Install KubeLB Manager and setup Management Cluster

Requirements

  • Service type LoadBalancer implementation. This can be a cloud solution or a self-managed implementation like MetalLB.
  • Network access to the tenant cluster nodes with node port range (default: 30000-32767). This is required for the envoy proxy to be able to connect to the tenant cluster nodes.

Installation for KubeLB manager

In case if Gateway API needs to be enabled for the cluster. Please set kubelb.enableGatewayAPI to true in the values.yaml. This is required otherwise due to missing CRDs, kubelb will not be able to start.

Prerequisites

  • Create a namespace kubelb for the CCM to be deployed in.
  • Create imagePullSecrets for the chart to pull the image from the registry in kubelb namespace.

At this point a minimal values.yaml should look like this:

imagePullSecrets:
  - name: <imagePullSecretName>

Install the helm chart

helm pull oci://quay.io/kubermatic/helm-charts/kubelb-manager-ee --version=v1.2.0 --untardir "." --untar
## Apply CRDs
kubectl apply -f kubelb-manager-ee/crds/
## Create and update values.yaml with the required values.
helm upgrade --install kubelb-manager kubelb-manager-ee --namespace kubelb -f kubelb-manager-ee/values.yaml --create-namespace

KubeLB Manager EE Values

KeyTypeDefaultDescription
affinityobject{}
autoscaling.enabledboolfalse
autoscaling.maxReplicasint10
autoscaling.minReplicasint1
autoscaling.targetCPUUtilizationPercentageint80
autoscaling.targetMemoryUtilizationPercentageint80
cert-manager.enabledboolfalseEnable cert-manager.
external-dns.enabledboolfalseEnable External-DNS.
fullnameOverridestring""
image.pullPolicystring"IfNotPresent"
image.repositorystring"quay.io/kubermatic/kubelb-manager-ee"
image.tagstring"v1.2.0"
imagePullSecrets[0].namestring"kubermatic-quay.io"
kkpintegration.rbacboolfalseCreate RBAC for KKP integration.
kubelb.debugbooltrue
kubelb.enableGatewayAPIboolfalseenableGatewayAPI specifies whether to enable the Gateway API and Gateway Controllers. By default Gateway API is disabled since without Gateway APIs installed the controller cannot start.
kubelb.enableLeaderElectionbooltrue
kubelb.enableTenantMigrationbooltrue
kubelb.envoyProxy.affinityobject{}
kubelb.envoyProxy.nodeSelectorobject{}
kubelb.envoyProxy.replicasint2The number of replicas for the Envoy Proxy deployment.
kubelb.envoyProxy.resourcesobject{}
kubelb.envoyProxy.singlePodPerNodebooltrueDeploy single pod per node.
kubelb.envoyProxy.tolerationslist[]
kubelb.envoyProxy.topologystring"shared"Topology defines the deployment topology for Envoy Proxy. Valid values are: shared and global.
kubelb.envoyProxy.useDaemonsetboolfalseUse DaemonSet for Envoy Proxy deployment instead of Deployment.
kubelb.propagateAllAnnotationsboolfalsePropagate all annotations from the LB resource to the LB service.
kubelb.propagatedAnnotationsobject{}Allowed annotations that will be propagated from the LB resource to the LB service.
kubelb.skipConfigGenerationboolfalseSet to true to skip the generation of the Config CR. Useful when the config CR needs to be managed manually.
kubelb.tunnel.connectionManager.affinityobject{}
kubelb.tunnel.connectionManager.healthCheck.enabledbooltrue
kubelb.tunnel.connectionManager.healthCheck.livenessInitialDelayint30
kubelb.tunnel.connectionManager.healthCheck.readinessInitialDelayint10
kubelb.tunnel.connectionManager.httpAddrstring":8080"Server addresses
kubelb.tunnel.connectionManager.httpRoute.annotationsobject{"cert-manager.io/cluster-issuer":"letsencrypt-prod","external-dns.alpha.kubernetes.io/hostname":"connection-manager.${DOMAIN}"}Annotations for HTTPRoute
kubelb.tunnel.connectionManager.httpRoute.domainstring"connection-manager.${DOMAIN}"Domain for the HTTPRoute NOTE: Replace ${DOMAIN} with your domain name.
kubelb.tunnel.connectionManager.httpRoute.enabledboolfalse
kubelb.tunnel.connectionManager.httpRoute.gatewayNamestring"gateway"Gateway name to attach to
kubelb.tunnel.connectionManager.httpRoute.gatewayNamespacestring""Gateway namespace
kubelb.tunnel.connectionManager.imageobject{"pullPolicy":"IfNotPresent","repository":"quay.io/kubermatic/kubelb-connection-manager-ee","tag":""}Connection manager image configuration
kubelb.tunnel.connectionManager.ingressobject{"annotations":{"cert-manager.io/cluster-issuer":"letsencrypt-prod","external-dns.alpha.kubernetes.io/hostname":"connection-manager.${DOMAIN}","nginx.ingress.kubernetes.io/backend-protocol":"HTTP","nginx.ingress.kubernetes.io/proxy-read-timeout":"3600","nginx.ingress.kubernetes.io/proxy-send-timeout":"3600"},"className":"nginx","enabled":false,"hosts":[{"host":"connection-manager.${DOMAIN}","paths":[{"path":"/tunnel","pathType":"Prefix"},{"path":"/health","pathType":"Prefix"}]}],"tls":[{"hosts":["connection-manager.${DOMAIN}"],"secretName":"connection-manager-tls"}]}Ingress configuration for external HTTP/2 access
kubelb.tunnel.connectionManager.nodeSelectorobject{}
kubelb.tunnel.connectionManager.podAnnotationsobject{}Pod configuration
kubelb.tunnel.connectionManager.podLabelsobject{}
kubelb.tunnel.connectionManager.podSecurityContext.fsGroupint65534
kubelb.tunnel.connectionManager.podSecurityContext.runAsNonRootbooltrue
kubelb.tunnel.connectionManager.podSecurityContext.runAsUserint65534
kubelb.tunnel.connectionManager.replicaCountint1Number of connection manager replicas
kubelb.tunnel.connectionManager.requestTimeoutstring"30s"
kubelb.tunnel.connectionManager.resourcesobject{"limits":{"cpu":"500m","memory":"256Mi"},"requests":{"cpu":"250m","memory":"128Mi"}}Resource limits
kubelb.tunnel.connectionManager.securityContextobject{"allowPrivilegeEscalation":false,"capabilities":{"drop":["ALL"]},"readOnlyRootFilesystem":true,"runAsNonRoot":true,"runAsUser":65534}Security context
kubelb.tunnel.connectionManager.serviceobject{"httpPort":8080,"type":"ClusterIP"}Service configuration
kubelb.tunnel.connectionManager.tolerationslist[]
kubelb.tunnel.enabledboolfalseEnable tunnel functionality
nameOverridestring""
nodeSelectorobject{}
podAnnotationsobject{}
podLabelsobject{}
podSecurityContext.runAsNonRootbooltrue
podSecurityContext.seccompProfile.typestring"RuntimeDefault"
rbac.allowLeaderElectionRolebooltrue
rbac.allowMetricsReaderRolebooltrue
rbac.allowProxyRolebooltrue
rbac.enabledbooltrue
replicaCountint1
resources.limits.cpustring"500m"
resources.limits.memorystring"512Mi"
resources.requests.cpustring"100m"
resources.requests.memorystring"128Mi"
securityContext.allowPrivilegeEscalationboolfalse
securityContext.capabilities.drop[0]string"ALL"
securityContext.runAsUserint65532
service.portint8001
service.protocolstring"TCP"
service.typestring"ClusterIP"
serviceAccount.annotationsobject{}
serviceAccount.createbooltrue
serviceAccount.namestring""
serviceMonitor.enabledboolfalse
tolerationslist[]

Install the helm chart

helm pull oci://quay.io/kubermatic/helm-charts/kubelb-manager --version=v1.2.0 --untardir "." --untar
## Apply CRDs
kubectl apply -f kubelb-manager/crds/
## Create and update values.yaml with the required values.
helm upgrade --install kubelb-manager kubelb-manager --namespace kubelb -f kubelb-manager/values.yaml --create-namespace

KubeLB Manager CE Values

KeyTypeDefaultDescription
affinityobject{}
autoscaling.enabledboolfalse
autoscaling.maxReplicasint10
autoscaling.minReplicasint1
autoscaling.targetCPUUtilizationPercentageint80
autoscaling.targetMemoryUtilizationPercentageint80
fullnameOverridestring""
image.pullPolicystring"IfNotPresent"
image.repositorystring"quay.io/kubermatic/kubelb-manager"
image.tagstring"v1.2.0"
imagePullSecretslist[]
kkpintegration.rbacboolfalseCreate RBAC for KKP integration.
kubelb.debugbooltrue
kubelb.enableGatewayAPIboolfalseenableGatewayAPI specifies whether to enable the Gateway API and Gateway Controllers. By default Gateway API is disabled since without Gateway APIs installed the controller cannot start.
kubelb.enableLeaderElectionbooltrue
kubelb.enableTenantMigrationbooltrue
kubelb.envoyProxy.affinityobject{}
kubelb.envoyProxy.nodeSelectorobject{}
kubelb.envoyProxy.replicasint2The number of replicas for the Envoy Proxy deployment.
kubelb.envoyProxy.resourcesobject{}
kubelb.envoyProxy.singlePodPerNodebooltrueDeploy single pod per node.
kubelb.envoyProxy.tolerationslist[]
kubelb.envoyProxy.topologystring"shared"Topology defines the deployment topology for Envoy Proxy. Valid values are: shared and global.
kubelb.envoyProxy.useDaemonsetboolfalseUse DaemonSet for Envoy Proxy deployment instead of Deployment.
kubelb.propagateAllAnnotationsboolfalsePropagate all annotations from the LB resource to the LB service.
kubelb.propagatedAnnotationsobject{}Allowed annotations that will be propagated from the LB resource to the LB service.
kubelb.skipConfigGenerationboolfalseSet to true to skip the generation of the Config CR. Useful when the config CR needs to be managed manually.
nameOverridestring""
nodeSelectorobject{}
podAnnotationsobject{}
podLabelsobject{}
podSecurityContext.runAsNonRootbooltrue
podSecurityContext.seccompProfile.typestring"RuntimeDefault"
rbac.allowLeaderElectionRolebooltrue
rbac.allowMetricsReaderRolebooltrue
rbac.allowProxyRolebooltrue
rbac.enabledbooltrue
replicaCountint1
resources.limits.cpustring"500m"
resources.limits.memorystring"512Mi"
resources.requests.cpustring"100m"
resources.requests.memorystring"128Mi"
securityContext.allowPrivilegeEscalationboolfalse
securityContext.capabilities.drop[0]string"ALL"
securityContext.runAsUserint65532
service.portint8001
service.protocolstring"TCP"
service.typestring"ClusterIP"
serviceAccount.annotationsobject{}
serviceAccount.createbooltrue
serviceAccount.namestring""
serviceMonitor.enabledboolfalse
tolerationslist[]

Setup the management cluster

The examples and tools shared below are for demonstration purposes, you can use any other tools or configurations as per your requirements.

Management cluster acts as the dataplane and central control plane for all your load balancing configurations. It is the place where all the components required for Layer 4 and Layer 7 load balancing, AI Gateways, MCP Gateways, Agent2Agent Gateways, and API Gateways etc. are deployed. The management cluster is multi-tenant by design which makes it a perfect for managing a fleet of clusters in a scalable, robust, and secure way.

KubeLB has introduced an addons chart to simplify the installation of the required components for the management cluster. The chart is already part of the KubeLB manager chart and can be installed by setting the kubelb-addons.enabled to true in the values.yaml.

kubelb:
  enableGatewayAPI: true
  debug: true

## Addon configuration
kubelb-addons:
  enabled: true

  gatewayClass:
    create: true

  # Ingress Nginx
  ingress-nginx:
    enabled: false
    controller:
      service:
        externalTrafficPolicy: Local

  # Envoy Gateway
  envoy-gateway:
    enabled: true

  # Cert Manager
  cert-manager:
    enabled: true
    crds:
      enabled: true
    config:
      apiVersion: controller.config.cert-manager.io/v1alpha1
      kind: ControllerConfiguration
      enableGatewayAPI: true

  # External DNS
  external-dns:
    domainFilters:
      - example.com
    extraVolumes:
      - name: credentials
        secret:
          secretName: route53-credentials
    extraVolumeMounts:
      - name: credentials
        mountPath: /.aws
        readOnly: true
    env:
      - name: AWS_SHARED_CREDENTIALS_FILE
        value: /.aws/credentials
    txtOwnerId: kubelb-example-aws
    registry: txt
    provider: aws
    policy: sync
    sources:
      - service
      - ingress
      - gateway-httproute
      - gateway-grpcroute
      - gateway-tlsroute
      - gateway-tcproute
      - gateway-udproute

  ## AI and Agent2Agent Gateways Integration
  # KGateway CRDs
  kgateway-crds:
    enabled: true

  # KGateway
  kgateway:
    enabled: true
    gateway:
      aiExtension:
        enabled: true
    agentgateway:
      enabled: true

TCP/UDP Load Balancing (Layer 4)

Refer to Layer 4 Load Balancing Setup for more details.

Application Layer Load Balancing (Layer 7)

For Application layer load balancing, kubeLB supports both Ingress and Gateway API resources.

Our default recommendation is to use Gateway API and use Envoy Gateway as the Gateway API implementation. Most of the upcoming and current features that KubeLB will focus on will prioritize Gateway API instead of Ingress. With Envoy Gateway being the product that we’ll actively support, test, and base our features on.

While KubeLB supports integration with any Ingress or Gateway API implementation, the only limitation is that we only support native Kubernetes APIs i.e. Ingress and Gateway APIs. Provider specific APIs are not supported by KubeLB and will be completely ignored. Also, we are only testing KubeLB with Envoy Gateway and Nginx Ingress, we can’t guarantee the compatibility with other Gateway API or Ingress implementations.

Ingress

Refer to Ingress Setup for more details.

Gateway API

Refer to Gateway API Setup for more details.

Certificate Management(Enterprise Edition)

Refer to Certificate Management Setup for more details.

DNS Management(Enterprise Edition)

Refer to DNS Management Setup for more details.