Skip to content

Latest commit

 

History

History
 
 

openstack

README.md

Flatcar Provisioning Automation for OpenStack

This repository provides tools to automate Flatcar provisioning on OpenStack using Terraform.

Features

  • Minimal configuration required (demo deployment works with default settings w/o any customisation, just run terraform apply!).
  • Deploy one or multiple servers.
  • Per-server custom configuration via separate container linux config files.

Prerequisites

  1. OpenStack credentials: username, tenant name, password, authentication URL, region.

Disclaimer

This Terraform code is tested against a DevStack instance, for actual OpenStack deployment the "network" attribute might need to be modified.

HowTo

This will create a server in 'RegionOne' using a medium instance size in the 'public' network with 'default' security group. See "Customisation" below for advanced settings.

  1. Clone the repo.
  2. Add credentials in a terraform.tfvars file, expected credentials name can be found in provider.tf
  3. Run 
    terraform init
  4. Edit server1.yaml and provide your own custom provisioning configuration in container linux config syntax.
  5. Plan and apply. This will also auto-generate an SSH key pair to log in to the server after provisioning. The public key of that key pair will be registered with OpenStack.
    Invoke Terraform: 
    terraform plan
terraform apply

Terraform will print server information (name, ipv4 and v6) after deployment concluded. The deployment will create an SSH key pair in .ssh/.

After provisioning concluded the private key of the key pair generated during provisioning can be used to connect to node. As mentioned in the disclaimer, the tested setup relies on DevStack so we need to proxy jump on the devstack instance before reaching the deployed instance:

ssh -J user@[DEVSTACK-IP] -i ./.ssh/provisioning_private_key.pem -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null core@[SERVER-IP]

NOTE:

  • Server IP address can be found at any moment after deployment by running terraform output
  • If you update server configuration(s) in server-configs and re-run terraform apply, the instance will be replaced. Consider adding create_before_destroy to the openstack_compute_instance_v2 resource in compute.tf to avoid services becoming unavailable during reprovisioning.
  • SSH key is passed twice: in the Butane configuration and in the key_pair argument of the instance - while the SSH keys in the Butane configuration are added by Ignition during the first boot of the instance, key_pair are processed by [email protected] (a.k.a afterburn) from the OpenStack metadata service. It means that you can install SSH keys without Ignition and change them as wanted.

Customisation

The provisioning automation can be customised via settings in [terraform.tfvars][terraform.tfvars]:

  • cluster_name: Descriptive name of your cluster, will be used to generate server names. flatcar-terraform by default.
  • machines: Add more machines to your deployment. Each machine name must be unique and requires a respective [NAME].yaml server configuration in server-configs. An example / default configuration for machine server1 is provided with server1.yaml. During provisioning, server names are generated by concatenating the cluster name and the machine name - the defaults above will create a single server named flatcar-terraform-server1.
  • ssh_keys: Additional SSH public keys to add to core user's authorized_keys. Note that during provisioning, a provisioning RSA key pair will be generated and stored in the local directory .ssh/provisioning_private_key.pem and .ssh/provisioning_key.pub, respectively. The private key can be used for ssh connections (core user) to all provisioned servers.
  • release_channel: Select one of "lts", "stable", "beta", or "alpha". Read more about channels here.
  • flatcar_version: Select the desired Flatcar version for the given channel (default to "current", which is the latest).
  • flavor_name: The spec of the machine, it default to ds1G (1vCPU, 10GB of disk and 1GB of memory)
  • ssh: A boolean to create and attach a security group to the instances to allow SSH connections. (NOTE: At this moment, when creating an instance with this variable enabled then turning it off does not work as the security group is not firstly detached from the instance so the security group can't be deleted)