Connecting to supercomputer#
Web interface#
For lightweight entry to supercomputers
In web interface the resources are limited -> suitable for developing code and small analysis tasks
Bigger analysis tasks should be run via batch jobs
Web interface can be used for starting batch jobs
Tools in web interface:#
View, download and upload files
Terminal to login node
Terminal to compute node
Info: running jobs, disk usage, project status and supercomputer’s general status
Launch interactive apps and open them directly from the browser:
Desktop with apps: QGIS, GRASS, SagaGIS, SNAP, Zonation etc
Jupyter
TensorBoard, MLFlow
Visual Studio Code
RStudio
MATLAB
Connecting to the supercomputer via SSH#
During the course we will access the supercomputer via the web interface in order to not overwhelm you with setups before the course. However, this way may not always be the most convenient. You can also connect to the supercomputer via SSH.
Connecting with SSH clients
SSH clients give command-line access to a supercomputer or any other Linux server.
Basic SSH connection will not allow displaying remote graphics, but it is possible to set up with extra settings.
SSH clients:
To avoid typing your password every time and to make your connection more secure, you can set up SSH-keys.
Developing scripts remotely#
Instead of developing code on your local machine and moving it as files to the supercomputer for testing, you can also consider to use a local editor and push edited files directly to the supercomputer. This works for example with Visual Studio Code or Notepad++. Note that Visual Studio Code is also available through the Puhti web interface.