Advanced Deployment

Community examples and templates for various deployment methods.

Ansible

We recommend using the community.beszel Ansible Collection.

• GitHub repository
• Ansible Galaxy
• Agent role documentation
• Hub role documentation
• Module documentation

Thanks to dbrennand and all contributors for maintaining this collection.

For other useful examples see this discussion on GitHub.

Docker Swarm

0.12.0 Update

This guide was written prior to the introduction of universal tokens and agent-initiated WebSocket connections.

It should now be simpler to deploy agents in cluster environments. Feel free to share feedback or updated examples on our GitHub Discussions page.

The recommended approach is to define each agent separately and constrain it to a unique host / port.

For more info please search our GitHub issues for "swarm" or see examples by aeoneros:

https://wiki.aeoneros.com/books/beszel/page/quickstart-guide

x-common-config: &common-config
  image: henrygd/beszel-agent:latest
  restart: unless-stopped
  network_mode: host
  volumes:
    - /var/run/docker.sock:/var/run/docker.sock:ro
  environment:
    KEY: 'YOUR_PUBLIC_KEY_FROM_HUB'
  deploy: &common-deploy
    mode: replicated
    replicas: 1

services:
  beszel-agent1:
    <<: *common-config
    ports:
      - 45876:45876
    environment:
      <<: *common-config.environment
      LISTEN: '45876'
    deploy:
      <<: *common-deploy
      placement:
        constraints:
          - node.hostname == host-one

  beszel-agent2:
    <<: *common-config
    ports:
      - 45877:45877
    environment:
      <<: *common-config.environment
      LISTEN: '45877'
    deploy:
      <<: *common-deploy
      placement:
        constraints:
          - node.hostname == host-two

HashiCorp Nomad

0.12.0 Update

This guide was written prior to the introduction of universal tokens and agent-initiated WebSocket connections.

It should now be simpler to deploy agents in cluster environments. Feel free to share feedback or updated examples on our GitHub Discussions page.

An example Nomad configuration can be found in the article below by blinkinglight:

https://dev.to/blinkinglight/tailscale-and-beszel-on-hashicorp-nomad-1jmo

Kubernetes

0.12.0 Update

This guide was written prior to the introduction of universal tokens and agent-initiated WebSocket connections.

It should now be simpler to deploy agents in cluster environments. Feel free to share feedback or updated examples on our GitHub Discussions page.

Source discussion

The following was shared by nodesocket in this discussion on GitHub.

Hub

Use a standard Kubernetes Deployment and expose via a loadBalancer service or ingress. Nothing special is really needed for the hub.

Agents

Use a Kubernetes DaemonSet (deploys exactly one pod on every node automatically) and then add tolerations to allow deploying to the master/control-plane nodes. Finally, a few additional networking attributes such as hostNetwork and ports are needed.

Example Agent DaemonSet YAML

apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: beszel-agent
  namespace: default
spec:
  selector:
    matchLabels:
      app: beszel-agent
  template:
    metadata:
      labels:
        app: beszel-agent
    spec:
      hostNetwork: true
      containers:
        - env:
            - name: LISTEN
              value: "45876"
            - name: KEY
              value: "YOUR-KEY-HERE"
          image: henrygd/beszel-agent:latest
          imagePullPolicy: Always
          name: beszel-agent
          ports:
            - containerPort: 45876
              hostPort: 45876
      restartPolicy: Always
      tolerations:
        - effect: NoSchedule
          key: node-role.kubernetes.io/master
          operator: Exists
        - effect: NoSchedule
          key: node-role.kubernetes.io/control-plane
          operator: Exists
  updateStrategy:
    rollingUpdate:
      maxSurge: 0
      maxUnavailable: 100%
    type: RollingUpdate

Adding Systems to Beszel

Since we are using hostNetwork: true you use the Kubernetes node IP address when adding the system. Note: This is NOT the internal Kubernetes IP but the physical IP of the node itself. Each Kubernetes node only runs a single agent pod thus why this works.