# IDEs & Notebooks Development environments and notebook servers available on the cluster. # Jupyter on Slurm Run JupyterLab interactively on a compute node via SSH port forwarding. For a simpler browser-based experience, see **PowerIDE**. ## Step 1: Start an Interactive Job ```bash srun -p -A --pty bash ``` Replace `` and `` with your values from `check_my_partitions`. Note the compute node name you land on (e.g. `compute-0-62`) — you'll need it later. ## Step 2: Load the Jupyter Module ```bash module load mamba-env158/jupyter ``` ## Step 3: Start JupyterLab ```bash jupyter lab --ip=* --port=8892 --no-browser ``` Once started, you'll see output containing a URL with a token: ```bash http://localhost:8892/lab?token= ``` Copy the full URL including the token. ## Step 4: SSH Port Forwarding Open a new terminal on your local machine and run: ```bash ssh -N -L 8892::8892 @slurmlogin.tau.ac.il ``` Replace `` with the node from Step 1 (e.g. `compute-0-62`) and `` with your TAU username. Keep this terminal open. ## Step 5: Open in Browser Paste the URL from Step 3 into your browser: ```bash http://localhost:8892/lab?token= ``` ## Step 6: Closing When finished, press `Ctrl+C` in the interactive session to stop JupyterLab, then: ```bash exit ``` Also close the SSH tunnel terminal. ## Notes - If port 8892 is in use, pick any other free port — use the same number in both the `jupyter lab` command and the SSH tunnel - The SSH tunnel must stay open while you work - For a simpler experience without SSH tunneling, use **PowerIDE** instead # PowerIDE User Guide PowerIDE provides interactive access to the HPC cluster through a web browser. Run Jupyter notebooks, VS Code, and RStudio directly on compute nodes without needing SSH access. Access PowerIDE at: [https://poweride.tau.ac.il/jupyter](https://poweride.tau.ac.il/jupyter) ## Getting Started Log in with your TAU university credentials (same as email and other university services). After logging in, click **Start My Server**. You'll see a Server Options form to configure your compute resources. When you start your server, PowerIDE submits a Slurm job to the PowerSlurm cluster. Your session runs on a compute node — not on the PowerIDE server itself. This means: - You get dedicated resources (CPUs, memory, GPUs) on a compute node - Your job runs through the same Slurm scheduler as other HPC jobs - Your session will queue if the cluster is busy ## Configuring Resources ### Partition Select which partition to run on. The dropdown shows only partitions you have access to. Common options: - `power-general-shared-pool` — general purpose computing - `gpu-general-pool` — GPU-enabled nodes Check with your PI or HPC admin if unsure which partition to use. ### QOS Controls priority and resource limits. **Default (owner)** is usually the right choice. Only valid QOS options for your selected partition are shown. ### GPUs Appears only when a GPU partition is selected. Specify how many GPUs you need (0 if none). ### Time (D-HH:MM:SS) How long your session should run. Default: `04:00:00`. Your session is terminated when time runs out — save your work regularly. - `02:30:00` — 2.5 hours - `1-12:00:00` — 1 day and 12 hours ### CPUs per task Default: 1. Increase for multi-threaded code. ### Memory Default: 1G. Examples: `2G`, `8G`, `500M`. Start small and increase if needed — over-requesting delays job start. ### Working Directory Default: your home directory. Change to your project directory to save navigation time after launch. ### Stdout / Stderr Directory Where job logs are written. Default (home directory) is fine for most users. ## Starting Your Session Click the orange **Start** button. PowerIDE submits a Slurm job and shows a progress page. Once a compute node is available (usually 10–60 seconds), you're automatically redirected to JupyterLab. If the cluster is busy, you can close the browser and come back — your session will start when resources are available. ## Using JupyterLab - **Left sidebar** — file browser, running kernels, extensions - **Main area** — notebooks, text files, terminals - **+** button — opens the launcher for new tools Common tasks: - **New notebook** — click **+** → choose a Python kernel - **Terminal** — click **+** → click Terminal (bash shell on the compute node) - **Upload files** — drag and drop into the file browser - **Download files** — right-click file → Download ## Using VS Code PowerIDE includes VS Code running in your browser: 1. Click **+** to open the launcher 2. Click the **VS Code** icon 3. VS Code opens in a new tab with access to all your files and the same resources as JupyterLab ## Using RStudio RStudio runs as a separate service on a dedicated compute node. It has its own launch form with R-specific options: 1. Click **+** to open the launcher 2. Click the **RStudio** icon 3. Fill in the resource form and select your R environment 4. Click **Start** — RStudio opens in a new tab once the job is running ### R Environment Select the R environment to load. Each environment is a named module (e.g. `webR-genomics-2024`) with R and a pre-installed set of packages. Contact HPC support if you need a package that isn't available. ### R Library Path (optional) If you have a personal R package library installed in a directory on the cluster, enter its full path here (e.g. `/home/user/R/library`). R will search this directory first, before the environment's default library. ### Stopping RStudio Use the **Stop** button in the PowerIDE topbar to terminate your RStudio job. **Do not use File → Quit Session** — that ends the R session but leaves the Slurm job running, continuing to consume resources. ## Python Environments PowerIDE provides one default kernel: **Python 3.12 (Base)**. You can register your own conda/mamba environments as kernels: ```bash module load mamba/mamba-2.1.1 mamba create -n my-project python=3.11 pandas matplotlib mamba activate my-project mamba install ipykernel # Register as kernel (only visible to you) python -m ipykernel install --user --name my-project --display-name "My Project (Python 3.11)" ``` Refresh your browser — the new kernel appears in the launcher. To remove a kernel: ```bash jupyter kernelspec uninstall kernel-name ``` ## Stopping Your Server **Always stop your server when done** to free resources for others. - **From JupyterLab** — File → Hub Control Panel → Stop My Server - **From PowerIDE home** — navigate to [https://poweride.tau.ac.il/jupyter/hub/home](https://poweride.tau.ac.il/jupyter/hub/home) → Stop My Server - **VS Code / RStudio** — use the **Stop** button in the PowerIDE topbar ## Best Practices - Request only what you need — over-requesting delays your job and others - Set a realistic time limit; restart if you need more - Only request GPUs if your code actually uses them - Store large datasets in scratch space, not your home directory - Use Git for code — not for large data files