Skip to content

Latest commit

 

History

History
152 lines (133 loc) · 8.66 KB

README.md

File metadata and controls

152 lines (133 loc) · 8.66 KB
Raw

rancher-kubernetes-workshop

Supporting code/diagrams for my devopsdays SLC workshop on running Kubernetes with Rancher given on May 16, 2017.

This repo has all the code necessary to create a k8s cluster in a Rancher environment with logging to Sumologic, DNS through AWS route53, and alerting through VictorOps. This is configured to match what we do at 3dsim. You will probably want to modify many things to match your needs.

In this workshop you will learn how to deploy a production grade, highly available Kubernetes cluster in AWS using Rancher and Ansible. Almost all steps will be automated so that the finished product truly embodies Infrastructure as Code. By the end of the workshop, we will have a fully running kubernetes cluster, authentication/authorization in place for cluster management, monitoring and alerting using Prometheus, graphs of server performance using Grafana, rolling deploys of application code, centralized logging, and more.

Read more about the various tools/platforms: https://kubernetes.io http://rancher.com https://www.ansible.com

Requirements: You will need to have an AWS account set up and able to provision new servers. You will also need to have git installed to download the workshop material. A recent version of Docker is also required (https://www.docker.com). If you don't have an AWS account, you can sign up for one here: https://portal.aws.amazon.com/gp/aws/developer/registration/index.html

Prerequisites

curl -LO https://storage.googleapis.com/kubernetes-release/release/v1.5.4/bin/darwin/amd64/kubectl
chmod +x ./kubectl
sudo mv ./kubectl /usr/local/bin/kubectl

Overview of workshop

  • Create Rancher server
  • Setup auth for rancher
  • Create Kubernetes cluster
  • Deploy applications to cluster
  • Update applications in cluster
  • Setup prometheus, grafana, and alertmanager for monitoring/alerting
  • Demo centralized logging
  • Auto scale your cluster
  • (Extra credit 1) Setup monitoring of application using prometheus
  • (Extra credit 2) Setup centralized logging using ELK stack

Create Rancher server (Optional)

  • Run the following command in the root of this repo to create a container called ansible
docker run --restart=always -d -v `echo ~`/.ssh:/root/.ssh -e AWS_ACCESS_KEY_ID=`echo $AWS_ACCESS_KEY_ID` -e AWS_SECRET_ACCESS_KEY=`echo $AWS_SECRET_ACCESS_KEY` --name ansible ryanwalls/ansible-aws:v2.2.1.0-1-k8s tail -f /dev/null
  • (Optional) Change password in rancher-server.yml lines 67 and 92
  • Create an ssh key pair in AWS and fill in keyName and ansible_ssh_private_key_file variables in group_vars/all/main.yml with name and location of an ssh key you've created in AWS and downloaded to your machine. Use a relative path. Download your key to ~/.ssh on your desktop. See http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-key-pairs.html#having-ec2-create-your-key-pair. Ideally this would be automated.
  • If you haven't already, you'll need to setup an SSL cert. See instructions in "Prerequisites" section.
  • Fill in default_ssl_cert_arn in group_vars/all/main.yml with the ARN of your SSL cert
  • Fill in default_hosted_zone in group_vars/all/main.yml with the domain of your route53 hosted zone (e.g. ryanwalls.com)
  • Let's create the rancher server now. Run the following from the root of this project.
docker cp . ansible:ansible && docker exec -it ansible ansible-playbook -i inventory/ -vvvv --extra-vars "env=slc" rancher-server.yml

If you have trouble creating own server

Rancher has created https://try.rancher.com which has a Rancher server (non production use) setup for you.

Setup auth for rancher (optional, skip if using try.rancher.com)

  • From Rancher click... Admin -> Access Control
  • Follow instructions to setup Github authentication

Create Kubernetes cluster

  • Fill in rancher_api variables in group_vars/all/main.yml.
  • Fill in default_ssl_cert_arn in group_vars/all/main.yml with the ARN of your SSL cert if you didn't do it earlier.
  • Fill in default_vpc_id in group_vars/all/main.yml with the VPC id of your default vpc. The default VPC can be found in your ec2 dashboard on the right hand side.
  • Fill in default_hosted_zone in group_vars/all/main.yml with the domain of your route53 hosted zone
  • Go to Rancher -> Environment -> Manage Environments -> Click Pencil next to Kubernetes at bottom of page -> Edit Config -> Select dropdown
    • Change cloud provider to AWS
    • Change "Plane Isolation" to "required"
  • Create environment in Rancher with Kubernetes as the orchestration tool
  • Create an environment specific API key (API -> Keys -> Advanced Options -> Add Environment API Key)
  • Copy API key into group_vars/all with the matching environment.
  • Also update the environment id in group_vars/all. Id is in the URL for rancher e.g. https://rancher.3dsim.com/env/1a644/api/keys, so 1a644 is the id.
  • Get the subnets for your AWS account and update the alb_subnets variable in roles/alb/defaults/main.yml. To get your subnets, navigate to https://us-west-2.console.aws.amazon.com/vpc/home?region=us-west-2#subnets:
  • Run kubernetes-cluster.yml with from the root of this repo. e.g.
docker cp . ansible:ansible && docker exec -it ansible ansible-playbook -i inventory/ -vvvv --extra-vars "env=slc" kubernetes-cluster.yml
  • Once kubernetes environment is up and running, generate a key for kubernetes. In Rancher navigate to Kubernetes -> CLI -> Generate Config. Put username and password for kubernetes in group_vars/all

Deploy applications to cluster

  • Setup default service route (used for healthchecks) kubernetes-default-service.yml
  • To deploy an application look at kubernetes-example1-deploy.yml. Make sure and add the ingress vars in roles/kubernetes_ingress/vars/main.yml to match the application you're deploying.
  • Then run:
docker cp . ansible:ansible && docker exec -it ansible ansible-playbook -i inventory/ -vvvv --extra-vars "env=slc" kubernetes-example1-deploy.yml

Update applications in cluster

  • Modify kubernetes-example1-deploy.yml
  • Redeploy it

Setup prometheus, grafana, and alertmanager for monitoring/alerting

  • Clone https://github.com/3DSIM/kube-prometheus.
  • Make sure you have setup your kube config file for the right environment. In rancher, Kubernetes -> CLI -> Generate config. Put config in ~/.kube/config.
    • If you have multiple contexts (See sample kubeconfig below), you can switch between them by running kubectl config set current-context <context name>
  • Execute ./hack/cluster-monitoring/deploy
  • (Optional, for Victorops alerts) Come back to this project and setup victorops routing keys in group_vars/all/main.yml
  • (Optional, for Victorops alerts) Fill in victorops_api_key in group_vars/all/main.yml with the key for your victorops account.
  • Run kubernetes-monitoring.yml

Demo centralized logging

Auto scale your cluster

  • Setup auto scaling by running kubernetes-cluster-autoscaler.yml

Sample kubeconfig

Sample kubeconfig:

apiVersion: v1
kind: Config
clusters:
- cluster:
    api-version: v1
    server: "https://rancher.3dsim.com/r/projects/1a644/kubernetes"
  name: "qa"
- cluster:
    api-version: v1
    server: "https://rancher.3dsim.com/r/projects/1a923/kubernetes"
  name: "prod"
contexts:
- context:
    cluster: "qa"
    user: "qa"
  name: "qa"
- context:
    cluster: "prod"
    user: "prod"
  name: "prod"
current-context: "qa"
users:
- name: "qa"
  user:
    username: "<Your QA username>"
    password: "<Your QA secret>"
- name: "prod"
  user:
    username: "<Your Prod username>"
    password: "<Your Prod secret>"