Instructor Notes

This is a placeholder file. Please add content here.

Introduction


Instructor Note

The first thing to do is to get all learners to plug in the hardware for their cluster. On its very first boot, the Pi automatically expands the root filesystem to fill the SD card. This can take a minute or two, during which the network interface will not yet be up.

This is a good time to deliver the lesson introduction. By the time you have finished, the Pi will be ready to connect to.



Preparing an SD Card


Instructor Note

The process of downloading the imager, getting students to flash it to the Pi, and navigating permissions errors on managed machines, means that this section can take some time. Consider pre-flashing SD cards for learners and skipping this section. If you do wish to run it, be prepared for this to take a while, and ensure you have enough helpers on-hand to assist with the software step.

If pre-flashing cards, consider setting up a multi-SD flashing rig using a USB-C hub with sufficient capacity, multiple fast SD card readers (USB-3 at minimum), fast SD cards (e.g. microSDHC with U3 speed rating at minimum), to flash multiple cards simultaneously. Software such as hypriot/flash can be useful for this purpose as it allows you to batch script and customise Pi images on the command line, rather than manually operating the GUI.



Booting and Updating


Instructor Note

.local resolution failures are most common on Windows. Linux learners in the same group are usually unaffected and can confirm the Pi is up. If a group is stuck, look up the Pi’s IP from the router interface and give it to them directly — this is the fastest recovery path.

Consider using a serial KVM controller software such as kvm-serial and HDMI capture interface to connect directly to the console of the user’s Raspberry Pi. The login screen prints the IP directly to the display.



Configuring the login node


Configuring a compute node


Some extra things that can be done


Testing & running your first job


Preparing compute nodes for eessi