How opensearch-project/helm-charts works
Project Type and Purpose
• This is an open-source repository of Helm charts for deploying the OpenSearch stack (OpenSearch core, OpenSearch Dashboards, Data Prepper) on Kubernetes.
• It’s not an application per se but a packaging/templating layer (Helm) that codifies Kubernetes manifests and CI/CD pipelines for chart linting, testing, and release.
Main Components
A. Source & CI/CD
• GitHub Repository (helm-charts)
• Developer (pushes commits/PRs)
• GitHub Actions workflows:
• lint-test.yaml (chart linting & basic validation)
• release.yaml (package and publish charts)
• chart-releaser (publishes to gh-pages)
B. Chart Artifacts & Registry
• GitHub Pages branch (gh-pages) hosting index.yaml and chart tarballs
• Helm repository endpoint: https://opensearch-project.github.io/helm-charts/
C. Consumers
• End users or automation pipelines running helm repo add & helm install
D. Kubernetes Cluster
• Kubernetes API Server
• Worker nodes running pods for:
• OpenSearch (StatefulSet, Services, ConfigMaps, PVCs)
• OpenSearch Dashboards (Deployment, Service, Ingress)
• Data Prepper (Deployment, ConfigMap, Service)
• Supporting Kubernetes primitives: Secrets, Ingress, HPA, PodDisruptionBudget, NetworkPolicy, ServiceAccount, RBAC
Relationships & Data/Control Flows
• Developer → push code → GitHub repo
• GitHub Actions (on push/PR) executes lint/test → upon tag or release event runs packaging
• Chart-Releaser uploads packaged .tgz + updates index.yaml in gh-pages branch
• End user helm repo update pulls index from gh-pages
• helm install fetches chart → renders templates → sends Kubernetes manifests to API server
• API server schedules pods on worker nodes → pods pull Docker images (OpenSearch, Dashboards, Data Prepper) from container registry
Architectural Patterns & Design Principles
• GitOps-style chart management: declarative manifests stored in Git and published via CI to a versioned Helm repo
• Chart templating: reusable parameterized YAML templates under charts/<chart>/templates
• Separation of concerns: three distinct charts targeting different layers of the search pipeline
• Kubernetes best practices: use of StatefulSet for OpenSearch, Deployments for stateless components, RBAC, network policies, health checks
Key Technologies & Frameworks
• Helm (Chart.yaml, values.yaml, templates)
• Kubernetes primitives (StatefulSet, Deployment, Service, Ingress, ConfigMap, Secret, PVC, ServiceAccount, RBAC)
• GitHub Actions for CI/CD
• Chart-Releaser (automates publishing to GitHub Pages)