¶ Proxmox Cluster
¶ Summary
This page explains where core virtual machines run. In simple terms: the MS-01 systems are the main compute layer for lab services, while supporting nodes handle helper workloads.
The technical node map, storage relationship, and automation touchpoints are documented below.
¶ Current Cluster Shape
The Proxmox environment is centered on the three MS-01 nodes. These systems act as the primary virtualization and shared-storage platform for the lab.
| Host Group | Count | CPU | Memory | Primary Function |
|---|---|---|---|---|
ms-01 series |
3 | Intel Core i9-12900H | 32 GB, 32 GB, 64 GB | Proxmox supervisors, Ceph, clustered workloads |
| supporting utility nodes | 6 | M900 + Lenovo Neo mix | varies | Supporting compute and helper workloads outside the core Proxmox tier |
The current Proxmox host inventory is documented in Hardware. The VM layer running on top of these hosts is documented in VMs.
¶ What Proxmox Is Used For
- Supervising the core VM layer.
- Providing the base platform for Ceph-backed workloads.
- Hosting infrastructure VMs and supporting lab services.
- Serving as the automation target for VM creation, deletion, and hardware-specific setup via Ansible.
- Providing a compact cluster foundation after moving away from larger older rack servers.
¶ Storage Relationship
The Proxmox cluster is tightly coupled to the MS-01 Ceph deployment.
- RBD pool: for block-backed workloads.
- CephFS pool: for shared filesystem-style workloads.
- Per-node media: each MS-01 contributes two Samsung NVMe drives to the storage tier.
¶ Automation Touchpoints
The HomeLab Ansible repo contains the main Proxmox-related workflows, including:
- VM lifecycle playbooks.
- Ceph tuning playbooks.
- IOMMU and passthrough-related playbooks.
- The MS-01 Thunderbolt/USB4 ring configuration playbook.
- Ongoing update and package-management playbooks.
¶ Node placement notes
| Workload type | Placement idea |
|---|---|
| Core infrastructure VMs | Prefer MS-01 tier where Ceph and stronger compute are available. |
| Lightweight helper workloads | Use supporting Tiny/Neo systems where practical. |
| Game workloads | Prefer nodes sized for steady CPU/RAM demand and storage access. |
| GPU/transcode workloads | Use Docker node labels for Intel GPU capability. |
¶ Maintenance flow
- Check Ceph health and running workload placement.
- Migrate or drain workloads if needed.
- Apply Proxmox or host maintenance to one node at a time.
- Confirm the node rejoins cleanly.
- Confirm Swarm and monitoring agree that services recovered.
¶ Related pages
¶ Screenshot Placeholders
¶ Operational Notes
- The cluster is intentionally compact and power-efficient compared to the older large-server design.
- Shared storage and automation matter more than oversized standalone hosts.
- Supporting nodes can absorb lighter or secondary roles, while the MS-01 tier remains the primary Proxmox platform.