v1.0.1

📋 Release Notes

🚀 New Features

Grafana Dashboard

• Added pre-built Grafana dashboard for comprehensive monitoring
• Dashboard sections include:
  • Controller Overview (Leader, Ready, Healthy pods, CRD count, IP stats)
  • Leader Election History timeline
  • CRD Status Table (Health, Reserved, Attached, Unattached IPs)
  • IP Allocation (Attachment rate gauge, IP-to-node mapping)
  • Node Status (Cordoned nodes, ip.ready label status)
  • Operations & Errors (Reconciliation rate, duration percentiles, API errors)
• ConfigMap provisioning support for Grafana sidecar auto-discovery
• Multi-cluster filtering with optional cluster variable

Optional Cluster Label for Metrics

• All metrics now support an optional cluster label for multi-cluster environments
• Set CLUSTER_NAME environment variable to enable cluster labeling
• Dashboard automatically detects and filters by cluster when available
• Works seamlessly without cluster label if not configured

Controller Health Metric

• Added netipallocation_controller_healthy gauge metric
• Non-leader pods now report ready=1 and healthy=1 (standby mode)
• Better visibility into all controller pod states

🔧 Improvements

• Metrics are now initialized on startup for all pods (leader and non-leader)
• Leader election status is updated immediately in lease_renewal_loop()
• Added pod_name parameter to reconcile_all() for proper metrics labeling
• Improved metric label consistency across all metrics

📊 Observability

• New Grafana dashboard ConfigMap for automatic provisioning
• Dashboard supports datasource variable selection
• All PromQL queries include optional cluster filtering (cluster=~"$cluster")

📝 Documentation

• Updated README with Grafana Dashboard section
• Added ConfigMap provisioning instructions
• Added dashboard preview diagram
• Updated Project Structure to match actual layout
• Added CLUSTER_NAME environment variable documentation
• Added PodMonitor example for Prometheus Operator
• Added troubleshooting for metrics and Grafana issues

⚠️ Breaking Changes

None

📦 Dependencies

No new dependencies

---

v1.1.0 (2025-12-15)

🚀 New Features

Prometheus Metrics Integration

• Added comprehensive Prometheus metrics for monitoring and alerting
• New metrics endpoint available at /metrics (port 9999)
• Metrics include:
  • netipallocation_crd_status - CRD health status
  • netipallocation_reserved_ips_total - Total reserved IPs per CRD
  • netipallocation_attached_ips_total - Currently attached IPs
  • netipallocation_unattached_ips_total - Unattached IPs
  • netipallocation_ip_attached - Per-IP attachment status
  • netipallocation_node_cordoned - Node cordon status
  • netipallocation_node_ip_ready - Node label status
  • netipallocation_controller_is_leader - Leader election status
  • netipallocation_controller_ready - Controller readiness
  • netipallocation_reconcile_total - Reconciliation counter
  • netipallocation_ip_attach_total - IP attach operations counter
  • netipallocation_ip_detach_total - IP detach operations counter
  • netipallocation_gcp_api_errors_total - GCP API error counter
  • netipallocation_reconcile_duration_seconds - Reconciliation latency histogram

Smart IP Reallocation for Cordoned Nodes

• Controller now detects cordoned nodes and reallocates IPs intelligently
• If a node is cordoned but deployment pods are still running, IP is preserved
• If a node is cordoned and no deployment pods are running, IP is detached and reallocated to a healthy node
• Prevents service disruption during node maintenance and cluster autoscaling

Enhanced Node Drain Detection

• Improved is_node_drained() function to be deployment-aware
• Skips system namespaces (kube-system, gke-system, istio-system)
• Skips DaemonSet pods when checking if node is drained
• Uses deploymentRef from CRD to check for specific deployment pods

🔧 Improvements

• Added deployment_ref parameter to detach_ip_from_node() for smarter detachment decisions
• Better logging with detailed context for debugging
• Added debug logs showing which nodes are being checked during reconciliation
• Improved error handling with GCP API error tracking

📊 Observability

• New PodMonitor support for Prometheus Operator
• Recommended PrometheusRule alerts included in documentation
• Example Prometheus queries provided

🐛 Bug Fixes

• Fixed variable shadowing issue where local node_cordoned variable conflicted with imported Prometheus metric
• Fixed IP not being reallocated when node is cordoned but has no running pods

📝 Documentation

• Updated README with comprehensive metrics documentation
• Added troubleshooting section
• Added architecture diagram
• Added cloud provider setup instructions

⚠️ Breaking Changes

None

📦 Dependencies

• Added prometheus-client>=0.19.0

---

v1.0.0 (2025-12-01)

🎉 Initial Release

Features

• Automatic static external IP allocation to Kubernetes nodes
• Support for GCP cloud provider
• Leader election using Kubernetes Lease
• Node labeling with ip.ready=true when IP is attached
• Pod eviction from misconfigured nodes
• Configurable reconciliation interval per CRD
• Health endpoints (/healthz, /readyz)

CRD Support

• NetIPAllocation custom resource for defining IP allocation rules
• Support for reservedIPs, deploymentRef, nodeSelector
• Cloud provider configuration (gcp, aws, azure)
• Strategy configuration (maxSurge, maxUnavailable)

---

Upgrade Guide

From v1.1.0 to v1.2.0

1. Update the controller image:

   kubectl set image deployment/ip-address-controller \
     controller=your-registry/ip-address-controller:v1.2.0

2. Optional: Set cluster name for multi-cluster metrics:

   env:
     - name: CLUSTER_NAME
       value: "gke-prod"  # or your cluster name

3. Optional: Deploy Grafana dashboard:

   kubectl apply -f k8s/grafana-dashboard-configmap.yaml

From v1.0.0 to v1.1.0

1. Update the controller image:

   kubectl set image deployment/ip-address-controller \
     controller=your-registry/ip-address-controller:v1.1.0

2. Add metrics port to deployment (if not present):

   ports:
     - name: health
       containerPort: 8080
     - name: metrics
       containerPort: 9999

3. Optional: Deploy PodMonitor for Prometheus Operator:

   kubectl apply -f k8s/podmonitor.yaml

4. Optional: Deploy PrometheusRule for alerts:

   kubectl apply -f k8s/prometheusrule.yaml

---

Roadmap

v1.3.0 (Planned)

• AWS support
• Azure support
• Webhook validation for CRD
• Status subresource for CRD

v1.4.0 (Planned)

• Multi-IP per node support
• IP pool management
• Automatic IP reservation from cloud provider
• Custom health checks before IP reallocation