K8s Installation/Process
- A detailed manual procedure to install K8s software and PKC data content.
- Incrementally document installation process in PKC, and slowly migrate manual processes into automated processes by Jenkins, Ansible, and Terraform. These actions should be hyperlinked in corresponding PKC pages.
Installation Outline
After watching many videos on installing Kubernetes, the following outline is extracted from Edureka's video tutorial. One should also have browsed through this book[1] at least once.
The suggested initial configuration is to start with one master node and one worker node.
The master node must have at least 2 CPU cores, 2 Gb of memory. The worker node (slave) should have at least 1 CPU core and 2 Gb of memory.
If one needs to add more machine, one can do it after the above two machines are working.
Ideally, set up the domain names of these two nodes with their public IPv4 addresses.
Set the right Security Groups
{{#lst:Input/K8s Installation/Security Groups|Security Groups}}
Procedure for both Master and Slave
To automate this repeated process on many machines, consider using Ansible or Terraform. The usual initial action is to update, but make sure that you switch to the super user mode and keep being in that role throughout most of the installation.
sudo su apt update
Swap Space must be turned off
Then, turn off swap space
swapoff -a
The /etc/fstab file must be edited to remove the line entry that specifies a /swapoff directory. It is possible that the /swapoff line doesn't exist already. (Ubuntu 20.04 Practice)
nano /etc/fstab
Install Net-Tools
Before inspecting the IP addresses of these machines, you will need to first install Net-Tools.
apt install net-tools
Afterwards, run the instruction:
ifconfig
This will show some information, such as follows:
eth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 9001
inet 172.31.20.148 netmask 255.255.240.0 broadcast 172.31.31.255
inet6 fe80::5d:ddff:fea4:9331 prefixlen 64 scopeid 0x20<link>
ether 02:5d:dd:a4:93:31 txqueuelen 1000 (Ethernet)
RX packets 14905 bytes 20660726 (20.6 MB)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 2247 bytes 270976 (270.9 KB)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0
lo: flags=73<UP,LOOPBACK,RUNNING> mtu 65536
inet 127.0.0.1 netmask 255.0.0.0
inet6 ::1 prefixlen 128 scopeid 0x10<host>
loop txqueuelen 1000 (Local Loopback)
RX packets 198 bytes 17010 (17.0 KB)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 198 bytes 17010 (17.0 KB)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0Update Hostname, Hosts, and Setup Static IP
Then, update the hostname file at the /etc/hostname location:
nano /etc/hostname
Change the master node and worker node to kmaster and worker1 respectively.
The string that shows up in this file /etc/hostname, will eventually show up in the command line interface.
It will show up in a terminal app with prompts looking like: root@kmaster: or root@worker1:.
Static IP address
Notice the inet value: 172.31.20.148. This is the static and private IP address, that can be used even after one reboots the machine.
One may set up the static IP address here:
nano /etc/network/interfaces
Then, put in the following textual content:
auto enp0s8
iface enp0s8 inet static
address <IP-Address-of-masters and slaves>In the above mentioned example, <IP-Address-of-masters and slaves> is 172.31.20.148. Note that this step needs to be executed for each master and slave/worker node.
Statically defined host names
After setting up the network interfaces, one needs to set up a static look up table for the /etc/hosts file:
nano /etc/hosts
Then, put in the following textual content:
Note that this step needs to be executed for each master and slave/worker node.
127.0.0.1 localhost
172.31.28.100 worker1
172.31.20.148 kmaster
# The following lines are desirable for IPv6 capable hosts
::1 ip6-localhost ip6-loopback
fe00::0 ip6-localnet
ff00::0 ip6-mcastprefix
ff02::1 ip6-allnodes
ff02::2 ip6-allrouters
ff02::3 ip6-allhostsNote that one may put in multiple entries at once, in this case, two entries, w1 and kmaster, are registered in the /etc/hosts file.
At this time, one may issue the command reboot, to restart the machine.
reboot
After rebooting, the machines should show the host name in command line. More importantly, use ifconfig to check if the locally-defined IP address has been kept stable.
Install SSH Server and Docker
If ssh server is not available, it can be set up by:
sudo apt-get install openssh-server
Afterwards, one may install docker
sudo apt-get update sudo apt-get install -y docker.io
Then, one must make sure curl and other things are available:
sudo apt-get update && sudo apt-get install -y apt-transport-https curl
Then, switch to root and use curl to install the following
sudo -su root curl -s https://packages.cloud.google.com/apt/doc/apt-key.gpg | apt-key add -
kubeadm, kubelet and kubectl
Then, one needs to add a file to the Debian package list:
cat <<EOF >/etc/apt/sources.list.d/kubernetes.list
deb http://apt.kubernetes.io/ kubernetes-xenial main
EOF
Then, run the following two commands:
apt-get update apt-get install -y kubelet kubeadm kubectl
These two commands could be replaced by compressing all the above into one instruction
echo "deb http://apt.kubernetes.io/ kubernetes-xenial main" \
| sudo tee -a /etc/apt/sources.list.d/kubernetes.list \
&& sudo apt-get update
Update the kubeadm.conf file
The following file must be edited with an extra data entry:
nano /etc/systemd/system/kubelet.service.d/10-kubeadm.conf
In this file, add the following entry to the last line.
Environment="cgroup-driver=systemd/cgroup-driver=cgroupfs"
Then, switch back to regular user by ctrl+D
Only do this for the Master Node
It is time to use the kubeadm init command. Note that the output of kubeadm init should be recorded for other workers to join the cluster.
sudo kubeadm init
Or one may specify more parameter using this instruction:
sudo kubeadm init --pod-network-cidr=<depends on calico or flannel pod network> --apiserver-advertise-address=<ip-address-of-master>
An example based on Calico Pod Network (192.168.0.0/16) is shown here:
sudo kubeadm init --pod-network-cidr=192.168.0.0/16 --apiserver-advertise-address=172.31.20.148
An example based on Flannel Pod Network (10.244.0.0/16) is shown here:
sudo kubeadm init --pod-network-cidr=10.244.0.0/16 --apiserver-advertise-address=172.31.20.148
After the Kubernetes master node has been successfully initialized, one must run the following three instructions before proceeding:
mkdir -p $HOME/.kube sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config sudo chown $(id -u):$(id -g) $HOME/.kube/config
The reverse operation of kubeadm init is kubeadm reset in root mode.
If the Internal IP was specified wrong, the process will stuck. Then you need to reset in root mode. (Not sure whether to clean the .kube directory or the .kube/config file)
Pod network based on Calico (or Flannel)
First download the calico.yaml file: (The following instruction is different from the original video.)
curl https://docs.projectcalico.org/manifests/calico.yaml -O
Use the kubectl apply command.
kubectl apply -f calico.yaml
An alternative Kubernetes networking substrate is Flannel. The URL to get the YAML file is here:
curl https://raw.githubusercontent.com/coreos/flannel/master/Documentation/kube-flannel.yml -O
Then you may just perform kubectl apply command to install the network substrate:
kubectl apply -f kube-flannel.yaml
Kubernetes Dashboard deployed on Master Node
To avoid running the Dashboard on worker nodes, the following command needs to be launched before any worker nodes joining the cluster: (It is important to note that the hyperlink in the original video is not working, the following instruction has been tested as of July 21, 2021)
kubectl apply -f https://raw.githubusercontent.com/kubernetes/dashboard/v2.2.0/aio/deploy/recommended.yaml
A service account must be created to make dashboard available
kubectl create serviceaccount dashboard -n default
Then, it needs to create the dashboard-admin
kubectl create clusterrolebinding dashboard-admin -n default --clusterrole=cluster-admin --serviceaccount=default:dashboard
Then, one needs to get the secret key to be pasted into the web-based interface of Kubernetes Dashboard.
kubectl get secret $(kubectl get serviceaccount dashboard -o jsonpath="{.secrets[0].name}") -o jsonpath="{.data.token}" | base64 --decode
This will generate a long token string, something like this (yours must be different, since it is a dynamically generated):
eyJhbGciOiJSUzI1NiIsImtpZCI6IjRPcQhLUU54QXE1R1dSdnhsZmdMaUxMN0NMdk8wZ1ZEUjhUMVFLZkdVaEkifQ.eyJpc3MiOiJrdWJlcm5ldGVzL3NlcnZpY2VhY2NvdW50Iiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9uYW1lc3BhY2UiOiJkZWZhdWx0Iiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9zZWNyZXQubmFtZSI6ImRhc2hib2FyZC10b2tlbi1wd3FwNSIsImt1YmVybmV0ZXMuaW8vc2VydmljZWFjY291bnQvc2VydmljZS1hY2NvdW50Lm5hbWUiOiJkYXNoYm9hcmQiLCJrdWJlcm5ldGVzLmlvL3NlcnZpY2VhY2NvdW50L3NlcnZpY2UtYWNjb3VudC51aWQiOiJhOGRkMDE0OC1jNDQzLTRkODctYTVjOC01MjJmYmRjZGMxMGQiLCJzdWIiOiJzeXN0ZW06c2VydmljZWFjY291bnQ6ZGVmYXVsdDpkYXNoYm9hcmQifQ.jbjvJgnLCMhsWbdg_7A83PJZ3sLskS9MPTFEZGvUr8eC-I0tosPXgMBkiWgoAgPcFNpYjXWRN3Ia66vYTEHAc0kqSDsZAMbUP48pszBQR0InPk_7tt7kQn3Scx6FEhkRbxVXaiVqYaafxLdOQlbAP_Xz9KOTjq2L-RU0Pxf83FsAITpJbTVmX7oz_trZSEeP1knqjnnKKn3ppYoYWpAwD-FDDXlPwqvHidxLB-Db9rbxkVGSI2yAibtW6mgldStEC9uv64zBpleUVMIw7ys6a9LxuHuVNg29oZrWaIL7qi6rvvQIqvzcneRRUSC2E7pV-Jl7x80leyRL8SRUiKhBsg
To allow access to this dashboard, the following command must be run:
kubectl proxy
Only at the Slave
At the end, run this command to get the worker node, or slave to join the cluster.
sudo kubeadm join 172.31.20.148:6443 --token h5c0bs.nt4vtekd1eb7qupd --discovery-token-ca-cert-hash sha256:088a39a92611a81f18f16bd6ab6b1b438a7015b68392fa8559487c9240c1d1b6
Check Installation Results
Check the final result on master node, all the nodes need to be in Ready status:
$ kubectl get nodes
NAME STATUS ROLES AGE VERSION kmaster Ready control-plane,master 121m v1.21.3 worker1 Ready <none> 36s v1.21.3
To see all the pods, use the following instruction:
kubectl get pods -o wide --all-namespaces
To let a worker leave the cluster, run
kubeadm reset
on the worker node and run
kubectl delete node worker1
Kubernetes on Ubuntu by Edureka
A complete 6 hour lecture series can be found here.
|UWg3ORRRF60}}
Some online installation tutorials
- Video/Installing Kubernetes
- https://blog.alexellis.io/kubernetes-in-10-minutes/
- https://kubernetes.io/docs/setup/production-environment/tools/kubeadm/install-kubeadm/
- https://computingforgeeks.com/deploy-kubernetes-cluster-on-ubuntu-with-kubeadm/
- https://blog.knoldus.com/how-to-install-kubernetes-on-ubuntu-20-04-kubeadm-and-minikube/
References
- ↑ Arundel, John; Domingus, Justin (2019). Cloud Native DevOps with Kubernetes. O'Reilly Media. ISBN 978-1-492-04076-7.