Kapitan Core Concepts

Kapitan essentially allows you to organise your data with the inventory, and uses it to configure and run multiple Input Types which generate files.

So you can essentially compose your configuration using your favourite tools.

    graph LR
        TARGETS ==  **kapitan compile** ==> INPUT_TYPES
        INPUT_TYPES --> TARGETS_OUTPUT
        TARGETS_OUTPUT --> GENERATED_FILES


        INPUT_TYPES("Input types")

        subgraph "inventory"
            direction LR
            CLASSES("classes") --> TARGETS
            TARGETS(["`targets/production.yml`"])
        end
        subgraph "compiled output"
            TARGETS_OUTPUT([production/])
            GENERATED_FILES("manifests")
        end

Kapitan supports multiple Input Types, and the support keeps growing.

• Helm Charts
• jsonnet
• Python
• Jinja2

Targets

Kapitan organises your configuration through targets.

A simple target is typically made of a classes and a parameters sections, and it looks like this:

# inventory/targets/production.yaml
classes:
  - common
  - acme.applications.super

parameters:
    description: My Super Application

Compiling targets

You can tell Kapitan to compile all your targets by running the following command in a valid kapitan repository root:

$ kapitan compile
Rendered inventory (0.06s): discovered 200 targets.
Compiling 200/200 targets using 64 concurrent processes: (64 CPU detected)
...
Compiled production (0.68s)
...
Compiled 200 targets in 20s

Alternatively, you can select just the target you want to compile, using the -t or --target flag:

$ kapitan compile -t production
Rendered inventory (0.06s): discovered 200 targets.
Compiling 1/200 targets using 1 concurrent processes: (64 CPU detected)
Compiled production (0.68s)
Compiled 1 targets in 0.73s

Targets outputs

The output of a target compilation gets copied under the compiled folder, with the same structure as the original target.

📁 inventory/
└── 📁 targets/
    └── 📄 production.yml    #  --->

📁 compiled/
├── 📄 ...
└── 📁 production/           # compiled output
    ├── 📄 README.md
    ├── 📁 scripts/
    ├── 📁 manifests/
    ├── 📁 terraform/
    └── 📄 ...

The files or directories being created depends on what is defined inside the targets, and will change based on the configuration you define in the Kapitan target.

Organising targets

You can organise targets in a directory structure, and Kapitan will mimic the structure in the compiled folder

How targets are named

Kapitan identifies targets by a name that is automatically constructed based on the directory structure under the inventory/targets folder.

teams/devops/grafana.yml becomes teams.devops.grafana

📁 inventory/
└── 📁 targets/
    └── 📁 teams/
        └── 📁 devops/
            └── 📄 grafana.yml # teams.devops.grafana

📁 compiled/
└── 📁 teams/
    ├── 📄 ...
    └── 📁 devops/
        └── 📁 grafana/    # teams.devops.grafana
            ├── 📁 scripts/
            ├── 📁 manifests/
            └── 📁 terraform/

Classes

Kapitan allows you to organise your shared inventory data using classes, which are found in the inventory/classes folder.

📁 inventory/
└── 📁 classes/
    ├── 📄 common.yml              # common
    └── 📁 teams/
        ├── 📁 project1/
        │   └── 📄 config.yml      # teams.project1.config
        └── 📁 project2/
            ├── 📄 config.yml      # teams.project2.config
            └── 📄 kubernetes.yml  # teams.project2.kubernetes

The structure of a class is very similar to target files. Classes can refer to other classes.

# inventory/classes/common.yaml
classes:
    - kapitan.common

parameters:
    organization: ACME
    headquarters: London
    timezone: Europe/London