Single source of truth for declarative workloads to be deployed on SAAP (Stakater App Agility Platform).

Stakater opinionated gitops repository structure:

• Mono repo; one repo for all clusters, tenants/teams and environments
• Single branch i.e. main and then separate folders per environment
• ArgoCD for application deployment
• Helm for templatization

As a software organization, I would like to:

• Audit all changes made to pipelines, infrastructure, and application configuration.
• Roll forward/back to desired state in case of issues.
• Consistently configure all environments.
• Reduce manual effort by automating application and environment setup and remediation.
• Have an easy way to manage application and infrastructure state across clusters/environments.

• GitOps is a natural evolution of DevOps and Infrastructure-as-Code.
• GitOps is when the infrastructure and/or application state is fully represented by the contents of a git repository. Any changes to the git repository are reflected in the corresponding state of the associated infrastructure and applications through automation.

GitOps is a set of practices that leverages Git workflows to manage infrastructure and application configurations. By using Git repositories as the source of truth, it allows the DevOps team to store the entire state of the cluster configuration in Git so that the trail of changes are visible and auditable.

GitOps simplifies the propagation of infrastructure and application configuration changes across multiple clusters by defining your infrastructure and applications definitions as “code”.

• Ensure that the clusters have similar states for configuration, monitoring, or storage.
• Recover or recreate clusters from a known state.
• Create clusters with a known state.
• Apply or revert configuration changes to multiple clusters.
• Associate templated configuration with different environments.

• Git is the source of truth.
• Separate application source code (Java/Go) from deployment manifests i.e the application source code and the GitOps configuration reside in separate git repositories.
• Deployment manifests are helm charts and standard Kubernetes (k8s) manifests i.e Kubernetes manifests in the GitOps repository can be simply applied with nothing more than a oc apply.
• Helm for defining the differences between environments.

Tenant (Product) owns Applications which are promoted to different Environments (Static and Dynamic)

Cluster >> Tenants (teams/products) >> Applications >> Environments

A cluster can hold multiple tenants; and each tenant can hold multiple applications; and each application be deployed into multiple environments.

This gitops structure supports:

• Multiple clusters
• Multiple tenants/teams/products
• Multiple apps
• Multiple environments (both static and dynamic)

We have 2 clusters:

1. Cluster # 1 (i.e. devtest) holds 3 environments: build, preview and dev
2. Cluster # 2 (i.e. prod) holds 2 environments: stage and prod

We have 2 product based tenants; who have exactly same structure

1. gabbar
2. veeru

We have 3 apps:

Gabbar has two apps:

1. stakater-nordmart-review,
2. stakater-nordmart-review-ui

Veeru has one app:

1. stakater-nordmart-inventory

We have 5 environments for each application:

1. build
2. preview
3. dev
4. stage
5. prod

A dedicated build namespace for a tenant which holds tekton pipleines; all tekton pipleines for all of applications of a given tenant run in this particular namespace

To support dynamic environments on PR (pull requests) we use Stakater Tronador; and Tronador's Environment custom resource is created in this namespace.

Environment names are prefixed by a number which depends upon the order in which the application should be promoted between the different environments; e.g. DTE > dev > stage > prod

We follow similar naming convention for naming the clusters.

The cluster scoped infrastructural configurations are deployed through a separate infra repository.

CI/CD/CD

The pipelines produce following outputs

Helm charts in GitOps repositories can be defined

• Option # 1: locally, with a templates folder including all the Kubernetes manifests or
• Option # 2: without a templates folder using a dependency and this dependency points to an external Helm Chart Repository where your chart is located.

We prefer the later approach (option # 2); with this setup, the only configuration present in the GitOps repository is the environment-specific configuration and dependencies, based on versions. This solution scales better and has better versioning capabilities, but requires some more Helm expertise from developers.

Your resources are structured as follows:

1. For this particular environment, you define a Chart.yaml with an external dependency to a Chart Version. The Helm chart is located in a remote Helm Chart repository

2. The values.yaml for this particular environment contains all the environment-specific configuration and includes an image tag that points to a tag in the Container Registry

Credits