Managing Kubernetes with Rancher

KSS CON'19

Adam Płaczek

Problem solving history

  • Virtual machines solved problems with hardware- one  OS running on one HW

 

  • Containers solved developers problems with Virtual Machines- light image, slow boot, big OS overhead, portability ....

 

  • Orchestrators solved operations problems with containers- managing multiple Docker Engines and application lifetime. 

 

  • Kubernetes platforms solved everyones problems with orchestrators- Difficult deployment and management

 

Container app development

Deployment and  testing done on a laptop

 

No worries about HA

No worries about deployment

No worries about scaling up

Why containers need orchestration ?

Provisioning and deployment of multiple containers

Ensuring containers availability

Scaling up or down

Load Balancing

Service Discovery

Secrets management

Health monitoring

Allocation of resources between containers

External exposure of services to the outside world

...............

Container Orchestrators

Kubernetes ( aka K8s )​

 

Docker swarm

 

Mesos + Marathon

 

Nomad

Kubernetes vs Nomad

https://www.nomadproject.io/intro/vs/kubernetes.html

Out of the box K8s provides collection of collaborating services which provide full container orchestration functionality.

Widely adopted -all major cloud vendors provide managed K8s clusters GKE, EKS, AKS

 

Nomad focus is workload scheduling and cluster management without additional stuff like Service Discovery, Secret Management etc.

Managed k8s comparision

KUBECTL

command line too to interact with K8 cluster

https://kubernetes.io/docs/tasks/tools/install-kubectl/

Concepts

  • Pod- smallest deployable unit

 

 

  • Service- DNS entry grouping multiple pods

 

 

  • Deployment- Create and update multiple pods

Deployment

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: super-saiyan-app-1.0
spec:
  replicas: 2
  template:
    metadata:
      labels:
        name: super-saiyan-app
        version: "1.0"
    spec:
      containers: 
        - name: super-saiyan-app
          image: brudnyhenry/super-saiyan-app:v1
          ports:
            - name: http
              containerPort: 8080
kubectl apply -f deployment.yaml

Service

apiVersion: v1
kind: Service
metadata: 
  name: super-saiyan-app
  labels: 
    name: super-saiyan-app
spec:
  ports:
    - name: http
      protocol: "TCP"
      port: 8080
      targetPort: 8080
  selector: 
    name: super-saiyan-app
    version: "1.10"
  type: LoadBalancer
kubectl apply -f service.yaml
kubectl expose deployment super-saiyan-app --type=LoadBalancer --name=my-service

Port Forward

kubectl port-forward deployment/demoapp 8080:8080  


curl localhost:8080

Access your deployments without exposing them to external network

 

https://kubernetes.io/docs/tasks/access-application-cluster/port-forward-access-application-cluster/

Rancher

multi-cluster Kubernetes manager

  • One level of abstraction above Kubernetes
  • User friendly UI
  • Well tested app catalog
  • Authentication across multiple clusters
  • Connect to existing K8s or create new cluster with RKE

Rancher

Run Rancher locally

docker run -p 80:80 -p 443:443   -v /opt/rancher:/var/lib/rancher   rancher/rancher:latest

Try Kubernetes locally

https://github.com/kubernetes/minikube

https://k3s.io/

Demo

Create Kubernetes cluster in AWS EC2

Deploy custom app

Upgrade

Create Rancher server  AWS EC2

App Monitoring

Made with Slides.com