Metallb can be used in Static (Bare Metal) clusters when there is no option to use cloud load balancers. Metallb can work in L2 or BGP modes.

Example of Metallb usage in L2 mode

Below is a small step-by-step guide on how to enable the metallb module, create a LoadBalancer and provide access to an application (Nginx web server).

  1. Specify node groups (NodeGroup) to run the applications to provide access to.

    For example, Ingress controllers are run on frontend nodes while the application is run on a worker node. Frontend nodes have a label node-role/metallb="".

    apiVersion: deckhouse.io/v1
kind: NodeGroup
metadata:
  name: frontend
spec:
  disruptions:
    approvalMode: Manual
  nodeTemplate:
    labels:
      node-role.deckhouse.io/metallb: ""
    taints:
    - effect: NoExecute
      key: dedicated.deckhouse.io
      value: metallb
  nodeType: Static
---
apiVersion: deckhouse.io/v1
kind: NodeGroup
metadata:
  name: worker
spec:
  disruptions:
    approvalMode: Manual
  nodeType: Static

  2. Make sure the nodes are labeled correctly:

    kubectl get nodes -l node-role.deckhouse.io/metallb=

    Your output should look something like this:

    $ kubectl get nodes -l node-role/metallb
NAME              STATUS   ROLES      AGE   VERSION
demo-frontend-0   Ready    frontend   61d   v1.21.14
demo-frontend-1   Ready    frontend   61d   v1.21.14

  3. Enable the metallb module and set the nodeSelector and tolerations parameters for the MetalLB speakers.

    Below is an example of the module configuration:

    apiVersion: deckhouse.io/v1alpha1
kind: ModuleConfig
metadata:
  name: metallb
spec:
  version: 1
  enabled: true
  settings:
    addressPools:
    - addresses:
      - 192.168.199.100-192.168.199.102
      name: frontend-pool
      protocol: layer2
    speaker:
      nodeSelector:
        node-role.deckhouse.io/metallb: ""
      tolerations:
      - effect: NoExecute
        key: dedicated.deckhouse.io
        operator: Equal
        value: metallb

  4. Deploy the application (nginx) and Service on port 8080:

    kubectl create deploy nginx --image=nginx
kubectl create svc loadbalancer nginx --tcp=8080:80

  5. Verify that the service has been created:

    kubectl get svc nginx

    Your output should look something like this:

    $ kubectl get svc nginx
NAME    TYPE           CLUSTER-IP     EXTERNAL-IP       PORT(S)          AGE
nginx   LoadBalancer   10.222.9.190   192.168.199.101   8080:31689/TCP   3m11s

  6. Check if the application is accessible.

    Example:

    $ curl -s -o /dev/null -w "%{http_code}" 192.168.199.101:8080
200