Skip to content

Latest commit

 

History

History
248 lines (175 loc) · 7.53 KB

README.md

File metadata and controls

248 lines (175 loc) · 7.53 KB

Positron over Remote SSH

Launch Positron on Alpine (CU Boulder) or amc-bodhi (CU Anschutz) HPC clusters.

Overview

These scripts allocate a compute node on your HPC cluster and provide SSH connection instructions for remote development with Positron. The scripts use a ProxyJump SSH configuration to connect through the login node to your allocated compute node.

  • Alpine (CU Boulder): Uses SLURM job scheduler (alpine-positron.sh)
  • amc-bodhi (CU Anschutz): Uses LSF job scheduler (bodhi-positron.sh)

Prerequisites

  • Access to Alpine or amc-bodhi HPC cluster
  • Positron installed on your local machine
  • SSH key configured for cluster access
  • amc-bodhi only: Connected to AMC VPN

Setup (One-time)

1. Add your local machine's SSH key to Alpine

If you haven't already, you need to add your local machine's public SSH key to Alpine's authorized_keys file. This allows your local machine to authenticate with Alpine compute nodes via the ProxyJump connection.

  1. On your local machine, copy your public key:

    cat ~/.ssh/id_rsa.pub

    Or if you use a different key:

    cat ~/.ssh/id_ed25519.pub

  2. Log into Alpine and add the key to authorized_keys:

    ssh <your-username>@login.rc.colorado.edu

  3. On Alpine, add your public key:

    mkdir -p ~/.ssh
chmod 700 ~/.ssh
echo "your-public-key-content" >> ~/.ssh/authorized_keys
chmod 600 ~/.ssh/authorized_keys

    Replace your-public-key-content with the output from step 1.

2. Configure Positron Server location on Alpine

By default, Positron installs its server components to ~/.positron-server, but $HOME on Alpine has limited space. To avoid filling up your home directory, set up Positron Server on /scratch/alpine:

On Alpine (login or compute node):

# Create directory on scratch (larger allocation)
mkdir -p /scratch/alpine/${USER}/.positron-server

# Create symlink from home to scratch
ln -s /scratch/alpine/${USER}/.positron-server ~/.positron-server

Important notes:

  • /scratch/alpine is purged every 90 days of files not accessed
  • If the directory is purged, Positron will automatically reinstall the server when you next connect
  • You may need to recreate the symlink if it's removed: ln -s /scratch/alpine/${USER}/.positron-server ~/.positron-server
  • For more details on how Positron Remote-SSH works, see: https://positron.posit.co/remote-ssh.html#how-it-works-troubleshooting

Quick Start (Alpine)

1. Submit the job to Alpine

sbatch alpine-positron.sh

2. Check job status

squeue -u $USER

Wait until your job is in the "R" (running) state.

3. View connection instructions

cat logs/positron-<JOB_ID>.out

Replace <JOB_ID> with your actual job ID from squeue.

4. Connect from Positron

  • Open Positron on your local machine

    • Press Cmd+Shift+P (Mac) or Ctrl+Shift+P (Windows/Linux)
    • Select "Remote-SSH: Open SSH Configuration File"

  • Paste in your SSH config (from the log file) and save:

    Host positron-<JOB_ID>
    HostName <compute-node>
    User <your-username>
    ProxyJump <your-username>@login-ci.rc.colorado.edu

  • Select "Remote-SSH: Connect to Host"

  • Choose positron-<JOB_ID> from the list

  • Positron will install its server components on the remote node automatically

5. When finished

Always cancel your job to free resources:

scancel <JOB_ID>

Quick Start (amc-bodhi)

Important: You must be connected to the AMC VPN before proceeding.

1. Submit the job to amc-bodhi

bsub < bodhi-positron.sh

2. Check job status

bjobs

Wait until your job is in the "RUN" state.

3. View connection instructions

cat logs/positron-<JOB_ID>.out

Replace <JOB_ID> with your actual job ID from bjobs.

4. Connect from Positron

  • Open Positron on your local machine

    • Press Cmd+Shift+P (Mac) or Ctrl+Shift+P (Windows/Linux)
    • Select "Remote-SSH: Open SSH Configuration File"

  • Paste in your SSH config (from the log file) and save:

    Host positron-bodhi-<JOB_ID>
    HostName <compute-node>
    User <your-username>
    ProxyJump <your-username>@amc-bodhi.ucdenver.pvt
    ForwardAgent yes

  • Select "Remote-SSH: Connect to Host"

  • Choose positron-bodhi-<JOB_ID> from the list

  • Positron will install its server components on the remote node automatically

5. When finished

Always cancel your job to free resources:

bkill <JOB_ID>

Configuration

Alpine (alpine-positron.sh)

Resources are configured via SLURM directives:

  • --time=08:00:00 - Maximum job duration (8 hours)
  • --mem=20gb - Memory allocation (20 GB)
  • --partition=amilan - Alpine partition for general compute
  • --qos=normal - Quality of service tier

See Alpine documentation for available options.

amc-bodhi (bodhi-positron.sh)

Resources are configured via LSF directives:

  • -W 8:00 - Maximum job duration (8 hours)
  • -R "rusage[mem=20000]" - Memory allocation (20 GB)
  • -q normal - Queue for general compute
  • -n 1 - Number of tasks

Troubleshooting

Job won't start

  • Check queue status: squeue -u $USER
  • Check available resources: sinfo
  • Verify your account has hours: curc-quota

Can't connect via SSH

  • Ensure SSH config was added to your local ~/.ssh/config (not on Alpine)
  • Verify job is running: squeue -u $USER
  • Check log file for correct hostname and job ID
  • Verify your local SSH public key is in Alpine's ~/.ssh/authorized_keys (see Setup section)

Connection drops

  • SSH connection may timeout if idle. The job itself will continue running.
  • Reconnect using the same SSH host entry.

R interpreter not discovered

R interpreter discovery can be unreliable on remote systems. If you don't see R under "Start Session" (even though Python interpreters appear):

If you find you are able to use R versions available in the modules system, please let me know.

  1. Install R through mamba/conda on the remote system:

    mamba install -c conda-forge r-base

  2. Enable conda discovery in Positron settings (on your local machine):

    • Press Cmd+, (Mac) or Ctrl+, (Windows/Linux) to open Settings
    • Search for "Positron R Interpreters Conda Discovery"
    • Enable the checkbox, or add to your settings.json: 
      {
    "positron.r.interpreters.condaDiscovery": true
}

  3. Manually trigger interpreter discovery:

    • Press Cmd+Shift+P (Mac) or Ctrl+Shift+P (Windows/Linux)
    • Select "Interpreter: Discover all interpreters"

After these steps, R should appear in the interpreter dropdown and start successfully.

Note: There are multiple discussions in the Positron repository about interpreter discovery issues, though solutions may vary by system configuration.

Notes

  • The compute node allocation will run for the full time requested or until you cancel it
  • Always remember to scancel your job when done to free resources
  • Log files are stored in the logs/ directory with the pattern positron-<JOB_ID>.out

Resources