Skip to content

Latest commit

 

History

History
123 lines (88 loc) · 3.47 KB

setting_up_the_cluster.md

File metadata and controls

123 lines (88 loc) · 3.47 KB
Raw

Setting up the workshop cluster

You should definitely do a test run before your workshop. This document will guide you through the process of running the cluster before, during and after your workshop.

🔥 Make sure to have alternative cloud vendors available in case of unexpected downtime or maintenance!

Weeks before the workshop

Have accounts and the corresponding API keys for the cloud services ready you are going to use. Acquire API access tokens and place them as environment variables on your machine:

1. Configure the trainers and participants in inventors/group_vars/all/workshop.yml. Contact the participants prior and collect their SSH public keys.

workshop:
  trainers:
    - name: "Torsten Köster"
      ssh_key: "ssh-rsa AAAAB3..."
    - name: "Nikolaus Winter"
      ssh_key: "ssh-rsa AAAAB3...."
  participants:
    - name: alice
      ssh_key: "ssh-rsa AAAAB3...."
    - name: bob
      ssh_key: "ssh-rsa AAAAB3...."
    - name: charlie
...

Adjust the Terraform instance_count variable in variables.tf according to the workshop participants:

variable "instance_count" {
  default = "1"
}

2. Prepare your local machine for cluster setup:

brew install terraform@0.12 ansible figlet
brew link terraform@0.12 --force

3. Prepare remote shared state

If you're not the only trainer, it makes sense to share Terraform state via their free backend service. You could use your own backend (e.g. a Postgres database as well).

alt

In order to use the backend, you need to sign up to their service and join (or share) a workspace. Configure access to the backend in your ~/.terraformrc:

credentials "app.terraform.io" {
  token = "..."
}

4. Download the Terraform plugins needed to spin up the cloud infrastructure:

terraform init

The day before the workshop

Set up the server and certificate infrastructure. The Terraform plan task checks what infrastructure items are already existing and which needs to be created. The apply task executes the plan.

$ terraform plan
...
Plan: 9 to add, 0 to change, 0 to destroy.
$ terraform apply

Now a bunch of servers have spun up. They have been assigned random pet names as hostnames. The corresponding Let's Encrypt TLS certificates are stored in roles/boostrap/tls/files for direct access out of Ansible.

Now start provisioning the servers using Ansible. First, a general bootstrap is applied (base os, users, Docker), then the workshop specific roles are applied:

ansible-playbook bootstrap.yml
ansible-playbook workshop.yml

alt

Workshop day

Assign each participant an individual server. Keys are distributed to all machines, so participants could switch (or share) servers.

ansible-playbook server-2-participant.yml

If you have some late arrivals, adjust the server count in the Terraform file and re-run the steps above. Only the one additional server will be created and provisioned.

Terminating misbehaving instances

If any instance is misbehaving, terminate and recreate it.

terraform taint <SERVERNAME>
terraform plan
terratorm apply

🐄 Think cattle not pets!

The day after the workshop

Shut down the whole infrastructure using Terraform:

terraform destroy