Infrable

Hanny's legendary infrastructure as code solution.

# Install
pip install -U infrable  # requires python >= 3.10

# Bootstrap a new project
infrable init

Table of contents

1. Prologue
2. Chapter 1 - Chaos
3. Chapter 2 - Hosts and Services
4. Chapter 3 - Templates
5. Chapter 4 - The Deploy or Recover Workflow
6. Chapter 5 - Commands, Tasks and Workflows
7. Chapter 6 - Environments and Switches
8. Chapter 7 - Meta and Secrets
9. Chapter 8 - Custom Modules

Prologue

In an infinite void surrounded by blinking lights, Hanny woke up with her memory hollow, only her name echoing in the dark corners of her mind. She found herself floating in a realm between the tangible world of Python code and the ethereal cosmos of the space. A synthetic voice resonated around her about a crucial infrastructure migration project but it was all Greek to her. Her memory, once the key to the Python kingdom, failed to provide any answers. She was stuck in an interstellar labyrinth, entrusted with a technology mission she could not recall, her only anchors being a mechanical keyboard and lines of code whizzing past on the nearby screen.

Chapter 1 - Chaos

In the void of her discontent, Hanny started to recall fragments of a past life: back on Earth, where a looming infrastructure migration project threatened to upend everything. Humanity grappled with a chaotic infrastructure and an ill-prepared toolkit. Python developers fought with convoluted tools, while the desperate search for declarative configuration management proved futile. Amidst this disarray, Hanny was catapulted into the cosmos, a reluctant knight tasked with an unenviable quest: to find an uncomplicated yet competent solution capable of taming this looming catastrophe.

Chapter 2 - Hosts and Services

In a moment of revelation, Hanny grasped the intricacies of infrastructure: hosts and the services they housed, the vital backbone of any system. She understood the criticality of maintaining a single reference source, a complete, harmonious documentation capturing these elemental relationships. Committing herself to uphold this single source of truth principle, she began the painstaking process of documentation, pouring every detail into a consolidated testament christened "infra.py".

infra.py

from infrable import Host, Service

# Hosts/ -----------------------------------------------------------------------
dev_host = Host(fqdn="dev.example.com", ip="127.0.0.1")
beta_host = Host(fqdn="beta.example.com", ip="127.0.0.1")
prod_host = Host(fqdn="prod.example.com", ip="127.0.0.1")
# /Hosts -----------------------------------------------------------------------


# Services/ --------------------------------------------------------------------
dev_web = Service(host=dev_host, port=8080)
beta_web = Service(host=beta_host, port=8080)
prod_web = Service(host=prod_host, port=8080)

dev_nginx = Service(host=dev_host, port=80)
beta_nginx = Service(host=beta_host, port=80)
prod_nginx = Service(host=prod_host, port=80)
# /Services --------------------------------------------------------------------

List the hosts and services:

infrable hosts
infrable services

Chapter 3 - Templates

Gradually piecing together her fragmented memories, Hanny realized configuration files for the host deployments ought to be maintained as templates, drawing values as needed from "infra.py". Back on Earth, the challenge had been a lack of a coherent system to document the destination of these files, a problem of organization that posed a significant hurdle. Then, in a moment of genius, Hanny conceived a groundbreaking solution. She'd document the files' path directly within the configuration templates themselves. And so, with renewed vigor, she started adding crucial details as header comments nestled within the configuration files, a legendary stroke promising to transform the face of infrastructure migration.

infra.py

template_prefix = "https://github.com/username/repository/blob/main"

templates/nginx/web.j2

# vim: syn=nginx

# ---
# src: {{ template_prefix }}/{{ _template.src }}
# dest: {{ dev_nginx.host }}:/etc/nginx/sites-enabled/web
# chmod: 644
# chown: root:root
# ---

server {
    listen {{ dev_nginx.port }};
    listen [::]:{{ dev_nginx.port }}

    server_name {{ dev_nginx.host.fqdn }} www.{{ dev_nginx.host.fqdn }};

    location / {
        proxy_pass http://127.0.0.1:{{ dev_web.port }};
        include proxy_params;
    }
}

NOTE: The _template.src is a special variable, available in all templates.

Chapter 4 - The Deploy or Recover Workflow

In the vast expanse of uncertainty, one thing became crystal clear to Hanny: the importance of reviewing the files made through "infra.py" and comparing these with the currently deployed configurations before pushing them. This process would intercept any live changes and ensure their inclusion in the templates. Her cautious nature also recognized the essential function of maintaining local backups of the utilized configurations, providing an insurance of sorts. To address any complications, she conceived a failsafe measure: The Deploy or Recover Workflow. It promised relief from human error while ensuring an approach to easily revert and recover the service, another triumphant stride in Hanny’s cosmic saga.

Deploy workflow:

infrable files deploy [path]

## Same as
# infrable files gen [path]
# infrable files pull
# infrable files backup
# infrable files push

For snake folks:

from infrable import files

files.deploy(path)

## Same as
# files.gen(path)
# files.pull()
# files.backup()
# files.push()

flowchart TD;
    A[gen: generate artifacts from templates and infra.py as .new files] --> B[pull: for each generated artifact pull the currently deployed version from server as .old files];
    B --> C[backup: copy the artifacts in a local backup directory for easy recovery in case of failure];
    C --> E[diff: compare the .new and .old files];
    E -- push: for each result --> F{is there any difference?};
    F -- yes --> H[display the diff];
    F -- no --> G[skip and delete artifacts];
    H --> I{confirm push?};
    I -- yes --> J[push the file onto the server];
    I -- no --> K[skip and delete artifacts];

Diff example:

--- .infrable/files/root@dev.example.com/etc/monit/conf.d/system.cfg.old

+++ .infrable/files/root@dev.example.com/etc/monit/conf.d/system.cfg.new

@@ -1,13 +1,14 @@

 check system dev.example.com
-    if memory usage > 85% then alert
+    # if memory usage > 85% then alert
     if cpu usage (user) > 80% for 3 cycles then alert
     if cpu usage (system) > 80% for 3 cycles then alert

Push? (y, n, all) [y]:

Recover workflow:

infrable files recover [path]

## Same as
# infrable files revert [path]
# infrable files push

For snake folks:

from infrable import files

files.recover(path)

## Same as
# files.revert(path)
# files.push()

flowchart TD;

    A[revert: copy the artifacts from the given or latest backup directory into artifacts directory but in reverse order] --> B[diff: compare the .new and .old files];
    B --> C[push: run the steps for the push workflow]

Chapter 5 - Commands, Tasks and Workflows

As Hanny delved deeper into the complexities of infrastructure migration, she encountered a critical realization: the need to extend beyond mere configuration pushing. To attain a seamless transition, testing, service restarts, and other post-deployment actions were imperative. With relentless determination, she integrated a feature within "infra.py" to execute remote commands on hosts, a capability enhancing the deployment process.

# Run a command on a host by name
infrable remote dev_host "sudo systemctl reload nginx"

# Or by service name
infrable remote dev_nginx "sudo systemctl reload nginx"

# Or all affected hosts (as per files diff)
infrable remote affected-hosts "sudo systemctl reload nginx"

# Or
infrable files affected-hosts | infrable remote - "sudo systemctl reload nginx"

But Hanny's pursuit of excellence didn't halt there; she grasped the inadequacy of loose command execution, acknowledging the necessity for structured organization. Thus, she envisioned a novel concept - the creation of "Tasks", groups of related commands designed to streamline operations.

infra.py

import typer

# Tasks/ -----------------------------------------------------------------------
dev_nginx.typer = typer.Typer(help="dev_nginx specific tasks.")

@dev_nginx.typer.command(name="reload")
def reload_dev_nginx():
    """[TASK] Reload nginx: infrable dev-nginx reload"""

    assert dev_nginx.host, "Service must have a host to reload"
    dev_nginx.host.remote().sudo.nginx("-t")
    dev_nginx.host.remote().sudo.systemctl.reload.nginx()
# /Tasks -----------------------------------------------------------------------

Running tasks:

infrable dev-nginx reload

As she toiled to realize this vision, a brilliant idea swept over her - to orchestrate these tasks into coherent sequences, she christened them "workflows", heralding the dawn of a new era in infrastructure management.

infra.py

from infrable import concurrent, paths

# Workflows/ -----------------------------------------------------------------------
deploy = typer.Typer(help="Deployment workflows.")

@deploy.command(name="dev-nginx")
def deploy_dev_nginx():
    """[WORKFLOW] Deploy dev_nginx files: infrable deploy dev-nginx"""

    files.deploy(paths.templates / "nginx")
    cmd = "sudo nginx -t && sudo systemctl reload nginx && echo success || echo failed"
    fn = lambda host: (host, host.remote().sudo(cmd))
    for host, result in concurrent(fn, files.affected_hosts()):
        print(f"{host}: {result}")
# /Workflows -----------------------------------------------------------------------

Running workflows:

infrable deploy dev-nginx

Chapter 6 - Environments and Switches

The exhilaration of progress was swiftly countered by the rigors of defining templates, tasks, and workflows independently for every host within multiple environments. Hanny's keen eyes blinked tiredly at lines of Python code, her fingers aching from relentless typing. It was then that she stumbled upon another vital realization: the service and host relationship hinged upon the environment. The environment — "dev", "beta", "prod", among others — determined where services were deployed. Wondering if environments could be switchable, she pondered a pivotal question: What if the "infra.py" file could adjust values based on the environment? This would phenomenally simplify the deployment process, utilizing the same template, task, and workflow definitions across different hosts. Seizing on this insight, Hanny developed 'Switch', an ingenious mechanism allowing environment-dependent value adjustment within "infra.py". With 'Switch', infrastructure migration wasn't a labyrinth to navigate but a journey of exploration and innovation. The weight of her mission lightened with every line of code she churned out. And so Hanny, engulfed by the silence of the cosmos, marched on, her spirit indomitable, her zeal untamed.

infra.py

from infrable import Switch, Host, Service

# Environments/ ----------------------------------------------------------------
dev = "dev"
beta = "beta"
prod = "prod"

environments = {dev, beta, prod}
env = Switch(environments, init=dev)  # <-- Defining a switch for different environments
current_env = env()
# /Environments ----------------------------------------------------------------

# Hosts/ -----------------------------------------------------------------------
dev_host = Host(fqdn="dev.example.com", ip="127.0.0.1")
beta_host = Host(fqdn="beta.example.com", ip="127.0.0.2")
prod_host = Host(fqdn="prod.example.com", ip="127.0.0.3")

managed_hosts = env(  # <-- Switching hosts based on the environment
    dev=[dev_host],
    beta=[beta_host],
    prod=[prod_host],
)
# /Hosts -----------------------------------------------------------------------

# Services/ --------------------------------------------------------------------
web = Service(
    host=env.strict(dev=dev_host, beta=beta_host, prod=prod_host),  # <-- Strict switch
    port=8080,
)

nginx = Service(port=80, host=web.host)  # <-- No need to use switch here
# /Services --------------------------------------------------------------------

Updating the templates to use the switchable values:

templates/nginx/proxy_params.j2

# vim: syn=nginx

# ---
# src: {{ template_prefix }}/{{ _template.src }}
# dest:
# {% for host in managed_hosts %}  # <-- Yes, you can
#   - loc: {{ host }}:/etc/nginx/proxy_params
# {% endfor %}
# chmod: 644
# chown: root:root
# ---
proxy_set_header Host $http_host;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header X-Forwarded-Proto $scheme;

Managing switches:

# Switch current environment
infrable switch env [dev|beta|prod]

# Check the current switch value
infrable switch env

# If you name a switch "env", you get an alias for convenience
infra env [dev|beta|prod]

# Check all switch values
infrable switches

Chapter 7 - Meta and Secrets

A veil of mystery cloaked Hanny's latest quest. She coveted the ability to access secret values, untouched by the eyes of version control systems, and juxtapose these to hosts and services. It was a daunting challenge — to embed securely these secret keys within the Python labyrinth without heightening complexity. She mused, her gaze focused on the stars scattered across the cosmos. If simplicity was the mother of invention, Hanny would be its devoted disciple. She engaged in the great Pythonic tradition of simplicity, armed with wisdom acquired from prior challenges, plunging into the task at hand. She devised an elegant strategy. Her solution was a poet’s dream and a programmer’s delight, the relief of her achievement echoing in the cosmic silence.

infra.py

from infrable import Meta, readfile

common_secret_key = readfile("secrets/common/secret_key")  # <-- Read a secret file

web = Service(
    meta=Meta(secret_key=common_secret_key),  # <-- Attach metadata to items
    host=env(dev=dev_host, beta=beta_host, prod=prod_host),
    port=8080,
)

Managing secrets:

# Hide secrets from git
echo /secrets/ >> .gitignore

# Update the secrets by hand
vim secrets/web/secret_key

Chapter 8 - Custom Modules

As the celestial silence enveloped her, Hanny stared at the last line of code and finally, exhaled. The Python maestro had defied the odds, charting a path through the sprawling galaxy of infrastructure migration. In her hands, humanity held the weapon to combat the specter of inevitable infrastructure displacement, not just once but potentially countless times henceforth. No longer would they flounder in the face of migration, for 'infrable' was here.

But if the endless cosmos stood as a testament to anything, it was the limitless potential for growth. Hanny, ever the visionary, saw beyond the painstaking creation of a toolkit. She envisioned a continuously evolving mechanism, a collaborative system enhanced by the collective genius of Python peers worldwide. That's when it struck her. Like a supernova in the silent night, she realized the adaptability imbued within Python. Fellow programmers could craft Python modules and stitch them into the existing "infra.py", thus extending and elevating its functionality, a synergy of skills and codes enhancing 'infrable' to uncharted territories.

lib/modules/mycloud.py

from dataclasses import dataclass
from typer import Typer
from infrable import infra
from infrable.distro.ubuntu import UbuntuLatestHost

@dataclass
class MyCloud:
    """MyCloud Python library."""

    secret_api_key: str
    typer: Typer | None = None

    def provision_ubuntu_host(self, fqdn: str):
        ip = self.api.create_ubuntu_host(fqdn)
        return MyCloudUbuntuHost(fqdn=fqdn, ip=ip)

@dataclass
class MyCloudUbuntuHost(UbuntuLatestHost):
    """MyCloud's customized Ubuntu server."""

    def setup(self, as_root: bool = False):
        super().setup(as_root=as_root)
        self.install_mycloud_agent()

    def install_mycloud_agent(self):
        raise NotImplementedError

workflows = Typer()

@workflows.command()
def provision_ubuntu_host(fqdn: str, setup: bool = True):
    """[WORKFLOW] Provision Ubuntu host."""

    # Get the MyCloud instance from infra.py
    cloud = next(iter(infra.item_types[MyCloud].values()))

    # Provision the host
    host = cloud.provision_ubuntu_host(fqdn)
    if setup:
        host.setup(as_root=True)

    name = fqdn.split(".")[0]
    print("Add the host to the infra.py file.")
    print(f"{name} = {repr(host}")

Plugging the module in infra.py:

infra.py

from lib.modules import mycloud

# Clouds/ ----------------------------------------------------------------------
cloud = mycloud.MyCloud(secret_api_key=readfile("secrets/mycloud/secret_api_key"))
cloud.typer = mycloud.workflows
# /Clouds ----------------------------------------------------------------------

Running the module workflows:

infra cloud --help