NowWhatHub

NowWhatHub provides browser-based access to NowWhat compute resources. Each server you start creates a Slurm job with the CPU, memory, GPU, and runtime requested for your work.

Start a server

1. Sign in to NowWhatHub with your institutional credentials.
2. Select Start My Server. If you already have a server, select New Server to create another named server.
3. Choose the resources required by your work:
   • CPU or GPU partition;
   • maximum runtime;
   • number of CPU cores;
   • memory;
   • GPU type and quantity, when using the GPU partition.
4. Select Start and wait for Slurm to allocate the requested resources.

JupyterLab opens when the server is ready. Smaller resource requests will generally start sooner.

Choose the GPU partition before starting the server if your notebook or application needs a GPU. Opening an application later does not change the server allocation.

Open an application

Use the JupyterLab Launcher to choose what to open. Select the + button in the File Browser if the Launcher is not visible.

Python or R notebook

Under Notebook, select the Python or R kernel associated with the environment you want to use. The environment name is shown in the kernel name.

Use a notebook for interactive analysis, data exploration, and code that is naturally organized into cells.

RStudio

Select an RStudio (environment-name) card. RStudio opens in a new browser tab and uses the indicated R environment.

Use RStudio for R projects, scripts, packages, and R Markdown or Quarto documents.

VS Code

Select VS Code. It opens in a new browser tab and runs inside the same server allocation.

Use VS Code for projects with multiple files, software development, debugging, and integrated terminal work.

Desktop

Select Desktop to open a remote Linux desktop in a new browser tab.

Use the Desktop for applications with a graphical interface, such as napari, or when you need a file manager and a complete desktop environment. The Desktop session starts on the login node; check the execution location before starting computationally intensive work.

Your browser may ask you to allow pop-ups when opening RStudio, VS Code, or Desktop.

Run commands on the correct node

A terminal opened from JupyterLab runs on the login node, not automatically on the compute node assigned to the NowWhatHub server. The Desktop session also starts on the login node.

Check your current location before running a command:

hostname

Use the login node only for lightweight operations such as navigating files, editing scripts, and connecting to the allocated compute node. To run a computation from a JupyterLab or Desktop terminal, connect through SSH to the node assigned to your server:

ssh <assigned-compute-node>
hostname

Start the computation only after hostname confirms that the terminal is on the expected compute node. Access to that node is tied to the active Slurm allocation and ends when the NowWhatHub server is stopped.

Micromamba environments

When a server starts, NowWhatHub loads the available Micromamba environments, including:

• shared environments provided under /software;
• your personal environments in your home directory.

Compatible environments appear in the Launcher as Python or R notebook kernels and RStudio entries. The environment name is included in each entry so you can select the correct one.

If you create or modify a personal environment while the server is running, stop and start the server again to reload the environment list.

Stop a server

Closing a notebook, an application tab, or the browser does not stop the server.

To release the Slurm resources:

1. Save your notebooks and files.
2. Return to the NowWhatHub home or control panel.
3. Find the running server and select Stop.
4. Wait until the server is shown as stopped.

Stop every named server that you no longer need, especially GPU servers. Files saved in persistent storage remain available; running processes, unsaved changes, and in-memory notebook state are lost when the server stops.