Kubernetes Instructions

K8S installation instruction:
Follow this https://kubernetes.io/docs/tasks/tools/install-kubectl/#install-kubectl-on-windows
(Download kubectl.exe & copy it to C:\windows\system32)
Minikube: https://github.com/kubernetes/minikube/releases (install minikube-installer.exe)
(Download version 1.2.0, from assets section)
Run: minikube start (Minikube is installed in C:\prog~\kubernetes\minikube)
Minikube dashboard
Ctrl+C
Minikube stop

Docker SWARM Vs Kubernetes:

Docker/SWARM:
(+) Light weight (“PS”)
(+) portable
(+) Infra(runtime) services

(-)Too many IPs
(-)Master is for orchestration + Ops
(-)Service availability is not 100% (M+R>SD)
(-)No elasticity
(-)Single Point of Failure (Admin)
(-)Invest on Reachable(s)
(-)JSON/YAML/SHA256/blrfts
(-)Orchestrated

Kubernetes:
(+) Any container orchestration
(+) Any kernel
(+) Any CSP/non-CSP
(+) Open source
(+) Master is only for orchestration
(+) one IP - master IP
(+) UoS(expand)
(+) Orchestration as a service(self heal)
(+) AI on orchestration
(+) scale-> service availability
(+) Manage change (Image change, Port change, Volume change)

In Kubernetes, only dynamic ports (no static port, -p)

Scaling:
Service & Infra 3 axis scaling (x,y,z)
X: Machine/Infra scaling (VMs)
Y: Services(with memory, CPUs) scaling (Containers)
Z: Dependency services scaling

Master & Node
Master:
- Notary?(RSA)
- Authentication
- Valid service?
- Service? Etcd?
- Servicable?
- LB (of node)
Runs in port no-8443
kubectl cluster-info

Node:
VM + Container runtime(default is docker. i.e. dockerd)
Runs in port no-443

K8S-> 1 master + 7 nodes
K3S-> 1 master + 3 nodes

Components of Master:
Services rigitry (by default, etcd). (What service is running)
Services gateway (How to reach the service): Default is Kube DNS, others for ex: nginx, nginx plus
Service discovery (where is service running): Default Kube API server, others: Stack, Ribbon.. Load balancing is done here to find which node
Workflow:
1)Service gateway -> 2)service registry -> 3)Memory & storage provider -> 4)service discovery ->5) Ingress controller

Components of Node:
Agent: Kubelet… Load balancing is is done here via kube proxy for which Pod
Proxy: Kube Proxy (IP tables)
Dockerd
Logs: Fulentd
Workflow:
6)Kubelet->7)kube proxy->8)POD->9)Logs(fluentd)->10)Ingress controller-> 11) service gateway

Namespace: Collection of PODs. Virtual cluster. For logical quota.
3 default namespaces available:
1.kube-system- master details
2.kube-public - Internal cluster details
3.default

CLIs:
Kubectl & Kubeadm(‘minikube’ when using minikube)

CMD1 terminal(For Ops):
Minikube status
Minikube start
Minikube ip
kubectl get nodes
kubectl get namespaces
kubectl get namespaces
kubectl describe namespace kube-public
kubectl create ns demo
kubectl describe namespace demo
kubectl get pods -n kube-system(kube-system is a name of the namespace for ex:)
kubectl get pods -> searched in default namespace
kubectl describe pod etcd-minikube -n kube-system
kubectl get pod etcd-minikube -n kube-system
kubectl get pod etcd-minikube -n kube-system -o json
kubectl get pod etcd-minikube -n kube-system -o yaml
kubectl run nginx-container1 –image=nginx –port=80 -n demo (POD name cannot be set)
kubectl get pods -n demo
kubectl describe pod nginx-container1-7cd546bffb-6627g -n demo
Note: 10.x.x.x is public IP & 172.x.x.x is private IP
kubectl exec -it nginx-container1-7cd546bffb-6627g bash -n demo
(Inside the container/POD machine, you can run commands like ‘hostname’ & exit)
kubectl get pods -n demo -o wide
kubectl run nginx-containererror –image=ngnx –port=80 -n demo (create a pod with a wrong image. POD/Deployment is still created)
kubectl get pods -n demo
kubectl delete pod nginx-containererror-6777d4b675-rm6l8 -n demo
kubectl get pods -n demo (Though we have deleted the container, through self healing another container will be created)
kubectl get deployment -n demo
kubectl get pods -n demo
kubectl delete deployment nginx-containererror -n demo (Now the pod is deleted & container is not re-created)
kubectl get roles -n kube-system
kubectl get clusterroles -n kube-system
kubectl describe deployment nginx-container1 -n demo
kubectl scale deployment nginx-container1 –replicas=5 -n demo
(When you scale via deployment, memory is grabbed first & then pods are created. So, if there are any memory issue, deployment itself will not be created. i.e. no half way creation of PODs)
kubectl get deployment -n demo
kubectl get pods -n demo
kubectl delete pod nginx-container-95d6675d8-g57zw -n demo
kubectl expose deployment nginx-container1 –type=NodePort -n demo
kubectl get services -n demo (This is service descovery: We are reading the service)
kubectl describe service nginx-container1 -n demo
minikube service nginx-container1 -n demo (In real time, we use ‘kubeadm’) (we are Accessing the service here. Here we donot access via PORT, but via service name)
minikube service nginx-container1 -n demo –url=true (Just to get the URL)
kubectl run nginx-cp –image=nginx –port=80 -n demo
kubectl get deployments -n demo
kubectl expose deployment nginx-cp –type=ClusterIP -n demo (Here, type is cluster IP. i.e. accessed via cluster IP)
kubectl get service -n demo
(Take cluster IP & curl the service URL from ‘minikube ssh’)
i.e. Minikube ssh
Sudo -i
curl http://10.97.75.105

kubectl run nginx-mp –image=nginx –port=80 –replicas=3 -n demo
kubectl get deployments -n demo
kubectl expose deployment nginx-mp –type=LoadBalancer -n demo (Here exposed(or accessed via) to master IP)
kubectl get services -n demo
kubectl describe service nginx-mp -n demo
minikube service nginx-mp -n demo
kubectl get services –all-namespaces (List all services across all namespaces)
kubectl delete deployment nginx-mp -n demo (Here, deployment & pod are deleted, but service is not deleted)

UseCases:	Node	LoadBalancer	ClusterIP
Replicas?	Point	Yes	No (not recommended)
LB?	Yes(Node)	Kube DNS + Kubelet	Not there
Access?	Kubelet	Master IP/External IP	Cluster IP
#Estimation?	Node IP	50-60%	10-15%
Benefit	only 25% of services will be here	Stable	Futuristic
Dynamic Port	Pilot	Yes	No
Customize	Yes

Secrets
kubectl get secret
kubectl describe secret default-token-hj8qh (<secret_name)
kubectl create secret generic db-pass-values –from-literal=user=root –from-literal=password=admin
kubectl get secret
kubectl describe secret db-pass-values
kubectl get secret db-pass-values -o json
kubectl create secret generic db-volume –from-file=user.txt –from-file=pwd.txt (create 2 files before hand user.txt & pwd.txt with user name & password resp)
kubectl get secretes

Config Map
kubectl get cm
kubectl get cm -n kube-system
kubectl describe cm kubeadm-config -n kube-system -> this gives property of cluster
kubectl create cm language-key –from-literal=language=”UK_ENGLISH”
Note: Object name cannot have _ (for ex: container name, service name, config map etc)
kubectl get cm
kubectl describe cm language-key

kubectl create namespace rollout
kubectl run nginx-roll –image=nginx –replicas=10 –port=80 -n rollout
kubectl get deployments -n rollout
kubectl expose deployments nginx-roll –type=NodePort -n rollout
minikube service nginx-roll -n rollout
kubectl describe deployment nginx-roll -n rollout

HPA(Horizontal POD availability)
Rollout strategy:

MAX	Surge	Unavailable (25% by default)	Min
10	10+3	3	6

Describe deployment will show these details like Max, Surge etc.

kubectl set image deployment/nginx-roll nginx-roll=bitnami/nginx -n rollout
kubectl rollout status deployment/nginx-roll -n rollout
kubectl get pods -n rollout
kubectl describe pod nginx-roll-566d564589-6qv4j -n rollout
kubectl rollout undo deployment/nginx-roll -n rollout
kubectl rollout history ….. -> to see the rollout history

In Docker perspective, POD is a paused container.
i.e.
Minikube ssh
Sudo -I
Docker ps | grep nginx-cp(containername)

Dashboard:
minikube addons list
minikube addons enable heapster
minikube addons enable ingress
kubectl get pods -n kube-system (Now you can see some more pods are created since we have enabled above 2 addons)
Minikube dashboard (dasboard URL will openup. Here you can scale the deployment etc from the browser)
minikube addons open heapster (it will open up dashboard link where you can see the graphs for Pods)

kubectl run ubuntu-cont –image=ubuntu -> POD without a service
kubectl get pods -> this will show the POD status as ContainerCreating->CrashLoopBackOff->Completed->ContainerCreating.. It will keep on restarting.. Because default restart policy of container in k8s is ‘restart_always’
kubectl describe pod ubuntu-cont-db87d5bc4-tbck7

Create PODs from YML with no restart
YML files asre available at: https://github.com/vinaydhegde/DockerStuff/tree/master/K8S
kubectl get services -n kube-system -> Get Kube DNS IP

Kubectl get pods
Kubectl logs command-demo
Kubectl get secrets
kubectl create -f pod1.yml
kubectl create -f pod2.yml
Kubectl exec -it secret-envirs-test-pod bash
Echo $SECRET_USERNAME
Echo $SECRET_PASSWORD
Kubectl get secrets

kubectl create -f pod3.yml
kubectl exec -it mypod bash
Cd /etc/vols
Ls
Cat user.txt
Cat pwd.txt

kubectl describe ns default -> It says no resource quota for default namespace
In the K8sDay5 git repo, refer file called ‘quota’
Set the quota now;
Kubectl create -f quota.txt
Kubectl get quota
Kubectl get resourcequota pod-demo -o yml
Kubectl run test-demo –image=nginx –replicas=3 –port=80
Kubectl get deployments
Kubectl delete quota pod-demo
Kubectl get deployments

Communication between PODs:
For example:
NS1: has 2 pods (POD1 has 2 containers) & NS2 has 1 pod
1. Container1 -> container 2 in POD1: via volume
2. POd1->POD2: via service
3. POD2->POD3: via namespace.service

Volume types:
1. Persistent volume(Master volume)
2. POD volume:(Empty dir)
3. Local volume(node volume)

Create 2 containers in one POD:
kubectl create -f pod5.yml
kubectl get pods
kubectl describe pod two-containers (2 containers are created.. Ngin cont is running & ubuntu is terminated)
kubectl -exec -it two-containers -c nginx-container bash

hostname (POD name is the hostname)
Exit

Ingress tutorial:
kubectl get pods -n kube-system
Create a POD & service:
kubectl create -f pod6.yml
kubectl create -f pod7.yml
kubectl create -f pod8.yml
kubectl get pods
kubectl get services -> note it’s a cluster ip service
We already have ingreass controller, we need to create ingress router
kubectl get ing
http:///crap http:///apple http:///verizon

Minikube stop

CMD2(for kubernetes Admin):
Minikube dashboard

CMD3(for admin):
Minikube ssh
Sudo -i