openshift · blublinsky · May 12, 2026
diff --git a/.cursor/skills/find-duplication/SKILL.md b/.cursor/skills/find-duplication/SKILL.md
@@ -93,7 +93,7 @@ For each duplicate found, classify:
 |----------|--------|
 | **Extract** — identical logic in 3+ places | Recommend a shared helper in utils |
 | **Parameterize** — same structure, different values | Recommend a common function with parameters |
-| **Acceptable** — similar but serving different domains (app server vs lcore) | Note it, no action needed |
+| **Acceptable** — similar but serving different domains (e.g. appserver vs postgres) | Note it, no action needed |
 | **Boilerplate** — kubebuilder/controller-runtime patterns | Skip, this is framework convention |
 | **Test-only** — repeated test setup/fixtures | Recommend shared test fixture (only if user asked) |
 

diff --git a/AGENTS.md b/AGENTS.md
@@ -43,7 +43,7 @@ OLSConfigReconciler.Reconcile() →
 ├── reconcileLLMSecrets()
 ├── reconcileConsoleUI()
 ├── reconcilePostgresServer()
-└── reconcileAppServer() OR reconcileLCore()  [MUTUALLY EXCLUSIVE via --enable-lcore flag]
+└── reconcileAppServer() (application server via `appserver` package)
     └── [12+ sub-tasks via ReconcileTask pattern]
 ```
 
@@ -73,8 +73,7 @@ make test-e2e   # E2E tests (requires cluster)
 
 ### Controllers
 - `internal/controller/olsconfig_controller.go` - Main reconciler with finalizer logic
-- `internal/controller/appserver/` - App server (LEGACY)
-- `internal/controller/lcore/` - Lightspeed Core (NEW)
+- `internal/controller/appserver/` - App server
 - `internal/controller/postgres/` - PostgreSQL
 - `internal/controller/console/` - Console UI
 - `internal/controller/watchers/` - External resource watching
@@ -83,8 +82,10 @@ make test-e2e   # E2E tests (requires cluster)
 
 ### Tests
 - `*_test.go` - Unit tests (co-located)
-- `test/e2e/` - E2E tests
-- `internal/controller/utils/testing.go` - Test utilities
+- `test/e2e/` - E2E tests (shared helpers in `assets.go`, `utils.go`, `client.go`, and related files)
+- `internal/controller/utils/test_fixtures.go` - Shared controller **unit-test** fixtures (default `OLSConfig` CR, random secret/configmap/TLS helpers, telemetry pull-secret create/delete, shared `With*` provider mutators). Add here when the same shape is reused across tests or packages.
+- For **one-off** CR or spec fragments used only in one file, **inline** next to the test (or use a file-local unexported helper in that `*_test.go`) instead of `test_fixtures.go`, so refactors do not leave unused exported helpers.
+- `internal/controller/utils/testing.go` - `TestReconciler`, `NewTestReconciler`, and `StatusHasCondition` for envtest-based suites
 - `internal/controller/suite_test.go` - Test suite setup, shared helpers
   - `cleanupOLSConfig()` - Reusable CR cleanup helper (removes finalizers, waits for deletion)
 

diff --git a/ARCHITECTURE.md b/ARCHITECTURE.md
@@ -4,7 +4,7 @@ This document describes the internal architecture of the OpenShift Lightspeed Op
 
 ## Overview
 
-The operator follows a modular, component-based architecture where each major component (application server, Lightspeed Core/Llama Stack, PostgreSQL, Console UI) is managed by its own dedicated package with independent reconciliation logic.
+The operator follows a modular, component-based architecture where each major component (application server, PostgreSQL, Console UI) is managed by its own dedicated package with independent reconciliation logic.
 
 ## Key Design Decisions
 
@@ -37,7 +37,7 @@ The operator follows a modular, component-based architecture where each major co
 **Core Orchestration:**
 - Main `Reconcile()` method coordinates all reconciliation phases
 - `SetupWithManager()` configures controller watches and event handlers
-- Selects backend: calls either `appserver.ReconcileAppServer()` OR `lcore.ReconcileLCore()` based on `--enable-lcore` flag
+- Reconciles the application server via `appserver` (see `ReconcileAppServerResources` / `ReconcileAppServerDeployment` in `olsconfig_controller.go`)
 
 **Support Functions:**
 - Implements `reconciler.Reconciler` interface (provides config/images to components)
@@ -58,7 +58,7 @@ The operator follows a modular, component-based architecture where each major co
 - Detect OpenShift version and select appropriate images
 - Start controller and handle graceful shutdown
 
-**Key Flags:** `--enable-lcore` (backend selection), `--controller-namespace`. See `cmd/main.go` for complete list.
+**Key Flags:** Image URLs, `--controller-namespace`, reconcile interval, and related runtime options. See `cmd/main.go` for the complete list.
 
 ### Reconciler Interface (`internal/controller/reconciler`)
 
@@ -70,22 +70,9 @@ Provides clean contract between main controller and component packages:
 
 ### Application Server Package (`internal/controller/appserver`)
 
-**Purpose:** Manages OpenShift Lightspeed application server (LEGACY backend - LLM API proxy)
+**Purpose:** Manages the OpenShift Lightspeed application server (LLM API, RAG, optional MCP, metrics).
 
-**Entry Point:** `ReconcileAppServer(reconciler.Reconciler, context, *OLSConfig)`
-
-### Lightspeed Core Package (`internal/controller/lcore`)
-
-**Purpose:** Manages Lightspeed Core + Llama Stack server (NEW backend - agent-based with MCP support)
-
-**Entry Point:** `ReconcileLCore(reconciler.Reconciler, context, *OLSConfig)`
-
-**Key Features:**
-- Dynamic LLM configuration (supports OpenAI, Azure OpenAI, others)
-- CA certificate support for custom TLS
-- RAG support with vector database
-- MCP (Model Context Protocol) integration
-- Metrics with K8s authentication
+**Entry Points:** `ReconcileAppServerResources` and `ReconcileAppServerDeployment` (invoked from `olsconfig_controller.go`).
 
 ### PostgreSQL Package (`internal/controller/postgres`)
 
@@ -138,7 +125,7 @@ High-level reconciliation sequence:
 6. Reconcile Components:
    - Console UI (if enabled)
    - PostgreSQL (if conversation cache enabled)
-   - Backend (AppServer OR LCore - mutually exclusive, controlled by --enable-lcore flag)
+   - Application server (`appserver` package)
 7. Update Status Conditions based on deployment readiness
 ```
 

diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
@@ -10,12 +10,12 @@ This guide provides instructions for contributing to the OpenShift Lightspeed Op
 ## Quick Context
 
 The operator uses a **modular, component-based architecture**:
-- Each component (appserver, lcore, postgres, console) is a self-contained package
+- Each component (`appserver`, `postgres`, `console`) is a self-contained package
 - Components use the `reconciler.Reconciler` interface (no circular dependencies)
 - Task-based reconciliation pattern (list of tasks executed sequentially)
 - Two resource approaches: **Owned** (operator-created, ResourceVersion tracking) vs **External** (user-provided, data comparison)
 
-**Best way to learn**: Read existing component code (`appserver/`, `postgres/`, `console/`, `lcore/`)
+**Best way to learn**: Read existing component code (`appserver/`, `postgres/`, `console/`)
 
 ---
 
@@ -58,7 +58,7 @@ func ReconcileMyComponent(r reconciler.Reconciler, ctx context.Context, cr *olsv
 }
 ```
 
-**Reference**: See `appserver/reconciler.go`, `postgres/reconciler.go`, or `lcore/reconciler.go`
+**Reference**: See `appserver/reconciler.go` or `postgres/reconciler.go`
 
 ### Step 3: Implement Asset Generation
 
@@ -70,7 +70,7 @@ func ReconcileMyComponent(r reconciler.Reconciler, ctx context.Context, cr *olsv
 - Always set owner references with `controllerutil.SetControllerReference()`
 - Use `utils.DefaultLabels()` for consistent labeling
 
-**Reference**: See `appserver/assets.go`, `postgres/assets.go`, or `lcore/assets.go`
+**Reference**: See `appserver/assets.go` or `postgres/assets.go`
 
 ### Step 4: Add Constants
 
@@ -139,7 +139,7 @@ The OLM bundle needs regeneration when changes affect how the operator is deploy
 **Bundle Update Required:**
 - **RBAC Changes**: Modified `//+kubebuilder:rbac` markers in Go code OR changed files in `config/rbac/`
 - **CRD Changes**: Modified API types in `api/v1alpha1/olsconfig_types.go`
-- **Image Changes**: New operator image, new operand images (appserver, lcore, postgres, console), or image version changes
+- **Image Changes**: New operator image, new operand images (appserver, postgres, console), or image version changes
 - **CSV Metadata**: Changed operator description, keywords, maintainers, links, or other metadata
 
 **Bundle Update NOT Required:**

diff --git a/api/v1alpha1/olsconfig_types.go b/api/v1alpha1/olsconfig_types.go
@@ -21,7 +21,6 @@ import (
 	corev1 "k8s.io/api/core/v1"
 	resource "k8s.io/apimachinery/pkg/api/resource"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	runtime "k8s.io/apimachinery/pkg/runtime"
 )
 
 // OLSConfigSpec defines the desired state of OLSConfig
@@ -343,9 +342,6 @@ type DeploymentConfig struct {
 	// MCP server container settings.
 	// +operator-sdk:csv:customresourcedefinitions:type=spec,displayName="MCP Server Container"
 	MCPServerContainer ContainerConfig `json:"mcpServer,omitempty"`
-	// Llama Stack container settings.
-	// +operator-sdk:csv:customresourcedefinitions:type=spec,displayName="Llama Stack Container"
-	LlamaStackContainer ContainerConfig `json:"llamaStack,omitempty"`
 	// Console container settings.
 	// +operator-sdk:csv:customresourcedefinitions:type=spec,displayName="Console Deployment"
 	ConsoleContainer Config `json:"console,omitempty"`
@@ -482,10 +478,6 @@ type ModelSpec struct {
 // ProviderSpec defines the desired state of LLM provider.
 // +kubebuilder:validation:XValidation:message="'deploymentName' must be specified for 'azure_openai' provider",rule="self.type != \"azure_openai\" || self.deploymentName != \"\""
 // +kubebuilder:validation:XValidation:message="'projectID' must be specified for 'watsonx' provider",rule="self.type != \"watsonx\" || self.projectID != \"\""
-// +kubebuilder:validation:XValidation:message="'providerType' and 'config' must be used together in llamaStackGeneric mode",rule="!has(self.providerType) || has(self.config)"
-// +kubebuilder:validation:XValidation:message="'config' requires 'providerType' to be set",rule="!has(self.config) || has(self.providerType)"
-// +kubebuilder:validation:XValidation:message="Llama Stack Generic mode (providerType set) requires type='llamaStackGeneric'",rule="!has(self.providerType) || self.type == \"llamaStackGeneric\""
-// +kubebuilder:validation:XValidation:message="Llama Stack Generic mode cannot use legacy provider-specific fields",rule="self.type != \"llamaStackGeneric\" || (!has(self.deploymentName) && !has(self.projectID) && !has(self.url) && !has(self.apiVersion))"
 // +kubebuilder:validation:XValidation:message="credentialKey must not be empty or whitespace",rule="!has(self.credentialKey) || !self.credentialKey.matches('^[ \\t\\n\\r\\v\\f]*$')"
 // +kubebuilder:validation:XValidation:message="googleVertexConfig is required for google_vertex provider",rule="self.type != \"google_vertex\" || has(self.googleVertexConfig)"
 // +kubebuilder:validation:XValidation:message="googleVertexAnthropicConfig is required for google_vertex_anthropic provider",rule="self.type != \"google_vertex_anthropic\" || has(self.googleVertexAnthropicConfig)"
@@ -515,7 +507,7 @@ type ProviderSpec struct {
 	// Provider type
 	// +kubebuilder:validation:Required
 	// +required
-	// +kubebuilder:validation:Enum=azure_openai;bam;openai;watsonx;rhoai_vllm;rhelai_vllm;fake_provider;llamaStackGeneric;google_vertex;google_vertex_anthropic
+	// +kubebuilder:validation:Enum=azure_openai;bam;openai;watsonx;rhoai_vllm;rhelai_vllm;fake_provider;google_vertex;google_vertex_anthropic
 	// +operator-sdk:csv:customresourcedefinitions:type=spec,displayName="Provider Type"
 	Type string `json:"type"`
 	// Deployment name for Azure OpenAI provider
@@ -540,24 +532,10 @@ type ProviderSpec struct {
 	// +kubebuilder:validation:Optional
 	// +operator-sdk:csv:customresourcedefinitions:type=spec,displayName="TLS Security Profile",xDescriptors={"urn:alm:descriptor:com.tectonic.ui:advanced"}
 	TLSSecurityProfile *configv1.TLSSecurityProfile `json:"tlsSecurityProfile,omitempty"`
-	// Llama Stack Generic provider type for provider configuration (e.g., "remote::openai", "inline::sentence-transformers")
-	// When set, this provider uses Llama Stack Generic mode instead of legacy mode.
-	// Must follow pattern: (inline|remote)::<provider-name>
-	// +kubebuilder:validation:Pattern=`^(inline|remote)::[a-z0-9][a-z0-9_-]*$`
-	// +operator-sdk:csv:customresourcedefinitions:type=spec,displayName="Llama Stack Provider Type"
-	ProviderType string `json:"providerType,omitempty"`
-	// Arbitrary configuration for the provider (Llama Stack Generic mode only)
-	// This map is passed directly to Llama Stack provider configuration.
-	// Credentials are automatically injected as environment variable substitutions.
-	// Example: {"url": "https://...", "custom_field": "value"}
-	// +kubebuilder:pruning:PreserveUnknownFields
-	// +operator-sdk:csv:customresourcedefinitions:type=spec,displayName="Llama Stack Provider Config"
-	Config *runtime.RawExtension `json:"config,omitempty"`
 	// Secret key name for provider credentials (defaults to "apitoken" if not set).
 	// Specifies which key inside credentialsSecretRef to read the credential value from.
-	// The credential value is always exposed to the container as env var {PROVIDER_NAME}_API_KEY
-	// (derived from the provider name, not this field), and referenced in the Llama Stack config
-	// YAML as ${env.PROVIDER_NAME_API_KEY}. This field only controls which secret data key is read.
+	// The credential value is exposed to the app server container as env var {PROVIDER_NAME}_API_KEY
+	// (derived from the provider name, not this field). This field only controls which secret data key is read.
 	// +kubebuilder:validation:Optional
 	// +operator-sdk:csv:customresourcedefinitions:type=spec,displayName="Credential Key Name"
 	CredentialKey string `json:"credentialKey,omitempty"`

diff --git a/api/v1alpha1/zz_generated.deepcopy.go b/api/v1alpha1/zz_generated.deepcopy.go
diff --git a/bundle/manifests/lightspeed-operator.clusterserviceversion.yaml b/bundle/manifests/lightspeed-operator.clusterserviceversion.yaml
@@ -93,19 +93,11 @@ spec:
           - description: API Version for Azure OpenAI provider
             displayName: Azure OpenAI API Version
             path: llm.providers[0].apiVersion
-          - description: |-
-              Arbitrary configuration for the provider (Llama Stack Generic mode only)
-              This map is passed directly to Llama Stack provider configuration.
-              Credentials are automatically injected as environment variable substitutions.
-              Example: {"url": "https://...", "custom_field": "value"}
-            displayName: Llama Stack Provider Config
-            path: llm.providers[0].config
           - description: |-
               Secret key name for provider credentials (defaults to "apitoken" if not set).
               Specifies which key inside credentialsSecretRef to read the credential value from.
-              The credential value is always exposed to the container as env var {PROVIDER_NAME}_API_KEY
-              (derived from the provider name, not this field), and referenced in the Llama Stack config
-              YAML as ${env.PROVIDER_NAME_API_KEY}. This field only controls which secret data key is read.
+              The credential value is exposed to the app server container as env var {PROVIDER_NAME}_API_KEY
+              (derived from the provider name, not this field). This field only controls which secret data key is read.
             displayName: Credential Key Name
             path: llm.providers[0].credentialKey
           - description: Deployment name for Azure OpenAI provider
@@ -156,12 +148,6 @@ spec:
           - description: Watsonx Project ID
             displayName: Watsonx Project ID
             path: llm.providers[0].projectID
-          - description: |-
-              Llama Stack Generic provider type for provider configuration (e.g., "remote::openai", "inline::sentence-transformers")
-              When set, this provider uses Llama Stack Generic mode instead of legacy mode.
-              Must follow pattern: (inline|remote)::<provider-name>
-            displayName: Llama Stack Provider Type
-            path: llm.providers[0].providerType
           - description: TLS Security Profile used by connection to provider
             displayName: TLS Security Profile
             path: llm.providers[0].tlsSecurityProfile
@@ -271,9 +257,6 @@ spec:
             path: ols.deployment.database.replicas
             x-descriptors:
               - urn:alm:descriptor:com.tectonic.ui:podCount
-          - description: Llama Stack container settings.
-            displayName: Llama Stack Container
-            path: ols.deployment.llamaStack
           - description: MCP server container settings.
             displayName: MCP Server Container
             path: ols.deployment.mcpServer
@@ -891,9 +874,6 @@ spec:
                       - --metrics-bind-address=:8443
                       - --secure-metrics-server
                       - --cert-dir=/etc/tls/private
-                      - --lcore-image=quay.io/lightspeed-core/lightspeed-stack:dev-latest
-                      - --use-lcore=false
-                      - --lcore-server=true
                       - --service-image=registry.redhat.io/openshift-lightspeed/lightspeed-service-api-rhel9@sha256:5287134ee84c4837e74db54a36f9abe71dcbd46d067b04d99bfaa972a4f149cd
                       - --console-image=registry.redhat.io/openshift-lightspeed/lightspeed-console-plugin-rhel9@sha256:c613e471cd3b77fb85dacb337b5b9c6b9fddd06563485b0189809690b142dd4f
                       - --console-image-pf5=registry.redhat.io/openshift-lightspeed/lightspeed-console-plugin-pf5-rhel9@sha256:86ac43e54a7d121762265cfbb9b32d3b75226e75ef25abaff2adb22c1171f00a