Kubernetes-training

Use env

Prerequisites:

• Docker 17.05+ from the edge channel
• Go 1.6 (better 1.7) if you do not use Mac
• The moby and linuxkit binaries. In bin/ the versions for OS X. To build them (i.e. who's on Linux):

git clone github.com:linuxkit/linuxkit $GOPATH/src/github.com/linuxkit/linuxkit
cd $GOPATH/src/github.com/linuxkit/linuxkit
make
cp bin/* $THIS-LOCAL-REPO/bin

Build a K8s env

• git clone [email protected]:fsoppelsa/kubernetes-training.git
• cd kubernetes-training/env
• Paste your SSH id_rsa.pub key file content in .yml files around the end, substitute # Your ssh key goes here
• cd .. ; mv bin .. # hardcoded stuff, because ../../bin/moby is invoked
• cd env ; make build-vm-images # may take up to 10 minutes
• ./boot-master.sh # This starts a master via hyperkit on OS X, on Qemu on Linux
• This will be up in 2 or 3 seconds. Type ENTER and then get the eth0 address: ip addr show dev eth0
• Open a new shell and: ./ssh_into_kubelet.sh <IP>
• In this new shell: kubeadm-init.sh # Takes 1, 2 minutes.
• Wait until the master is Ready!!! Check with kubectl get nodes until you see moby-xxxxx Ready
• Take note of the token in output, something like cd9b45.501fc7d4fc481545
• Only when the Master node is Ready, open a new shell and join a node: ./boot-node.sh 1 --token <T> <IP>:6443
• Open an extra new shell and join another node: ./boot-node.sh 2 --token <T> <IP>:6443
• For example ./boot-node.sh 1 --token cd9b45.501fc7d4fc481545 192.168.0.5:6443
• You can add as many minions as you want. For training, 2 are ok

Now you have 4 shells: 1 for Master boot, 1 is SSH to Master, 1 is Node1 boot, 1 is Node2 boot.

For the training tests we'll use always the one with SSH into the Kubelet, the second in order. Do not close others. To shutdown a machine in the other 3 shells at the end of training, just type halt in it.