3. How everything connects (source to production)

The 12-Factor app support in Charmcraft and Rockcraft is an opinionated framework based on the 12-Factor methodology. If you web app uses one of the supported framework and follows the conventions of the tool, you can easily containerize, charm, and deploy it with Juju.

        flowchart TD
%% -- Styling Classes --
classDef nodeBox fill:#fff,stroke:#666,stroke-width:2px,color:#333,rx:5,ry:5,text-align:center;
classDef groupContainer fill:none,stroke:#7c4dff,stroke-width:2px,stroke-dasharray: 5 5,color:#333,font-size:12px;

%% -- Main Flow Styling --
linkStyle default stroke:#7c4dff,stroke-width:2px,fill:none;

%% -- Phase 1 --
subgraph P1 ["CODE"]
    direction TB
    Node1["Source:<br/>12-factor web app"]:::nodeBox
end

%% -- Phase 2 --
subgraph P2 ["BUILD + TEST"]
    direction TB
    Node2["Container:<br/>12-factor app rock"]:::nodeBox
    Node3["Software operator:<br/>12-factor app charm"]:::nodeBox
end

%% -- Phase 3 --
subgraph P3 ["RELEASE"]
    direction TB
    Node4["Published:<br/>12-factor app rock<br/>and charm in the<br/>Charmhub store"]:::nodeBox
end

%% -- Phase 4 --
subgraph P4 ["DEPLOY + OPERATE"]
    direction LR
    Node5["Production:<br/>12-factor app deployed<br/>to end users"]:::nodeBox
    Node6["Day 1 Operations:<br/>Integrate with database, ingress, SSO, etc."]:::nodeBox

    %% Internal Link (Grey)
    Node5 --> Node6
end

%% -- Phase 5 --
subgraph P5 ["MONITOR"]
    direction TB
    Node7["Observe: <br/> Track and monitor<br/>with COS"]:::nodeBox
end

%% -- Connections --
%% Connect Source to the first item in Build
Node1 --> Node2
Node1 --> Node3

%% Connect both Build items to Release
Node2 --> Node4
Node3 --> Node4

%% Connect Release to Deploy
Node4 --> Node5

%% Connect Deploy to Monitor
Node6 --> Node7

%% -- Apply Group Styles --
class P1,P2,P3,P4,P5 groupContainer

The diagram above shows the 12-factor tooling in the context of the development and operations lifecycle. The developer creates the app, uses Rockcraft and Charmcraft to build a container image and software operator, publishes the contents to the Charmhub store, and deploys the app to a Kubernetes cloud using Juju. The charming ecosystem provides everything needed for operating the charm, including databases, ingresses, and more.

We recommend to put the Rockcraft project file (rockcraft.yaml) in the same repository as your code, created with the Rockcraft extension for your specific framework. The rock generated with this project file is a fully compliant OCI-image that can be used outside of the Juju ecosystem.

Your web app containerized in a rock will be managed by a charm, a software operator orchestrated by Juju. You can create the charm using the init profile, that will use the appropriate Charmcraft extension. We recommend to place the charm code inside the charm directory in the same repository as your code.

The 12-Factor app support in Charmcraft and Rockcraft does not enforce any specific CI/CD pipeline. Some recommendations and useful tools are:

  • For the build stage, the rockcraft and charmcraft tools are used to create the rock and charm artifacts.

  • For integration tests involving charms, use the Jubilant library.

  • concierge is an opinionated utility to provision testing machines.

  • Charmcraft’s test command, based on Spread is a convenient full-system test (task) distribution.

  • Once your artifacts are ready, they can be uploaded to Charmhub and promoted to the desired channel. This is not a mandatory step, as you can deploy charms locally without Charmhub.

  • For the deployment, the current recommendation is to use the Juju Terraform Provider.

Juju is the engine that will orchestrate the software operators. The Juju ecosystem provides a multitude of curated software operators for your observability stack, database, SSO, and many more and allows their deployment and lifecycle management on metal, on VMs, on K8s and on cloud providers (see substrates).

Your web app will be able to integrate seamlessly with other charms, that can be running in Kubernetes or in Machines, and on-premises or in the cloud.

For the operation of your applications, it is strongly recommended to use the Canonical Observability Stack, an out-of-the-box solution for improved day-2 operational insight.