Skip to content

Latest commit

 

History

History
367 lines (266 loc) · 12.5 KB

README.md

File metadata and controls

367 lines (266 loc) · 12.5 KB
Raw

Provision

Code Owners ci GitHub release (latest by date)

Simple tasks to provision and tear down containers, instances and virtual machines.

Table of Contents

  1. Description
  2. Setup - The basics of getting started with provision
  3. Usage - Configuration options and additional functionality
  4. Limitations - OS compatibility, etc.
  5. Development - Guide for contributing to the module
  6. License - Additional information in regards to licensing

Description

Bolt tasks allowing a user to provision and tear down systems. It also maintains a Bolt inventory file. Provisioners so far:

  • ABS (AlwaysBeScheduling)
  • Docker
  • Vagrant
  • Provision Service

Setup

Setup Requirements

Bolt has to be installed to run the tasks. Each provisioner has its own requirements. From having Docker to installed or access to private infrastructure.

Running the tasks as part of puppet_litmus

Please follow the litmus documentation here.

Running the module stand-alone call the tasks/plans directly

For provisioning to work you will need to have a number of other modules available. You can use bolt to install the modules for you from your Puppetfile, by following this guide. The required modules are:

cat $HOME/.puppetlabs/bolt/Puppetfile
mod 'puppetlabs-puppet_agent'
mod 'puppetlabs-facts'
mod 'puppetlabs-puppet_conf'

Usage

There is a basic workflow for the provision tasks.

  • provision - creates/initiates a platform and edits a bolt inventory file.
  • tear_down - creates/initiates a system/container and edits a bolt inventory file.

For extended functionality please look at the provision wiki.

ABS

(internal to puppet) Allows you to provision machines on puppets internal pooler. Reads the '~/.fog' file for your authentication token.

Setting up your Token

In order to run ABS you first require an access token stored within your '.fog' file. If you already have one you may skip this section, otherwise request one by running the following command, changing the username.

$ curl -X POST -d '' -u tp --url https://test-example.abs.net/api/v2/token
Enter host password for user 'tp':
{
  "ok": true,
  "token": "0pd263lej948h28493692r07"
}%

Now that you have your token, check that it works by running:

$ curl --url https://test-example.abs.net/api/v2/token/0pd263lej948h28493692r07
{
  "ok": true,
  "user": "tp",
  "created": "2019-01-04 14:25:55 +0000",
  "last_accessed": "2019-01-04 14:26:27 +0000"
}%

Finally all that you have left to do is to place your new token into your '.fog' file as shown below:

$ cat ~/.fog
:default:
  :abs_token: 0pd263lej948h28493692r07

Running the Commands

Setting up a new machine
$ bundle exec bolt task run provision::abs --targets localhost action=provision platform=ubuntu-1604-x86_64

Started on localhost...
Finished on localhost:
  {
    "status": "ok",
    "node_name": "yh6f4djvz7o3te6.delivery.puppetlabs.net"
  }
Successful on 1 node: localhost
Ran on 1 node in 1.44 seconds
Tearing down a finished machine
$ bundle exec bolt task run provision::abs --targets localhost action=tear_down node_name=yh6f4djvz7o3te6.delivery.puppetlabs.net

Started on localhost...
Finished on localhost:
  Removed yh6f4djvz7o3te6.delivery.puppetlabs.net
  {"status":"ok"}
  {
  }
Successful on 1 node: localhost
Ran on 1 node in 1.54 seconds

Docker

Given an docker image name it will spin up that container and setup external ssh on that platform. For helpful docker tips look here

Containers by default will be managed in the current docker context, on the DOCKER_HOST, or on localhost if nether are configured.

Provision

$ bundle exec bolt task run provision::docker --targets localhost action=provision platform=ubuntu:14.04

Started on localhost...
Finished on localhost:
  Provisioning ubuntu_14.04-2222
  {"status":"ok","node_name":"localhost"}
  {
  }
Successful on 1 node: localhost
Ran on 1 node in 33.96 seconds

Provision allows for passing additional command line arguments to the docker run when specifying vars['docker_run_opts'] as an array of arguments.

When running Debian or Ubuntu containers, the following flags will be added to the $docker_run_opts by default.

--volume /sys/fs/cgroup:/sys/fs/cgroup:rw --cgroupns=host

These defaults can be overriden by passing the flags with different values i.e.

--volume /sys/fs/cgroup:/sys/fs/cgroup:ro --cgroupns=private
bundle exec bolt task run provision::docker --targets localhost action=provision platform=ubuntu:14.04 vars='{ "docker_run_opts": ["-p 8086:8086", "-p 3000:3000"]}'

Tear down

$ bundle exec bolt task run provision::docker --targets localhost action=tear_down node_name=localhost:2222

Started on localhost...
Finished on localhost:
  Removed localhost:2222
  {"status":"ok"}
  {
  }
Successful on 1 node: localhost
Ran on 1 node in 2.02 seconds

Vagrant

Tested with vagrant images:

  • ubuntu/trusty64
  • ubuntu/xenial64
  • ubuntu/bionic64
  • debian/jessie64
  • centos/7

provision

$ bundle exec bolt task run provision::vagrant --targets localhost action=provision platform=ubuntu/xenial64

Started on localhost...
Finished on localhost:
  {
    "status": "ok",
    "node_name": "127.0.0.1:2222"
  }
Successful on 1 node: localhost
Ran on 1 node in 51.98 seconds

For multi-node provisioning, you can assign arbitrary tags to the nodes you deploy, by passing an optional YAML-string 'vars' to the bolt task. In the example below we are assigning the role of k8s-controller to the provisioned node.

$ bundle exec bolt task run provision::vagrant --targets localhost action=provision platform=ubuntu/xenial64 inventory=/Users/tp/workspace/git/provision vars='role: k8s-controller'

sudo secure_path fix

As some Vagrant boxes do not allow ssh root logins, the vagrant user is used to login and sudo is used to execute privileged commands as root user. By default the Puppet agent installation does not change the systems' sudo secure_path configuration. This leads to errors when anything tries to execute puppet commands on the test system. To add the Puppet agent binary path to the secure_path please run the provision::fix_secure_path Bolt task:

$ bundle exec bolt task run provision::fix_secure_path path=/opt/puppetlabs/bin -i inventory.yaml -t ssh_nodes

Started on 127.0.0.1:2222...
Finished on 127.0.0.1:2222:
  Task completed successfully with no result
Successful on 1 target: 127.0.0.1:2222
Ran on 1 target in 0.84 sec

tear_down

$ bundle exec bolt task run provision::vagrant --targets localhost action=tear_down node_name=127.0.0.1:2222

Started on localhost...
Finished on localhost:
  Removed 127.0.0.1:2222
  {"status":"ok"}
  {
  }
Successful on 1 node: localhost
Ran on 1 node in 4.52 seconds

Additional Vagrant options

Box URL

If you need to use a local/private Vagrant box requiring you to specify a URL to access, you can pass a URL by setting the VAGRANT_BOX_URL environment variable. If you put %BOX% in the URL it will be replaced with the name of the box/image. You can also set the variable by setting the params hash in provision.yaml.

In provision.yaml:

vagrant_boxes:
  provisioner: vagrant
  images:
  - mybox1
  params:
    vagrant_box_url: 'https://boxes.example.com/box/%BOX%.json'
Password

If you use a custom Vagrant box which requires the use of a password instead of the standard Vagrant keypair then you can set it in the VAGRANT_PASSWORD environment variable, or in the provision.yaml params hash with the vagrant_password key.

Provision_service

The provision service task is meant to be used from a Github Action workflow.

Example usage: Using the following provision.yaml file:

test_serv:
  provisioner: provision::provision_service
  params:
    cloud: gcp
    region: europe-west1
    zone: europe-west1-d
  images: ['centos-7-v20200618', 'windows-server-2016-dc-v20200813']

In the provision step you can invoke bundle exec rake 'litmus:provision_list[test_serv]' and this will ensure the creation of two VMs in GCP.

Manual invocation of the provision service task from a workflow can be done using:

bundle exec bolt task run provision::provision_service --targets localhost action=provision platform=centos-7-v20200813 inventory=/Users/tp/workspace/git/provision/inventory.yaml vars='role: puppetserver'

Or using Litmus:

bundle exec rake 'litmus:provision[provision_service, centos-7-v20200813, role: puppetserver]'

Synced Folders

By default the task will provision a Vagrant box with the synced folder disabled. To enable the synced folder you must specify the parameter enable_synced_folder as true. Instead of passing this parameter directly you can instead specify the environment variable LITMUS_ENABLE_SYNCED_FOLDER as true.

Hyper-V Provider

This task can also be used against a Windows host to utilize Hyper-V Vagrant boxes. When provisioning, a few additional parameters need to be passed:

  • hyperv_vswitch, which specifies the Hyper-V Virtual Switch to assign the VM. If you do not specify one the Default Switch will be used.
  • hyperv_smb_username and hyperv_smb_password, which ensure the synced folder works correctly (only necessary is enable_synced_folder is true). If these parameters are omitted when provisioning on Windows and using synced folders Vagrant will try to prompt for input and the task will hang indefinitely until it finally times out. The context in which a Bolt task is run does not allow for mid-task input.

Instead of passing them as parameters directly they can also be passed as environment variables:

  • LITMUS_HYPERV_VSWITCH for hyperv_vswitch
  • HYPERV_SMB_USERNAME for hyperv_smb_username
  • HYPERV_SMB_PASSWORD for hyperv_smb_password

provision

PS> $env:LITMUS_HYPERV_VSWITCH = 'internal_nat'
PS> bundle exec bolt task run provision::vagrant --targets localhost action=provision platform=centos/7 hyperv_smb_username=tp hyperv_smb_password=notMyrealPassword

Started on localhost...
Finished on localhost:
  {
    "status": "ok",
    "node_name": "127.0.0.1:2222"
  }
Successful on 1 node: localhost
Ran on 1 node in 51.98 seconds

Using the tear_down task is the same as on Linux or MacOS.

Limitations

  • The docker task only supports Linux
  • The docker task uses port forwarding, not internal IP addresses. This is because of limitations when running on the mac.

Development

Testing/development/debugging it is better to use ruby directly, you will need to pass the JSON parameters. Depending on how you are running (using a puppet file or as part of a puppet_litmus). The dependencies of provision will need to be available. See the setup section above.

# powershell
 echo '{ "platform": "ubuntu-1604-x86_64", "action": "provision", "inventory": "c:\\workspace\\puppetlabs-motd\\" }' | bundle exec ruby .\spec\fixtures\modules\provision\tasks\abs.rb
# bash / zshell ...
 echo '{ "platform": "ubuntu-1604-x86_64", "action": "provision", "inventory": "/home/tp/workspace/puppetlabs-motd/" }' | bundle exec ruby spec/fixtures/modules/provision/tasks/abs.rb

Testing using bolt, the second step

bundle exec bolt task run provision::docker --targets localhost action=provision platform=ubuntu:14.04

License

This codebase is licensed under Apache 2.0. However, the open source dependencies included in this codebase might be subject to other software licenses such as AGPL, GPL2.0, and MIT.