How splunk/splunk-operator works
Project type and purpose
• This is a Kubernetes Operator (the “Splunk Operator”) built with the Operator SDK in Go.
• Its purpose is to manage Splunk Enterprise deployments on Kubernetes by defining Custom Resource Definitions (CRDs) and reconciling desired state (Clusters, Standalones, LicenseManagers, etc.) via controller loops.
• It bundles OLM manifests, Helm charts, integration tests (kuttl), client libraries for Splunk REST API, and end-to-end tests.
a. Kubernetes Control Plane
• API Server (CRD and built-in resource registry)
• etcd (state store)
b. Splunk Operator Pod
• Manager (main.go)
• Controller-runtime machinery
• Reconcilers (one per CR type) in internal/controller/*
• Shared libraries: pkg/splunk/common, pkg/splunk/client, pkg/splunk/enterprise logic
c. Splunk Enterprise Workloads
• StatefulSets/Deployments/Pods for Splunk Clustermanager, IndexerCluster, SearchHeadCluster, MonitoringConsole, Standalone
• Services, ConfigMaps, Secrets (generated by controllers)
d. External Storage and Services
• AWS S3 / Azure Blob / GCP Buckets (via pkg/splunk/client/* clients)
• Optional sidecars for metrics, debug
e. Operator Installation Artifacts
• CRD YAMLs (config/crd/bases)
• RBAC manifests (config/rbac)
• OLM bundle (bundle/manifests)
• Helm charts (helm-chart/)
f. CI/CD & Testing
• GitHub workflows (.github/workflows)
• KUTTL integration tests (kuttl/, config/scorecard/)
• Unit tests (internal/controller/*_test.go, pkg/**/*.go tests)
Interactions and dataflow
• Operator watches CRDs & built-in resources (ServiceAccount, RoleBinding, etc.) via controller-runtime → events feed into reconciliation loops
• Reconcilers call into pkg/splunk/enterprise to compute desired Kubernetes objects → operator issues create/update to API Server
• Operator uses pkg/splunk/client to reach out to Splunk REST API endpoints on Pods for configuration, scaling, upgrades, smartstore actions
• Controllers reconcile loop until actual cluster state matches desired CR spec
• KUTTL and integration tests drive CR application and assert expected Kubernetes and Splunk side effects
Architectural patterns & principles
• Operator Pattern: controller/event-driven reconciliation loop
• CRD-First design: custom types under api/v* drive all behavior
• Shared library layering:
• api/v*/ types → internal/controller → pkg/splunk/common & client → pkg/splunk/enterprise business logic
• Separation of concerns: controller orchestration vs. Splunk REST logic vs. K8s resource templating
• Test coverage: unit tests for every package, integration via kuttl, smoke tests, CI pipelines
Technologies & frameworks
• Go modules, Operator SDK, controller-runtime
• Kubernetes API machinery (CRDs, RBAC, Namespaces)
• Helm for packaging operator & Splunk charts
• OLM bundle for marketplace integration
• KUTTL testing framework for end-to-end
• Storage clients: AWS S3 SDK, Azure Blob SDK, GCP Storage client
Essential architectural elements
• CRD Registration & API Server (config/crd, api/v*)
• Operator Manager & Controllers (main.go + internal/controller)
• Shared libraries & clients (pkg/splunk/common, pkg/splunk/client)
• Splunk Enterprise resource managers (pkg/splunk/enterprise)
• Generated Kubernetes resources (StatefulSets, Services, ConfigMaps)
• External storage integration
• Packaging artifacts (Helm, OLM bundle)