Fleet Awesome

A curated list of minimal reproducible examples, usable via SUSE Rancher Fleet. All assets are quick and minimal, focusing on a local lab environment. Customization with flexible config through secrets or downstream cluster ConfigMap objects. By default the options for the Bundles, or snippets, recipes, are opnionated towards operational efficiency, proof-of-concepts or pilot experiments. Unlike other "awesome-something" lists, the role of Fleet is as a facilitator for gitops operations with the repository using fleet-focused conventions and formatting.

QuickStart

Git

• Fork the repository
• Add a git remote with a unique name for your new fork
• (Optional) Create a branch for your changes and testing
• (Optional) Create Pull Requests (PRs) in GitHub for submitting your branch changes

Fleet

• If you have Rancher installed, you already have a Fleet Controller on the Local Cluster
  • The UI option for Continuous Delivery is Fleet's "SailBoat" icon
  • connections established to Downstream Clusters with their Fleet-Agent
• For each project, create one or more GitRepo resources, see GenericExample-ClusterGroup.yaml and GenericExample-GitRepo.yaml in the main repo root folder
  • more examples under /misc/location/parslab/sublevel/fleet-controller
• Each GitRepo can have one or more paths to where the recipes live under your new forked repo
• Assign Clusters (clusters.fleet.cattle.io) to ClusterGroups with labels
  • Fleet pulls the repo data with a GitJob, then creates Bundles from fleet.yaml files for HelmCharts & Kustomizations
• Final result is a BundleDeployment sent and deployed by the fleet-agent in the Downstream cluster for the target defined in the ClusterGroup
  • Here Target is a reference to a downstream Cluster and/or Namespaces on that cluster, depending on wether Bundles need to deploy namespaced or cluster-wide resources

CRD Flow

Each snippet or recipe relies on Fleet CRDs to generate and deilver a Bundle to the target Clusters.

• See the Fleet CRD workflow outlined in the diagram below.

See the Fleet docs about bundle lifecycle stages for more information.

Category Structure

• Infra - infrastructure-related assets, including CNI configurations, K8s distributions, dual-stack, and ingresses.
• Misc - any example or asset that does not fit into other categories, or upstream open source projects that might be covered in AppCo.
• Security - security-focused and other related topics, secrets, service mesh, mtls.
• Storage - CSI-related assets, homelab-focused storage options like minio and NFS.
• Vendor - any vendor-specific assets which may or may not be open source, or have a vendor-component in the asset, manifests, or charts. The primary example is Application Collection.

Features

• Avoid longer names for labels and charts with a 63 character limit. Achieved through helm.releaseName option.
• Avoid clobbering existing resources, through use of helm.takeOwnership: false option, for example when a Bundle is deployed into a pre-existing Namespace.
• Use of Kustomization configuration options for safer deployment and targeting.
• Day Two operations respecting existing Namespaces, Secrets and ConfigMap resources for targeted ClusterGroups.

Next Steps

• Future bundle topics like Elemental, GatewayAPI, ClusterAPI
• wrap-up and enduser testing, teammate feedback for QA
• QA, unit tests for charts and bundle templates