|
| 1 | +# Introduction |
| 2 | + |
| 3 | +At [osscameroon](https://osscameroon.com) we often build software solutions that need to be deployed on different environments. |
| 4 | +This applications are packaged in several forms such as **docker images**, **static files** or **modules**. |
| 5 | + |
| 6 | +A packaged deployable application is called an **artifact**. Those artifacts are generated on CI jobs and developers will manually deploy them on our infrastructure. |
| 7 | +We currently have no nice and easy way to deploy applications automatically on both our **staging** and **production** environments. |
| 8 | + |
| 9 | +That's where **Sami** comes in. **Sami** is a system that will help us better manage the creation and deployment of our artifacts. |
| 10 | + |
| 11 | + |
| 12 | +# Architecture |
| 13 | + |
| 14 | +**Sami** is composed of several subsystems that handle the creation, storage and deployment of artifacts. |
| 15 | + |
| 16 | +- A **Repository** where the code lives with **CI integrations** triggered on pull request merged. |
| 17 | +- The **Artifact Management System** that will create, store and request a deployment of your artifacts. |
| 18 | +- The **Infrastructure** your application runs on. |
| 19 | + |
| 20 | + |
| 21 | + |
| 22 | + |
| 23 | +*Artifact Managment System* |
| 24 | + |
| 25 | +## Repository |
| 26 | + |
| 27 | +Where your application code lives. At [osscameroon](https://osscameroon.com) we store our source code on **GitHub**. |
| 28 | +This repository contains `CI` workflows that interacts with the **Artifact Managment System**. |
| 29 | + |
| 30 | +## Artifact Managment System |
| 31 | + |
| 32 | +The **Artifact Managment System** is responsible for the creation, storage of your artifacts, |
| 33 | + this is also where we can request the deployment of an artifact to any environment. |
| 34 | + |
| 35 | +In the case of an application packaged as a **docker image**, |
| 36 | + the application repository will contain several **CI workflows** that will be triggered |
| 37 | + when a pull request gets merged into the `master/main` branch. |
| 38 | + |
| 39 | +- **CI Artifact creation** workflow creates a docker image and pushes it to a **docker container registry**. |
| 40 | +- **CI Artifact deployment** request a deployment of the newly created docker image in our infrastructure. |
| 41 | + |
| 42 | + |
| 43 | + |
| 44 | + |
| 45 | +*Artifact Managment System Containers* |
| 46 | + |
| 47 | +### CI Artifact creation |
| 48 | +The Artifact creation CI is a reusable [GitHub Action](https://github.com/features/actions) that creates your artifacts, tags it and pushes it to an Artifact Registry. |
| 49 | +For a docker image that action will create the image and push it either on the [GitHub Container Registry](https://docs.github.com/en/packages/working-with-a-github-packages-registry/working-with-the-container-registry) or to [DockerHub](https://hub.docker.com/) |
| 50 | + |
| 51 | +*Note: we have built an Artifact Creation workflow [here](https://github.com/osscameroon/camerdevs/blob/682256603a8bd3cb58752a66e81ce36b37a1e6d6/.github/workflows/backend-build-api-image.yaml#L41)* |
| 52 | + |
| 53 | +### Artifact Registry |
| 54 | +An artifact registry designate the place your artifacts are being stored. For docker images it can be a container registry and for static files can be an s3 bucket. |
| 55 | + |
| 56 | + |
| 57 | +### CI Artifact deployment request |
| 58 | +This CI integration is a reusable [GitHub Action](https://github.com/features/actions) to request for your application artifact to be deployed on a specific environment. |
| 59 | + |
| 60 | +*Note: we have built an Artifact Deployment workflow [here](https://github.com/osscameroon/camerdevs/blob/682256603a8bd3cb58752a66e81ce36b37a1e6d6/.github/workflows/backend-build-api-image.yaml#L49)* |
| 61 | + |
| 62 | +## Infrastructure |
| 63 | + |
| 64 | +The application you build and package need to run on an infrastructure. The infrasture is a set of several components that allow us deploy, run, monitor and observe our applications/services. |
| 65 | + |
| 66 | +The infrastructure consists of these components: |
| 67 | + |
| 68 | +- A Git repository containing your applications deployment manifests. |
| 69 | +- An instance of [Webhook](https://github.com/adnanh/webhook). |
| 70 | +- A [Swarm](https://docs.docker.com/engine/swarm/) cluster which is where application will be run and scheduled. |
| 71 | + |
| 72 | +.png) |
| 73 | +*Infrastructure* |
| 74 | + |
| 75 | +### GitHub Repository |
| 76 | +Your applications deployment manifests should live in a GitHub repository. |
| 77 | +This repository also contains few **CI workflows** that will update your artifact sha in the manifest |
| 78 | +and/or trigger an api call to the **Webhook** that will trigger the actual deployment on your swarm clusters. |
| 79 | + |
| 80 | +**Examples:** |
| 81 | +- CI Updating the artifact sha in your manifest can found [here](https://github.com/osscameroon/deployments/blob/main/.github/workflows/deploy-service.yaml) |
| 82 | +- CI making the api call to the webhook can be found [here](https://github.com/osscameroon/deployments/blob/main/.github/workflows/apply-config.yaml) |
| 83 | + |
| 84 | +#### Deployment folder structure |
| 85 | +The deployment folder should follow a very strict structure and contain a `sami.yaml` file |
| 86 | +that describe your service's deployment. |
| 87 | + |
| 88 | +##### Folder |
| 89 | + |
| 90 | +At the root of the [deployments folder](https://github.com/osscameroon/deployments) |
| 91 | + |
| 92 | + |
| 93 | +###### L0 - root |
| 94 | + |
| 95 | +You can find these folder: |
| 96 | + |
| 97 | +- `conf` folder container configuration for applications running on the VPS such as `nginx`. |
| 98 | +- `services` which contains a list of services. |
| 99 | +Each service folder name should match the name of your service otherwise the `sami-cli` won't be able to pick them up correctly. |
| 100 | +- `README.md` that should contain a description of the deployments folder. |
| 101 | + |
| 102 | + |
| 103 | +###### L1 - service_name |
| 104 | +Inside the `services/<service_name>/` folder are contained folders named after the type of deployment you want to perform |
| 105 | +The supported type of deployment can be `swarm`, `compose`, `k8s`, `cron`, `static_file` etc... |
| 106 | + |
| 107 | +###### L2 - deployment_type |
| 108 | +The `services/<service_name>/<deployment_type>/` folder contain a folder named |
| 109 | +after the environment you want your application to be deployed to. |
| 110 | +The supported environment are `stage`, `production`, `pull_request`. |
| 111 | + |
| 112 | +###### L3 - environment |
| 113 | +In the `service/<service_name>/<deployment_type>/<environment>` can be found |
| 114 | +a `sami.(yaml|yml)` file that describes the service we would like to deploy. |
| 115 | + |
| 116 | +###### Tree |
| 117 | +``` |
| 118 | +├── README.md |
| 119 | +├── conf |
| 120 | +│ └── nginx |
| 121 | +│ └── sites-enabled |
| 122 | +│ ├── camerdevs.com |
| 123 | +│ ├── grafana.osscameroon.com |
| 124 | +│ ├── portainer.osscameroon.com |
| 125 | +│ └── webhooks.osscameroon.com |
| 126 | +└── services |
| 127 | + ├── camerdevs |
| 128 | + │ └── compose |
| 129 | + │ ├── prod |
| 130 | + │ └── stage |
| 131 | + │ └── camerdevs-stack.yml |
| 132 | + ├── caparledev-bot |
| 133 | + │ └── compose |
| 134 | + │ └── stage |
| 135 | + │ └── caparledev-bot-stack.yml |
| 136 | + ├── portainer |
| 137 | + │ └── compose |
| 138 | + │ └── prod |
| 139 | + │ └── portainer-agent-stack.yml |
| 140 | + └── prometheus |
| 141 | + └── compose |
| 142 | + └── prod |
| 143 | + ├── grafana |
| 144 | + ├── prometheus |
| 145 | + └── prometheus-stack.yml |
| 146 | +``` |
| 147 | + |
| 148 | +##### Sami's file |
| 149 | + |
| 150 | +the `sami.(yaml|yml)` file describes the service we want to deploy. |
| 151 | +The file is picked up and used by the `sami-cli` in order the deploy the application on our infrastructure. |
| 152 | + |
| 153 | +*Manifest* |
| 154 | +```yaml |
| 155 | +#service/<service_name>/<deployment_type>/<environment>/sami.(yml|yaml) |
| 156 | +version: v1beta |
| 157 | + |
| 158 | +service_name: "camerdevs" |
| 159 | +type: "swarm" |
| 160 | + |
| 161 | +#This description is not exhaustive and will be enriched as we design |
| 162 | +#the system |
| 163 | +``` |
| 164 | + |
| 165 | + |
| 166 | +### Webhook |
| 167 | + |
| 168 | +The [webhook](https://github.com/adnanh/webhook) is a peice of software that runs on your swarm manager node, usually your VPS and is listening for https requests. |
| 169 | +This webhook will receive endpoint https **POST** or **GET** request on the `/deployment_request` endpoint |
| 170 | +and deploy your application on the cluster using the `sami-cli`. |
| 171 | + |
| 172 | + |
| 173 | +### Swarm |
| 174 | + |
| 175 | +[Swarm](https://docs.docker.com/engine/swarm/) is our scheduler, we choose to use it because it runs well on low resources VPS. |
| 176 | +Which makes it very good for developers or teams without too much resources. |
| 177 | +We deploy a series of services on our swarm clusters for observability and monitoring. |
| 178 | +- Portainer |
| 179 | +- Prometheus |
| 180 | +- Grafana |
| 181 | + |
| 182 | +But also [Traefik](https://traefik.io/) a reverse proxy that will proxy our requests to the services running on swarm. |
0 commit comments