feat: add MetricsCollector interface, mock collector, e2e tests, and Helm NOTES.txt

- Extract MetricsCollector interface in pkg/gpu/collector.go for testability
- Add MockCollector in pkg/gpu/mock.go for GPU-free testing
- Update GPUExternalScaler to depend on interface instead of concrete type
- Add comprehensive e2e integration tests in tests/e2e/:
  - gRPC health check verification
  - IsActive with activation thresholds and profiles
  - GetMetricSpec for all configurations
  - GetMetrics across single/multi-GPU, all metric types
  - Scale-out/scale-in simulation
  - All pre-built profiles validation
- Add Helm chart NOTES.txt with post-install instructions
- Add test-e2e and helm-test Makefile targets

Signed-off-by: pmady <pavan4devops@gmail.com>

Author: pmady

Makefile

@@ -1,4 +1,4 @@
-.PHONY: build proto test lint clean docker-build docker-push docker-release deploy undeploy help
+.PHONY: build proto test test-e2e lint clean docker-build docker-push docker-release deploy undeploy helm-lint helm-template helm-test help
 
 BINARY_NAME := keda-gpu-scaler
 IMAGE_REPO := ghcr.io/pmady/keda-gpu-scaler
@@ -20,6 +20,9 @@ proto: ## Generate protobuf Go code
 test: ## Run unit tests
 	go test -v -race ./pkg/...
 
+test-e2e: ## Run e2e integration tests (no GPU required — uses mock collector)
+	go test -v -tags=e2e -race ./tests/e2e/...
+
 lint: ## Run linter
 	golangci-lint run ./...
 
@@ -52,3 +55,9 @@ helm-lint: ## Lint Helm chart
 
 helm-template: ## Render Helm templates
 	helm template keda-gpu-scaler deploy/helm/keda-gpu-scaler
+
+helm-test: ## Validate Helm chart renders correctly with default and custom values
+	helm lint deploy/helm/keda-gpu-scaler
+	helm template keda-gpu-scaler deploy/helm/keda-gpu-scaler > /dev/null
+	helm template keda-gpu-scaler deploy/helm/keda-gpu-scaler --set grpc.port=50051 --set logLevel=debug > /dev/null
+	@echo "Helm chart validation passed"

deploy/helm/keda-gpu-scaler/templates/NOTES.txt

@@ -0,0 +1,24 @@
+keda-gpu-scaler has been deployed as a DaemonSet on GPU nodes.
+
+  Service:  {{ include "keda-gpu-scaler.fullname" . }}.{{ .Release.Namespace }}.svc.cluster.local
+  gRPC Port: {{ .Values.grpc.port }}
+
+To use with KEDA, create a ScaledObject with an external trigger:
+
+  triggers:
+    - type: external
+      metadata:
+        scalerAddress: "{{ include "keda-gpu-scaler.fullname" . }}.{{ .Release.Namespace }}.svc.cluster.local:{{ .Values.grpc.port }}"
+        profile: "vllm-inference"
+
+Available profiles: vllm-inference, triton-inference, training, batch
+
+Verify the scaler pods are running on GPU nodes:
+
+  kubectl get pods -n {{ .Release.Namespace }} -l {{ include "keda-gpu-scaler.selectorLabels" . | replace ": " "=" | replace "\n" "," }}
+
+Check gRPC health:
+
+  kubectl exec -n {{ .Release.Namespace }} <pod-name> -- grpc_health_probe -addr=:{{ .Values.grpc.port }}
+
+Documentation: https://github.com/pmady/keda-gpu-scaler

pkg/gpu/collector.go

@@ -0,0 +1,26 @@
+/*
+Copyright 2026 The keda-gpu-scaler Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+package gpu
+
+// MetricsCollector is the interface for collecting GPU metrics.
+// The real implementation uses NVML; tests can inject a mock.
+type MetricsCollector interface {
+	CollectAll() ([]Metrics, error)
+	CollectDevice(index int) (Metrics, error)
+	DeviceCount() (int, error)
+	Close() error
+}

pkg/gpu/mock.go

@@ -0,0 +1,49 @@
+/*
+Copyright 2026 The keda-gpu-scaler Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+package gpu
+
+import "fmt"
+
+// MockCollector implements MetricsCollector with configurable GPU metrics.
+// Used for unit tests, integration tests, and e2e tests without GPU hardware.
+type MockCollector struct {
+	Devices []Metrics
+}
+
+// NewMockCollector creates a mock collector with the given device metrics.
+func NewMockCollector(devices []Metrics) *MockCollector {
+	return &MockCollector{Devices: devices}
+}
+
+func (m *MockCollector) CollectAll() ([]Metrics, error) {
+	return m.Devices, nil
+}
+
+func (m *MockCollector) CollectDevice(index int) (Metrics, error) {
+	if index < 0 || index >= len(m.Devices) {
+		return Metrics{}, fmt.Errorf("device index %d out of range (0-%d)", index, len(m.Devices)-1)
+	}
+	return m.Devices[index], nil
+}
+
+func (m *MockCollector) DeviceCount() (int, error) {
+	return len(m.Devices), nil
+}
+
+func (m *MockCollector) Close() error {
+	return nil
+}

pkg/scaler/server.go

@@ -32,12 +32,12 @@ import (
 // GPUExternalScaler implements the KEDA ExternalScaler gRPC interface.
 type GPUExternalScaler struct {
 	pb.UnimplementedExternalScalerServer
-	collector *gpu.Collector
+	collector gpu.MetricsCollector
 	logger    *zap.Logger
 }
 
 // NewGPUExternalScaler creates a new GPU external scaler server.
-func NewGPUExternalScaler(collector *gpu.Collector, logger *zap.Logger) *GPUExternalScaler {
+func NewGPUExternalScaler(collector gpu.MetricsCollector, logger *zap.Logger) *GPUExternalScaler {
 	return &GPUExternalScaler{
 		collector: collector,
 		logger:    logger,