Resilient Multiregion Global Control Planes With Crossplane and K8gb

This example demonstrates how to build resilient, scalable multicluster environments using Crossplane's declarative infrastructure provisioning integrated with k8gb for DNS-based failover.

Overview

This reference architecture showcases a Crossplane-based Global Control Plane with Active/Passive setup:

Key Components

Files

Usage

Local Testing 

# Render the composition locally
make run

# Or using crossplane render directly
crossplane render xr.yaml composition.yaml functions.yaml -r

Local Installation 

# In the main k8gb repo
make deploy-full-local-setup

# Check current context. We expect second k8gb cluster to be active after the
# installation
k config current-context
k3d-test-gslb2

# Install Crossplane
helm install crossplane \
--namespace crossplane-system \
--create-namespace crossplane-stable/crossplane

# Install Crossplane Providers
k apply -f crossplane-providers

# Wait for providers readiness
k get providers
NAME                            INSTALLED   HEALTHY   PACKAGE                                                 AGE
provider-azure-cache            True        True      xpkg.upbound.io/upbound/provider-azure-cache:v1         35s
provider-helm                   True        True      xpkg.upbound.io/upbound/provider-helm:v0                35s
provider-kubernetes             True        True      xpkg.upbound.io/upbound/provider-kubernetes:v0          35s
upbound-provider-family-azure   True        True      xpkg.upbound.io/upbound/provider-family-azure:v1.13.0   30s

# Install ProviderConfigs
k apply -f crossplane-providers/providerconfigs
providerconfig.azure.upbound.io/default created
providerconfig.helm.crossplane.io/default created
providerconfig.kubernetes.crossplane.io/default created

# Setup Azure Credentials Secret
kubectl create secret generic azure-account-creds -n crossplane-system --from-literal=credentials="$(cat credentials.json)" --dry-run=client -o yaml | kubectl apply -f -
# If you don't have Azure credentials json file, check Quickstart documentation
at https://marketplace.upbound.io/providers/upbound/provider-family-azure/latest

# Apply demo GlobalApp material
k apply -f definition.yaml,composition.yaml,functions.yaml
compositeresourcedefinition.apiextensions.crossplane.io/globalapps.example.crossplane.io created
Warning: CustomResourceDefinition.apiextensions.k8s.io "GlobalApp.example.crossplane.io" not found
composition.apiextensions.crossplane.io/global-app created
function.pkg.crossplane.io/crossplane-contrib-function-kcl created
function.pkg.crossplane.io/crossplane-contrib-function-auto-ready created

# Switch context to first k8gb cluster
k config use-context k3d-test-gslb1
Switched to context "k3d-test-gslb1".
# Repeat all installation steps in identical manner on the first cluster!

Deploy Examples 

# Deploy the basic XR
kubectl apply -f xr.yaml

# Deploy auto-mode XR with intelligent failover
kubectl apply -f xr-auto.yaml

# Deploy passive-mode XR for standby regions
kubectl apply -f xr-passive.yaml

Architecture Details

The KCL function implements the core failover monitoring logic:

  1. Observe GSLB: Creates Kubernetes Object to monitor GSLB resource health
  2. Health Assessment: Extracts serviceHealth status from k8gb
  3. Status Propagation: Updates XR status to reflect regional control plane health (Healthy/UNHEALTHY)
  4. Failover Enablement: Provides health data for automated Active/Passive transitions

The composition uses managementPolicies: ["Observe"] for read-only health monitoring without modifying the underlying GSLB resources, enabling safe observation across multiple regions.

Important Notes