Post

What Do I Do With 4 Prodesks

Posted
By Usman Masood Ashraf
views 5 min read

What Do I Do With 4 Prodesks

INTRODUCTION

You’ve just scored four HP ProDesk 600 G3 mini PCs – a sysadmin’s dream windfall. These compact workhorses (typically equipped with 6th/7th-gen Intel Core i5 processors, 8-32GB DDR4 RAM, and NVMe/SATA SSD support) present an ideal playground for building enterprise-grade infrastructure at home. But now comes the critical question: How do you transform this hardware quartet into a cohesive, production-resilient system that advances your DevOps skills?

Homelabs have become the proving ground for modern infrastructure strategies. A 2023 SysAdmin Survey by Spiceworks revealed that 68% of infrastructure professionals use personal labs to test technologies before deployment. With four identical nodes, you’re uniquely positioned to explore fault-tolerant architectures that mirror real-world environments – without cloud costs or corporate red tape.

In this comprehensive guide, we’ll transform your ProDesk fleet into:

  • A high-availability Proxmox VE cluster with Ceph distributed storage
  • A bare-metal Kubernetes lab using Talos Linux
  • An automated backup/replication target with ZFS
  • A distributed services platform for CI/CD pipelines and self-hosted apps

By the end, you’ll have a battle-tested environment for experimenting with infrastructure-as-code, container orchestration, and disaster recovery – using 100% open-source tooling.

UNDERSTANDING THE TOPIC

Homelab Architectures: From Single Node to Distributed Systems

Four identical nodes unlock cluster topologies that are impractical with heterogeneous hardware. Consider these configurations:

PatternNodes RequiredUse CaseProDesk Implementation
Active/Active HA3+Proxmox VE Cluster3 nodes for quorum + 1 backup
Primary/Replica2+PostgreSQL Streaming Replication2 DB nodes + 2 app nodes
Control/Worker Split4+Kubernetes1 control plane + 3 workers
Erasure Coding4+Ceph Object Storage4 OSD nodes

Technology Comparison: Virtualization vs Orchestration

Proxmox VE (Virtualization)

  • Strengths: Full hardware abstraction, live migration, ZFS/Ceph integration
  • Weaknesses: Overhead for containerized workloads, steeper RAM requirements
  • ProDesk Fit: Ideal when running mixed workloads (VMs + LXCs)

Kubernetes (Orchestration)

  • Strengths: Declarative infrastructure, cloud-native ecosystem, scalability
  • Weaknesses: Steep learning curve, hardware abstraction limitations
  • ProDesk Fit: Optimal for container-focused pipelines and microservices

Hybrid Approach

  • Proxmox hosts K8s VMs: Combines hardware flexibility with orchestration
  • Requires nested virtualization (enable in BIOS: Intel VT-x/AMD-V + IOMMU)

Real-World Cluster Sizing

Each ProDesk 600 G3 (i5-6500T, 32GB RAM, 512GB NVMe + 1TB SATA SSD) can handle:

  • Proxmox: 6-8 LXC containers or 3-4 lightweight VMs per node
  • Kubernetes: 15-20 pods/node (with 1GB RAM buffer for system)
  • Ceph: ~400MB/s OSD throughput (SATA SSD), 1GbE as bottleneck

PREREQUISITES

Hardware Preparation

  1. BIOS Configuration (All Nodes):
    1
2
3
4
5
6

    Advanced → Integrated Peripherals → 
  [X] SR-IOV Support
  [X] VT-d
  [X] AES-NI
Power Management → 
  [X] Wake on LAN
  2. Disk Layout (Recommended):
    1
2

    /dev/nvme0n1     # Proxmox OS / Kubernetes OS (128GB partition)
/dev/sda         # Ceph OSD / ZFS Mirror (1TB SSD)

Network Requirements

  • Switch: Managed L2+ with 802.1Q VLAN support (e.g., Ubiquiti USW-Lite-8)
  • IP Schema:
    1
2
3
4

    Node1: 192.168.10.11/24
Node2: 192.168.10.12/24
Node3: 192.168.10.13/24
Node4: 192.168.10.14/24
  • Firewall Rules:
    1
2
3
4
5
6
7
8

    # Proxmox Cluster Communication
ufw allow from 192.168.10.0/24 proto vrrp
ufw allow 8006/tcp  # Web UI
ufw allow 5404:5405/udp  # Corosync

# Kubernetes
ufw allow 6443/tcp  # Kube API
ufw allow 2379:2380/tcp  # etcd

Software Matrix

| Tool | Version | Purpose | |—————-|————-|———————————| | Proxmox VE | 8.1 | Virtualization Platform | | Talos Linux | 1.5 | Kubernetes OS | | Ceph Quincy | 17.2.6 | Distributed Storage | | k0s | 1.28 | Lightweight Kubernetes |

INSTALLATION & SETUP

Proxmox VE Cluster with Ceph

  1. Base Installation (Repeat per Node):
    1
2

    # Download ISO: https://www.proxmox.com/en/downloads
# Write to USB: dd if=proxmox-ve_8.1.iso of=/dev/sdX bs=4M status=progress
    • Partitioning: /dev/nvme0n1 as ext4, dev/sda as unused disk
    • Hostname: proxmox-01, IP: 192.168.10.11
  2. Cluster Initialization (First Node):
    1

    pvecm create PROXCLUSTER
  3. Join Subsequent Nodes:
    1

    pvecm add 192.168.10.11 -fingerprint [SHA-256]
  4. Ceph Deployment:
    1
2
3
4
5
6
7

    # On first node
pveceph install
pveceph init --network 192.168.10.0/24
pveceph createmon

# Add OSDs to each node
pveceph osd create /dev/sda
  5. Verify Cluster Status:
    1
2

    pvecm status
ceph -s

Bare-Metal Kubernetes with Talos

  1. Prepare Bootstrap Node (Proxmox VM):
    1
2
3
4
5

    qm create 9000 -name talos-bootstrap -memory 2048 -cores 2 \
  -net0 virtio,bridge=vmbr0 -scsihw virtio-scsi-pci
qm importdisk 9000 talos-amd64.iso local-lvm
qm set 9000 -scsi1 local-lvm:vm-9000-disk-0
qm start 9000
  2. Generate Machine Configs:
    1
2
3
4
5
6
7
8
9
10
11
12
13
14

    # controlplane.yaml
version: v1alpha1
machine:
  type: controlplane
  install:
    disk: /dev/nvme0n1
  network:
    interfaces:
      - interface: eno1
        dhcp: true
cluster:
  clusterName: prodesk-cluster
  controlPlane:
    endpoint: https://192.168.10.11:6443
  3. Apply Configuration:
    1

    talosctl apply-config -n 192.168.10.11 -f controlplane.yaml
  4. Bootstrap Cluster:
    1

    talosctl bootstrap -n 192.168.10.11

CONFIGURATION & OPTIMIZATION

Proxmox Tuning

  1. ZFS ARC Size Adjustment:
    1
2

    echo "options zfs zfs_arc_max=4294967296" > /etc/modprobe.d/zfs.conf
update-initramfs -u
  2. Ceph CRUSH Map Optimization:
    1
2

    ceph osd crush tunable minimal
ceph osd set-require-min-compat-client jewel
  3. VM CPU Pinning (For NUMA-aware ProDesks):
    1
2

    qm set 101 -cpuflags +pdpe1gb
qm set 101 -numa 1

Kubernetes Hardening

  1. Pod Security Admission: ```yaml

    psps.yaml

    apiVersion: apiserver.config.k8s.io/v1 kind: AdmissionConfiguration plugins:

    • name: PodSecurity configuration: apiVersion: pod-security.admission.config.k8s.io/v1 kind: PodSecurityConfiguration defaults: enforce: “restricted” enforce-version: “latest” ```
  2. Network Policies: ```yaml

    default-deny.yaml

    kind: NetworkPolicy apiVersion: networking.k8s.io/v1 metadata: name: default-deny spec: podSelector: {} policyTypes:

    • Ingress
    • Egress ```

USAGE & OPERATIONS

Proxmox Day-2 Operations

  1. Live Migration Test:
    1

    qm migrate 101 proxmox-02 --online
  2. Backup Automation:
    1
2
3
4
5

    # /etc/pve/vzdump.conf
tmpdir: /mnt/backups/tmp
storage: nas-backups
mode: snapshot
compress: lzo

Kubernetes Workload Deployment

  1. MetalLB Configuration: ```yaml apiVersion: metallb.io/v1beta1 kind: IPAddressPool metadata: name: prodesk-pool spec: addresses:
    • 192.168.10.200-192.168.10.220 ```
  2. Persistent Storage with Ceph CSI:
    1
2

    kubectl create -f https://raw.githubusercontent.com/ceph/ceph-csi/master/deploy/rbd/kubernetes/csi-config-map.yaml
kubectl create -f https://raw.githubusercontent.com/ceph/ceph-csi/master/deploy/rbd/kubernetes/csi-provisioner-rbac.yaml

TROUBLESHOOTING

Common Cluster Issues

  1. Proxmox Corosync Problems:
    1
2
3

    systemctl status corosync
corosync-cmapctl | grep members
pvecm expected 3 # Reset quorum
  2. Ceph HEALTH_WARN:
    1
2
3

    ceph health detail
ceph osd df tree
ceph osd pool autoscale-status
  3. Kubernetes Node Not Ready:
    1
2
3

    kubectl describe node proxmox-03
journalctl -u kubelet -n 100 --no-pager
talosctl logs -k kubelet

CONCLUSION

Four ProDesk 600 G3s provide an ideal balance between capability and manageability for serious homelab experimentation. By implementing either a Proxmox HA cluster or bare-metal Kubernetes environment (or both via nesting), you’ve created a platform to safely test:

  • Infrastructure-as-Code workflows with Terraform/Ansible
  • GitOps pipelines using Flux or Argo CD
  • Distributed storage systems like Ceph/GlusterFS
  • High-availability service topologies

To deepen your expertise:

The true value lies not in the hardware itself, but in the operational discipline gained from maintaining production-like systems on constrained resources – skills that directly translate to enterprise environments.

Open Source, Reddit Guides, Kubernetes
homelab sysadmin how-to tools kubernetes k8s docker containers
This post is licensed under CC BY 4.0 by the author.