OS Installation on Turing Worker

This page describes simple steps to install the OS (HypriotOS) on the PIs.

Overview

Turing uses HypriotOS because the quickest to set up. SSH and Docker supported by default

OS on Turing worker/slave

If your NAT setup was successfull on the master PI, it is time to push the SD card on the worker PIs, plug them to the ethernet switch and enable the power.

Access the node

  • Insert the SD card in the Turing Worker and power it on.
  • SSH to the Turing Master.
  • From the Turing Master, ssh to the new Turing Worker. And IP address in 192.168.2.1xx will have been assigned to the new node.

Freeze your configuration

Cloud init is perfect for the first boot. Once the node is up, it can be challenging not to preserve the fine tuning done to the OS.

sudo apt-get remove --purge cloud-init
sudo apt-get autoremove

Update OS

Let’s update to the latest version

sudo apt-get update
sudo apt-get upgrade

Gain access to latest docker-ce

sudo -i

curl -fsSL https://get.docker.com -o get-docker.sh
sh get-docker.sh
usermod -aG docker pirate

At the time this article is written 18.09 is not supported yet.

sudo apt-cache policy docker-ce
sudo apt-get remove --purge docker-ce
sudo apt-get autoremove

sudo apt-get install docker-ce=18.06.1~ce~3-0~debian

Update Turing Worker name

As root, replace black-pearl by kube-nodeXX, in the two following files:

sudo -i

vi /etc/hosts

It seems on HypriotOS 64, you need to do

sudo hostnamectl set-hostname kubemaster-pi

The /etc/resolv.conf on the new node should contain the IP address of your home router if dhcpd is setup properly on your master node.

Update the static IP on the Turing Master

It is important to edit the dhcpd.conf file in order to enfore the same internal IP address on the 192.168.2.x network every time the cluster is rebooted. Be sure to check then isc-dhcp-server is still working and that you need not corrupt the configuration file.

arp -a

sudo vi /etc/dhcpd/dhcpd.conf

sudo service isc-dhcp-server restart