[WIP] OTel+Prometheus sample and Grafana dashboard

.github/workflows/pr.yml

@@ -11,6 +11,7 @@ on:
     paths-ignore:
       - 'docs/**'
       - 'adr/**'
+      - 'observability/**'
   workflow_dispatch:
 jobs:
   check_format_and_unit_tests:

observability/README.md

@@ -0,0 +1,252 @@
+# Observability Stack for Java Operator SDK
+
+This directory contains the setup scripts and Grafana dashboards for monitoring Java Operator SDK applications.
+
+## Installation
+
+Run the installation script to deploy the full observability stack (OpenTelemetry Collector, Prometheus, and Grafana):
+
+```bash
+./install-observability.sh
+```
+
+This will install:
+- **cert-manager** - Required for OpenTelemetry Operator
+- **OpenTelemetry Operator** - Manages OpenTelemetry Collector instances
+- **OpenTelemetry Collector** - Receives OTLP metrics and exports to Prometheus
+- **Prometheus** - Metrics storage and querying
+- **Grafana** - Metrics visualization
+
+## Accessing Services
+
+### Grafana
+```bash
+kubectl port-forward -n observability svc/kube-prometheus-stack-grafana 3000:80
+```
+Then open http://localhost:3000
+- Username: `admin`
+- Password: `admin`
+
+### Prometheus
+```bash
+kubectl port-forward -n observability svc/kube-prometheus-stack-prometheus 9090:9090
+```
+Then open http://localhost:9090
+
+## Grafana Dashboards
+
+Two pre-configured dashboards are **automatically imported** during installation:
+
+### 1. JVM Metrics Dashboard (`jvm-metrics-dashboard.json`)
+
+Monitors Java Virtual Machine health and performance:
+
+**Panels:**
+- **JVM Memory Used** - Heap and non-heap memory consumption by memory pool
+- **JVM Threads** - Live, daemon, and peak thread counts
+- **GC Pause Time Rate** - Garbage collection pause duration
+- **GC Pause Count Rate** - Frequency of garbage collection events
+- **CPU Usage** - System CPU utilization percentage
+- **Classes Loaded** - Number of classes currently loaded
+- **Process Uptime** - Application uptime in seconds
+- **CPU Count** - Available processor cores
+- **GC Memory Allocation Rate** - Memory allocation and promotion rates
+- **Heap Memory Max vs Committed** - Heap memory limits and commitments
+
+**Key Metrics:**
+- `jvm.memory.used`, `jvm.memory.max`, `jvm.memory.committed`
+- `jvm.gc.pause`, `jvm.gc.memory.allocated`, `jvm.gc.memory.promoted`
+- `jvm.threads.live`, `jvm.threads.daemon`, `jvm.threads.peak`
+- `jvm.classes.loaded`, `jvm.classes.unloaded`
+- `system.cpu.usage`, `system.cpu.count`
+- `process.uptime`
+
+**Filtering:**
+All panels filter by `service_name="josdk"` to show metrics only from your operator.
+
+### 2. Java Operator SDK Metrics Dashboard (`josdk-operator-metrics-dashboard.json`)
+
+Monitors Kubernetes operator performance and health:
+
+**Panels:**
+- **Reconciliation Rate (Started)** - Rate of reconciliation loops triggered
+- **Reconciliation Success vs Failure Rate** - Success/failure ratio over time
+- **Currently Executing Reconciliations** - Active reconciliation threads
+- **Reconciliation Queue Size** - Pending reconciliation work
+- **Total Reconciliations** - Cumulative count of reconciliations
+- **Error Rate** - Overall error rate across all reconciliations
+- **Reconciliation Execution Time** - P50, P95, P99 latency percentiles
+- **Event Reception Rate** - Kubernetes event processing rate
+- **Failures by Exception Type** - Breakdown of errors by exception class
+- **Controller Execution Success vs Failure** - Controller-level success metrics
+- **Delete Event Rate** - Resource deletion event frequency
+- **Reconciliation Retry Rate** - Retry attempts and patterns
+
+**Key Metrics:**
+- `operator.sdk.reconciliations.started`, `.success`, `.failed`
+- `operator.sdk.reconciliations.executions` - Current execution count
+- `operator.sdk.reconciliations.queue.size` - Queue depth
+- `operator.sdk.controllers.execution.reconcile` - Execution timing histograms
+- `operator.sdk.events.received`, `.delete` - Event reception
+- Retry metrics and failure breakdowns
+
+**Filtering:**
+All panels filter by `service_name="josdk"` to show metrics only from your operator.
+
+## Importing Dashboards into Grafana
+
+### Automatic Import (Default)
+
+The dashboards are **automatically imported** when you run `./install-observability.sh`. They will appear in Grafana within 30-60 seconds after installation. No manual steps required!
+
+To verify the dashboards were imported:
+1. Access Grafana at http://localhost:3000
+2. Navigate to **Dashboards** → **Browse**
+3. Look for "JOSDK - JVM Metrics" and "JOSDK - Operator Metrics"
+
+### Manual Import Methods
+
+If you need to re-import or update the dashboards manually:
+
+#### Method 1: Via Grafana UI
+
+1. Access Grafana at http://localhost:3000
+2. Login with admin/admin
+3. Navigate to **Dashboards** → **Import**
+4. Click **Upload JSON file**
+5. Select `jvm-metrics-dashboard.json` or `josdk-operator-metrics-dashboard.json`
+6. Select **Prometheus** as the data source
+7. Click **Import**
+
+#### Method 2: Via kubectl ConfigMap
+
+```bash
+# Re-import JVM dashboard
+kubectl create configmap jvm-metrics-dashboard \
+  --from-file=jvm-metrics-dashboard.json \
+  -n observability \
+  -o yaml --dry-run=client | \
+  kubectl label --dry-run=client --local -f - grafana_dashboard=1 -o yaml | \
+  kubectl apply -f -
+
+# Re-import Operator dashboard
+kubectl create configmap josdk-operator-metrics-dashboard \
+  --from-file=josdk-operator-metrics-dashboard.json \
+  -n observability \
+  -o yaml --dry-run=client | \
+  kubectl label --dry-run=client --local -f - grafana_dashboard=1 -o yaml | \
+  kubectl apply -f -
+```
+
+The dashboards will be automatically discovered and loaded by Grafana within 30-60 seconds.
+
+## Configuring Your Operator
+
+To enable metrics export from your JOSDK operator, ensure your application:
+
+1. **Has the required dependency** (already included in webpage sample):
+   ```xml
+   <dependency>
+     <groupId>io.micrometer</groupId>
+     <artifactId>micrometer-registry-otlp</artifactId>
+   </dependency>
+   ```
+
+2. **Configures OTLP export** via `otlp-config.yaml`:
+   ```yaml
+   otlp:
+     url: "http://otel-collector-collector.observability.svc.cluster.local:4318/v1/metrics"
+     step: 15s
+     batchSize: 15000
+     aggregationTemporality: "cumulative"
+   ```
+
+3. **Registers JVM and JOSDK metrics** (see `WebPageOperator.java` for reference implementation)
+
+## OTLP Endpoints
+
+The OpenTelemetry Collector provides the following endpoints:
+
+- **OTLP gRPC**: `otel-collector-collector.observability.svc.cluster.local:4317`
+- **OTLP HTTP**: `otel-collector-collector.observability.svc.cluster.local:4318`
+- **Prometheus Scrape**: `http://otel-collector-prometheus.observability.svc.cluster.local:8889/metrics`
+
+## Troubleshooting
+
+### Check OpenTelemetry Collector Logs
+```bash
+kubectl logs -n observability -l app.kubernetes.io/name=otel-collector -f
+```
+
+### Check Prometheus Targets
+```bash
+kubectl port-forward -n observability svc/kube-prometheus-stack-prometheus 9090:9090
+```
+Open http://localhost:9090/targets and verify the OTLP collector target is UP.
+
+### Verify Metrics in Prometheus
+Open Prometheus UI and search for metrics:
+- JVM metrics: `jvm_*`
+- Operator metrics: `operator_sdk_*`
+
+### Check Grafana Data Source
+1. Navigate to **Configuration** → **Data Sources**
+2. Verify Prometheus data source is configured and working
+3. Click **Test** to verify connectivity
+
+## Uninstalling
+
+To remove the observability stack:
+
+```bash
+kubectl delete configmap -n observability jvm-metrics-dashboard josdk-operator-metrics-dashboard
+kubectl delete -n observability OpenTelemetryCollector otel-collector
+helm uninstall -n observability kube-prometheus-stack
+helm uninstall -n observability opentelemetry-operator
+helm uninstall -n cert-manager cert-manager
+kubectl delete namespace observability cert-manager
+```
+
+## Customizing Dashboards
+
+The dashboard JSON files can be modified to:
+- Add new panels for custom metrics
+- Adjust time ranges and refresh intervals
+- Change visualization types
+- Add templating variables for filtering
+- Modify alert thresholds
+
+After making changes, re-import the dashboard using one of the methods above.
+
+## Example Queries
+
+### JVM Metrics
+```promql
+# Heap memory usage percentage
+(jvm_memory_used_bytes{area="heap"} / jvm_memory_max_bytes{area="heap"}) * 100
+
+# GC throughput (percentage of time NOT in GC)
+100 - (rate(jvm_gc_pause_seconds_sum[5m]) * 100)
+
+# Thread count trend
+jvm_threads_live_threads
+```
+
+### Operator Metrics
+```promql
+# Reconciliation success rate
+rate(operator_sdk_reconciliations_success_total[5m]) / rate(operator_sdk_reconciliations_started_total[5m])
+
+# Average reconciliation time
+rate(operator_sdk_controllers_execution_reconcile_seconds_sum[5m]) / rate(operator_sdk_controllers_execution_reconcile_seconds_count[5m])
+
+# Queue saturation
+operator_sdk_reconciliations_queue_size / on() group_left() max(operator_sdk_reconciliations_queue_size)
+```
+
+## References
+
+- [Java Operator SDK Documentation](https://javaoperatorsdk.io)
+- [Micrometer OTLP Documentation](https://micrometer.io/docs/registry/otlp)
+- [OpenTelemetry Collector](https://opentelemetry.io/docs/collector/)
+- [Grafana Dashboards](https://grafana.com/docs/grafana/latest/dashboards/)