pbv

Puppet, Bolt, and Vagrant

(Bolt is not in action yet)

Puppet

Manage and automate the configuration of servers.

Puppet code is a declarative DSL (via Ruby).
Declarative languages describe "what" instead of "how".

The Puppet's primary server automates the configuration of its multiple agents' states. The primary server stores the Puppet code that defines the desired state. An agent retrieves, translates, and executes the code as commands to its specified systems (model-driven, pull-based workflow).

1. Agent ---> Facts ---> Primary
2. Primary -> Catalog -> Agent
3. Agent ---> Report --> Primary
4. Primary -> Report ✓

Puppet is available as puppetserver, puppet-agent, and puppetdb.

Goals

Consistency

validate applied config to achieve desired state
assume state is applied, identify the cause of failure
update the model to ensure the problem doesn't occur again

Automation

goal: maintain servers to achieve desired state
but what if there's hundreds/thousands of servers?
ease scaling infrastructure without doing so manually

Infrastructure as Code

combine software development and its structural operations
adopt version control, peer review, automated testing, CI/CD

Idempotent Workflow

run continuously, always get the same result
ensure the infrastructure's state matches the desired state
guarantee a system's desired state by repeatedly applying code
if a system's state deviates, Puppet will reapply its code

Using Puppet Bolt (with Vagrant)

orchestration tool to automate infrastructure maintenance
deploy apps, troubleshoot servers, patch systems, stop/restart services
agentless: connect via SSH, no agent installation required
use commands, scripts, and applied Puppet code (manifest blocks)
pluggable inventory system (nice to use with Vagrant)

Assuming your real machine is running Ubuntu:

See installation instructions here
Like Ansible, let's use Vagrant to prototype infrastructure with Bolt
Ensure VirtualBox or another VM provider is installed

Let's setup the project structure:

Let's also decide to use generic/rocky8 on the VM (CentOS 8 is EOL). Rocky Linux uses rpm instead of apt, if you're from Debian.

Initial artifact generation.
To devs using this repo: no need to regenerate, skip it.

bolt project init pbv
vagrant init generic/rocky8

Start the VM vagrant up.

Additional (puppetserver) Configuration

Install puppetserver:

sudo su -

yum update -y
rpm -Uvh https://yum.puppet.com/puppet-release-el-8.noarch.rpm
yum install -y puppetserver git

# Check version
puppetserver --version

Configure memory usage:

sudo su -

vim /etc/sysconfig/puppetserver

# Change:
# JAVA_ARGS="-Xms2g -Xmx2g
# to:
# JAVA_ARGS="-Xms512m -Xmx512m

systemctl start puppetserver
systemctl status puppetserver
# Active: active (running)
systemctl enable puppetserver

Append vm.hostname as agent:

vim /etc/puppetlabs/puppet/puppet.conf

[agent]
server = master.puppet.vm

Append path to puppet binary:

cd ~
vim .bash_profile

PATH=$PATH:$HOME/bin
PATH=$PATH:/opt/puppetlabs/puppet/bin

Install r10k, a code management tool:
https://puppet.com/docs/pe/2019.8/r10k.html

gem install r10k

Test puppet agent:

puppet agent -t

Add a control repository:

mkdir /etc/puppetlabs/r10k
vim /etc/puppetlabs/r10k/r10k.yaml

---
# Look for modules w/ r10k
:cachedir: '/var/cache/r10k'

:sources:
        :dj-mc:
                # Source control
                remote: 'https://github.com/dj-mc/puppet-control.git'
                # r10k will look here for code and modules
                basedir: '/etc/puppetlabs/code/environments'

# Deploy the env
r10k deploy environment -m

# Repository location
cd /etc/puppetlabs/code/environments/

dj-mc/pbv

pbv