Kubernetes Topics

Resource Management

Hands on lab to deep dive into the following concepts
- API Object
- K8S Cluster
- Pod
- ReplicaSet
- Service
- Deployment
Features of k8s
- Auto Scalable
- High Available

Network

Configuration
Service
Service Discovery
DNS

Cluster Management and Monitoring

Cluster management
- Node management
- Resource Label
- Namespace management
- Resource management
- High availability management
Log collection and analysis
- Solution
  - Heapster
  - Prometheus
  - ELK
  - Splunk
- Logs and metrics
  - System event
  - Container log
  - Service log
Troubleshooting

Storage

Volume
Persistent Volume
ConfigMap
Storage Class

Security

Store Confidential Information
Securing Kubernetes Clusters
Authorizing/RBAC/RoleBindings/ClusterRoleBindings

Domain Application

Microservice
Bigdata
- Spark
Deep Learning
- Tensorflow
Serverless application
CI/CD

Compare with Other Orchestration Platform

Introduction to docker swarm
Comparing docker swarm with K8S
Introduction to Mesos and Marathon
Comparing Marathon with K8S

Resource Management

Hands on lab to deep dive into the following concepts
- API Object
- K8S Cluster
- Pod
- ReplicaSet
- Service
- Deployment
Features of k8s
- Auto Scalable
- High Available

Recap of Kubernetes 101

Kubernetes Cluster

k8s cluster

##Kubernetes Concept

Pod
ReplicaSet
Service
Volume
Namespace

How to get started

Minikube
Docker Native Support (Since 17.12.0)
Google Cloud Kubernetes Engine

Kubernetes Resource Object

Persistent entities in Kubernetes system
Kubernetes use these entities to represent the state of cluster
- What containerized applications are running on which nodes
- Resources avaiable to those applications
- The policies around how application behave, restart/upgrade/fault-tolerant
Manipulated via Kubernetes API
- GO Client
- Python Client

Category	Name
Resource Object	Pod、ReplicaSet、ReplicationController、Deployment、StatefulSet、DaemonSet、Job、CronJob、HorizontalPodAutoscaling
Configuration Object	Node、Namespace、Service、Secret、ConfigMap、Ingress、Label、ThirdPartyResource、 ServiceAccount
Storage Object	Volume、Persistent Volume
Strategy Object	SecurityContext、ResourceQuota、LimitRange

Application Type

Application Type	Resource Object
Long-running	Deployment
Batch	Job/CronJob
Node-daemon	DaemonSet
Stateful Application	StatefulSet

What is a Pod?

Minimal unit for Kubermetes
- Bricks to build a house
- Container for Docker, Swarm
- Running process for a cluster
Encapsulates one or more containers
- Unique network IP
- Attach storage resources
- Decide how containers should run
Everything is tightly coupled

Pod Lifecycle

Create a Pod

kubectl run db --image mongo

kubectl get pods

eval $(minikube docker-env)

docker ps | grep 'mongo'

kubectl delete deployment db

Imperative VS Declarative

Yet this is not the best way to run Pods.

Benefits of declarative approach:

Idempotency
Predictability
Traceability

Declarative Way to Create a Pod

cat pod/db.yml

apiVersion: v1
kind: Pod
metadata:
  name: db
  labels:
    type: db
    vendor: MongoLabs
spec:
  containers:
  - name: db
    image: mongo:3.3
    command: ["mongod"]
    args: ["--rest", "--httpinterface"]

Kind
- What you want to do? Create a Pod.
apiVersion
- Which version do you want to use?
metadata
- Addional data for the object. Does not affect the behaviour of object.
spec
- The desired state of the object

kubectl create -f pod/db.yml

Get pods information:

kubectl get pods

kubectl get pods -o wide

Or you can use -o json or -o yaml to define the output format.

Get pod logs:

kubectl logs db

Execute command in the pod conainer:

kubectl exec db ps aux

Get detail information of the specified resource:

kubectl describe pod db

Better way to describe , no need to remember the resource name:

kubectl describe -f pod/db.yml

Involving Components and Process

API Server
- Most of the coordination in Kubernetes consists of a component writing to the API Server resource that another component is watching. The second component will then react to changes almost immediately.
Scheduler
- Watch for unassigned pods and assign to nodes with available resources matching pod requirements.
Kubelet
- Make sure the assigned pods are running on the node.

Running Multiple Containers In A Single Pod

cat pod/go-demo-2.yml

Output:

apiVersion: v1
kind: Pod
metadata:
  name: go-demo-2
  labels:
    type: stack
spec:
  containers:
  - name: db
    image: mongo:3.3
  - name: api
    image: vfarcic/go-demo-2
    env:
    - name: DB
      value: localhost

Create and get detail of the pods:

kubectl create -f pod/go-demo-2.yml

kubectl get -f pod/go-demo-2.yml

kubectl get -f pod/go-demo-2.yml -o jsonpath="{.spec.containers[*].name}"

Execute command in the db container

kubectl exec -it -c db go-demo-2 ps aux

List logs of the db container

kubectl logs go-demo-2 -c db

Remove the Pod:

kubectl delete -f pod/go-demo-2.yml

Monitoring Health

cat pod/go-demo-2-health.yml

Output:

apiVersion: v1
kind: Pod
metadata:
  name: go-demo-2
  labels:
    type: stack
spec:
  containers:
  - name: db
    image: mongo:3.3
  - name: api
    image: vfarcic/go-demo-2
    env:
    - name: DB
      value: localhost
    livenessProbe:
      httpGet:
        path: /this/path/does/not/exist
        port: 8080
      initialDelaySeconds: 5
      timeoutSeconds: 2 # Defaults to 1
      periodSeconds: 5 # Defaults to 10
      failureThreshold: 1 # Defaults to 3

kubectl create -f pod/go-demo-2-health.yml

kubectl describe -f pod/go-demo-2-health.yml

Scaling Pods

cat rs/go-demo-2.yml

Output:

apiVersion: apps/v1
kind: ReplicaSet
metadata:
  name: go-demo-2
spec:
  replicas: 2
  selector:
    matchLabels:
      type: backend
      service: go-demo-2
  template:
    metadata:
      labels:
        type: backend
        service: go-demo-2
        db: mongo
        language: go
    spec:
      containers:
      - name: db
        image: mongo:3.3
      - name: api
        image: vfarcic/go-demo-2
        env:
        - name: DB
          value: localhost
        livenessProbe:
          httpGet:
            path: /demo/hello
            port: 8080

Create the ReplicaSet:

kubectl create -f rs/go-demo-2.yml

Get ReplicaSet:

kubectl get rs

Get Pods:

kubectl get pods --show-labels

Process of creating a ReplicaSet:

Cluster View:

Delete the ReplicaSet:

kubectl delete -f rs/go-demo-2.yml --cascade=false

ReplicaSet is removed but pods are not:

kubectl get rs
kubectl get pods

Create the ReplicaSet again with --save-config:

kubectl create -f rs/go-demo-2.yml --save-config

kubectl get pods

cat rs/go-demo-2-scaled.yml

kubectl apply -f rs/go-demo-2-scaled.yml

Assign the pod name:

POD_NAME=$(kubectl get pods -o name | tail -1)

Delete the pod:

kubectl delete $POD_NAME

See what happens:

kubectl get pods

Reget the pod name:

POD_NAME=$(kubectl get pods -o name | tail -1)

Remove the service label:

kubectl label $POD_NAME service-

Describe the pod and add the label back”

kubectl describe $POD_NAME
kubectl get pods
kubectl label $POD_NAME service=go-demo-2
kubectl get pods

Kubernetes Session One

Kubernetes Topics

Resource Management

Network

Cluster Management and Monitoring

Storage

Security

Domain Application

Compare with Other Orchestration Platform

Resource Management

Recap of Kubernetes 101

Kubernetes Cluster

How to get started

Kubernetes Resource Object

Application Type

What is a Pod?

Pod Lifecycle

Create a Pod

Imperative VS Declarative

Declarative Way to Create a Pod

Involving Components and Process

Running Multiple Containers In A Single Pod

Monitoring Health

Scaling Pods

References