Skip to main content
Version: latest


MetalLB is a load-balancer implementation for bare metal Kubernetes clusters that uses standard routing protocols. This integration is recommended for self-hosted clouds and helps external services obtain an IP address when the service type is set to LoadBalancer.

MetalLB deploys a controller and a speaker. The speaker is deployed as a DaemonSet on all nodes.

Versions Supported


  • A Kubernetes cluster with Kubernetes version 1.13.0 or later that does not already have network load-balancing functionality.

  • A cluster network configuration that does not conflict with MetalLB. For more information, refer to the official Kubernetes Cluster Networking documentation

  • Ensure sufficient IPv4 addresses for MetalLB are available for the number of services in each cluster.

  • When using the Border Gateway Protocol (BGP), one or more BGP-capable routers are required.

  • When using the Layer 2 (L2) operating mode, Transmission Control Protocol (TCP) and User Datagram Protocol (UDP) traffic on port 7946 must be allowed between nodes, as required by the HashiCorp memberlist. You can configure other port as needed.


The addresses parameter applies to the manifest-based MetalLB pack. You can provide multiple entries, but usually only one is needed.

addressesThis can be a range of addresses or a CIDR address. Examples:


The lb-metallb manifest-based pack supports direct configuration of an L2 IP address set. The lb-metallb-helm Helm-based pack provides an L2 address pool.


Manifest-based MetalLB supports direct configuration of an L2 IP address set. You can set either a range of addresses or use CIDR format, such as A more advanced MetalLB configuration, such as Border Gateway Protocol (BGP) routing requires you to write your own manifests and add them to the Palette cluster profile.

The example shows the syntax used to set an address range.

controller: ""
speaker: ""
namespace: "metallb-system"
avoidBuggyIps: true

Helm Chart

Helm-based MetalLB default gives you an L2 address pool by default. It has two sections, charts:metallb-full:configuration and charts:metallb-full:metallb, as shown in the following examples.

Use the charts:metallb-full:configuration parameter section to set resource types that MetalLB supports. The pack default gives you an L2 address pool. To set up a more advanced scenario, you can use the other resource types provided in the pack. The pack includes a commented-out example for each resource.

avoidBuggyIPs: true
autoAssign: true

- first-pool

bgpadvertisements: {}
bgppeers: {}
communities: {}
bfdprofiles: {}

The charts:metallb-full:metallb parameter section provides access to all the options of the base chart that installs MetalLB. You don’t need to change anything unless you want to install MetalLB in Free Range Routing (FRR) mode. To use FRR mode, set the option to true, as the example shows.

enabled: true

Airgap Palette and VerteX

In the context of an airgap Palette or VerteX installation, you must add the following labels to the MetalLB namespace. These labels allow the speaker pods to come up successfully. Otherwise, depending on the Kubernetes version, the speaker pods may get blocked by security policies.

  • privileged
  • privileged
  • privileged

The following example shows how to update the pack YAML with the required labels.

- image:
- image:
- image:
- repo:
name: metallb
version: 0.13.11
namespace: metallb-system
.spectro.system.kubernetes.version | substr 0 4 }}"

Refer to the Profile Customization page to learn more about additional namespace labels and annotations.


If controller and speaker pods are not assigning new IP addresses that you provided in the MetalLB pack, it is likely pods in existing deployments do not have the latest configMap file.

IP addresses you specify in MetalLB pack values go into a configMap called config in the metallb-system namespace. The MetalLB controller and speakers use the configMap as a volume mount.

Any changed IP addresses will get updated in the configMap. You can confirm this by issuing the following command.

kubectl describe cm config --namespace metallb-system

Since the controller and speaker pods are using a previous copy of the configMap, existing deployments are unaware of the changes made to configMap.

To ensure updated addresses are reflected in the configMap, you need to restart the controller and speaker pods so they fetch the new configMap and start assigning new addresses correctly. Issue the commands below to restart the controller and speaker.

kubectl rollout restart deploy controller --namespace metallb-system
kubectl rollout restart ds speaker --namespace metallb-system


data "spectrocloud_pack" "MetalLB" {
name = "lb-metallb"
version = "0.13.5"
data "spectrocloud_pack" "MetalLB-Helm" {
name = "lb-metallb-helm"
version = "0.13.7"