to remove conda and pip environments from the home directory and check if this will resolve the issue.

Python

• • Recommended extensions: Python, Anaconda Extension Pack

    • When those are installed you can switch python env or conda env by clicking on the left bottom line specifying python.

    • Support for python scripts and Jupyter Notebooks

  • Syntax highlighting: extension MagicPython