IDE

Download

• Download Virtual Box.
• Download Vagrant.
• Download Putty.
• Download SuperPutty.
• Download Kubernetes Starter Kit.

Run Kubernetes

• Open ~\_Lecture_k8s_starter.kit-main\ch1\1.2\k8s-min-5GiB on commend
• Enter vagrant up

• Open Virtual Box to check 4 machines

Run Kubernetes on Terminal

• Open SuperPutty - [File] - [Import Sessions] - Import ~\_Lecture_k8s_starter.kit-main\ch1\1.3\Sessions.Xml

Setting

• To easy exit from the virtual machine, Open Virtual Box - [Environment Settings] - [Input] - [Virtual Machine] - Click Host Key and Press Ctrl + Alt
• To change font size of SuperPutty, Open SuperPutty and Click m-k8s from Sessions - Right click the session name in Commands - [Change Settings] - [Window] - [Apperance]

• If you want to change Putty colour, copy and paste _puttycolor.js and k8s(github_SysNet4Admin) Terminal,reg in ~\_Lecture_k8s_starter.kit-main\ch8\012 to your putty folder.

Pod

• Pod is union of containers.
• One pod has one or several containers to do one specific work.

Release Pod with NGINX

• Enter kubectl run nginx --image=nginx
• Enter Kubectl get pod and wait until Status is changed to Running
• Enter Kubectl get pod -o wode and copy nginx IP
• Enter curl [nginx IP]

Connect Pod from Outside

Service

• We use a service to connect a pod from Outside.
• At first, service will be connected with a NodePort.
• Then the NodePort communicates with the node.
• As a result, the service finds the pod in the node.

Expose Service

• Enter kubectl expose pod nginx --type=NodePort --port=80 to expose a service with a port.
• Enter kubectl get service and check your port number.

• Enter kubectl get nodes -o wide and check INTERNAL-IP.

• Check one of IP is connected from outside like commands or web browser.

Deployment

• Union of Pods

Create Deployment

• Enter kubectl create deployment deploy-nginx --image=nginx.

Scale Deployment

• Enter kubectl scale deployment deploy-nginx --replicas=3 because default replicas set is 1.

Expose Deployment

• This way is not the best way to connect with pods because you have to share your Internal IP.
• Enter kubectl expose deployment deploy-nginx --type=NodePort --port=80.

Load Balancer

• We use Load Balancer to expose the Deployment with External IP.
• Enter kubectl apply -f ~/_Lecture_k8s_starter.kit/ch2/2.4/metallb.yaml to install MetaLib.

• Enter kubectl create deployment chk-hn --image=sysnet4admin/chk-hn.
• Enter kubectl scale deployment chk-hn --replicas=3.

• Enter kubectl expose deployment chk-hn --type=LoadBalancer --port=80.

• Check your External IP with web browser

Delete Deployment, Service, Pod and MetaLib

• Enter kubectl delete deployment [Your Deployment Name].

• Enter kubectl delete service [Your Service Name].

• Enter kubectl delete Pod [Your Pod Name].

• Enter kubectl delete -f ~/_Lecture_k8s_starter.kit/ch2/2.4/metallb.yaml.

Components

Native components

• Enter kubectl get pods -n kube-system.

Kubernetes Cloud Service

• EKS : Elastic Kubernetes Service from AWS
• AKS : Azure Kubernetes Service
• GKE : Google Kubernetes Engine

Issues

About Deleteing

• If your deleted pod was a real pod, then you cannot rewind.
• But if your deleted pod was in a deployment, then kubernetes rewind the pod automatically.
• If you delete a pod in mater node, kubernetes recreate it automatically.

Pods

• At first, you have to make pods and a deployment for practice with kubectl apply -f ~/_Lecture_k8s_starter.kit/ch4/4.1/.
• Now, you have 3 deployment’s pods and 1 just pod.

• When you delete del-pod with kubectl delete pod [Your Pod], three pods will be remained.
• And del-pod is removed eternally.

• When you delete a pod which is in the deployment with kubectl delete pod [Your Delpoyment's Pod], three pods will be remained.
• Your deployment will rewind your deleted pod automatically.

Deployment

• When you want to delete a pod which is in the deployment, enter kubectl delete deployment [Your Deployment].
• Your deployment will remove your all pods in that deployment.

Scheduler

• To delete Scheduler in master node, enter kubectl delete pod kube-scheduler-m-k8s -n kube-system.
• Then you can see that your kubernetes create new scheduler immediately.

About Termination

• If you terminate a worker node, kubernetes seperate pods in that worker node to others.

Kubelet

• At first, terminate first worker node with systemctl stop kubelet

• You have to make pods for practice with kubectl apply -f ~/_Lecture_k8s_starter.kit/ch4/4.1/del-deploy.yaml.
• Now, you have 3 deployment’s pods unfairly.

• In master node, you can also practice terminating Kubelet with systemctl stop kubelet and kubectl delete pod kube-scheduler-m-k8s -n kube-system
• Now your scheduler in master node is always Terminating, because your master node kubelet is stopped.

• You can check, that kubelet in master node is still working well, with kubectl create deployment nginx --image=nginx

• You can check, that scheduler in master node is still working well, with kubectl scale deployment nginx --replicas=3

• By the way, to restart our kubelet and scheduler, enter systemctl start kubelet

ContainerD

• At first, terminate containerD in first worker node with systemctl stop containerd

• You need to scale pods to 6 with kubectl scale deployment del-deploy --replicas=6.
• Now, you have 6 deployment’s pods and those pods are working in worker node 2 and 3, not 1.

• To restart containerD, enter systemctl start containerd.

• You have to rescale to see the worker node 1 in pod list with kubectl scale deployment del-deploy --replicas=9.

• In master node, you can also practice terminating ContainerD with systemctl stop containerd

• You can check, that ContainerD in master node is still working well, with kubectl create deployment nginx --image=nginx and kubectl delete deployment nginx

• By the way, to restart our ContainerD and scheduler, enter systemctl start containerd and

Docker

• In worker node #1, you will stop ContainerD and Docker with systemctl stop containerd and systemctl stop docker.
• Then your new scaled deployment is not working in worker node #1.

• When you stoped worker node #1 more than 5 minutes, your pods in worker node #1 are terminated.
• But, those are not deleted, because there is no ContainerD.

Object

• Most objects in kubernetes has the status and spec.
• Pod : union of container
• Service : Connection Pod with outside
• Namespace : Place to publish pods, services and delpoyments
• Volume : Place to save eternal data, like pod

Edit Status

• To edit status of deployment, enter kubectl edit deployment del-deploy

Apply Volume

• At first, create symlink to make log file about volume with ~/_Lecture_k8s_starter.kit/ch5/5.2/nfs-exporter.sh log, cat /etc/exports and cat ~/_Lecture_k8s_starter.kit/ch5/5.2/dpy-chk-log.yaml.

• Now, create deployments with kubectl apply -f ~/_Lecture_k8s_starter.kit/ch5/5.2/dpy-chk-log.yaml.

• Now, create deployments with kubectl apply -f ~/_Lecture_k8s_starter.kit/ch5/5.2/dpy-chk-log.yaml.

• To check your log file, first of all, you have to execute a deployment with curl [Your Deployment IP].
• Then you will access this deployment with kubectl exec dpy-chk-log-655668ffb8-jl9fr -it -- /bin/bash.
• And you can access the log file with cat audit/audit_dpy-chk-log-655668ffb8-jl9fr.log.
• You can see your execution after that command line.

• This log file data is eternally saved, even when we deleted deployments.

Short Cut

Alias

• With alias, we can make short cut in kubernetes.

Master Node

• You can see all line of master node code with cat ~/_Lecture_k8s_starter.kit/ch1/1.2/k8s-min-5GiB/master_node.sh
• In this master node, for example, you can use k like kubectl

• Now, we will use keq as kubectl exec [Your Pod Number].
• You can see the code with cat ~/_Lecture_k8s_starter.kit/ch6/6.1/k8s_rc.sh

• To run this file, enter ` ~/_Lecture_k8s_starter.kit/ch6/6.1/k8s_rc.sh`.
• When you type keq, you can see the selection.
• The default command line environment is shell, but if you want, you can change to bash with /bin/bash.

Upgrade

• Before upgrading, our master node version is 1.25.0.

Master node

• To upgrade your kubernetes, you should upgrade kubeadm with yum upgrade kubeadm-[Your Upgrade Version] -y, cluster with kubeadm upgrade apply [Your Upgrade Version] -y, kubelet with yum upgrade kubelet-[Your Upgrade Version] -y.
• Now, you need to restart kubelet with systemctl restart kubelet and systemctl daemon-reload.

Worker node

• In worker node, you should upgrade kubeadm with yum upgrade kubeadm-[Your Upgrade Version] -y, cluster with kubeadm upgrade node, kubelet with yum upgrade kubelet-[Your Upgrade Version] -y
• When you have different cluster between master node and worker node, you can use kubeadm upgrade node and kubectl -n kube-system get cm kubeadm-config -o yaml.

• Now, you need to restart kubelet with systemctl restart kubelet and systemctl daemon-reload.
• You can see upgrade result in master node with kubectl get nodes.

Automatic Upgrade with Ansible

• You can check the installer code for ansible with cd _Lecture_k8s_starter.kit/ch8/009 and vi ansible-installer.sh
• yum install ansible-2.9.27-1.el7 -y will install ansible.
• cat <<EOF > /etc/ansible/hosts [Master] 192.168.1.10 [Workers] 192.168.1.[101:103] EOF describes upgrading hosts.
• known_host makes automatic authorization in ssh without password.

• You should run this file with ./ansible-installer.sh,

• There is an upgrade yaml code and you can check this with vi k8s-upgrade.yml.
• hosts means host for upgrading.
• In task, you can see yum and this describes we want to upgrade kubeadm, cluster and kubelet.
• ansible.builtin.systemd shows, we want to restart our kubelet and reload daemon.

• After you check your upgrade version, run this upgrade code with ansible-playbook k8s-upgrade.yml.

DNS

• Reference : Kubernetes DNS Query
• Use that manifest to create a Pod with kubectl apply -f https://k8s.io/examples/admin/dns/dnsutils.yaml.

• Take a look inside the resolv.conf file with kubectl exec -ti dnsutils -- cat /etc/resolv.conf

• Use the kubectl get pods command to verify that the DNS pod is running with kubectl get pods --namespace=kube-system -l k8s-app=kube-dns.
• The command line will show pods, that their label is kube-dns and they are belong to namespace kube-system.

• Use the kubectl logs command to see logs for the DNS containers with kubectl logs --namespace=kube-system -l k8s-app=kube-dns.

etcd

• In mater node, there is Api server and etcd.
• etcd communicates with Api server for cluster status.
• You can install etcd with ~/_Lecture_k8s_starter.kit/ch8/003/install_etcdctl.sh.

• Create deployments with k apply -f _Lecture_k8s_starter.kit/ch8/003/sysnet4admin.yaml
• Sync Api server and etcd with ETCDCTL_API=3 etcdctl --endpoints=https://[192.168.1.10]:2379 --cacert=/etc/kubernetes/pki/etcd/ca.crt --cert=/etc/kubernetes/pki/etcd/server.crt --key=/etc/kubernetes/pki/etcd/server.key get --prefix=true "" > /tmp/prefix
• Then you can see two command line, cat /tmp/prefix | nl | tail and cat /tmp/prefix | nl | grep -i 'pod":"sysnet4admin'

• When you scaled deployments, you should sync again.

• When you delete deployments, you should sync again, too.

Application

• Every application use a specific namespace, because it’s easier to delete.

Polaris

• Reference : Fairwinds
• Before practice, you need 6 normal deployments and 1 specific deployment, which belong to metallb-system namespace.

• Then, you need to download helm in your repository with belows.

curl -fsSL -o get_helm.sh https://raw.githubusercontent.com/helm/helm/master/scripts/get-helm-3
chmod 700 get_helm.sh
./get_helm.sh

• Now, you can download polaris with belows.

helm repo add fairwinds-stable https://charts.fairwinds.com/stable
helm upgrade --install polaris fairwinds-stable/polaris --namespace polaris --create-namespace

• Finally, you will expose polaris dashboard with LoadBalancer in port 8080 with k expose -n polaris deployment polaris-dashboard --type=LoadBalancer --name=polaris --port=8080.

• Use http://192.168.1.11:8080/ in your web browser to see your dashboard.

Chaos Mesh

• Reference : Chaos Mesh
• Download with curl -sSL https://mirrors.chaos-mesh.org/v2.5.1/install.sh | bash.

• Use k get svc -n chaos-mesh to check your node port number.

• And you can access Chaos Mesh dashboard in your web browser on [One of Your Node IP]:[Chaos Mesh Node Port]
• Chaos Mesh is used as a lab with experiment tab.