proxmox-vms

Proxmox VM Management

Use the proxmox-manager MCP server tools to manage VMs, containers, nodes, storage, and snapshots on a Proxmox VE cluster.

General Approach

• When a node is required and the user hasn't specified one, call pve_list_nodes first and ask which node to target — or list VMs across all nodes with pve_list_vms (omit node arg).
• For destructive actions (delete, force stop, rollback), always confirm with the user before proceeding.
• Long-running operations (clone, migrate) return a task ID — the user can track it with pve_list_tasks.

Common Workflows

Viewing the cluster

1. pve_cluster_status — overall health and quorum
2. pve_list_nodes — nodes with CPU/RAM usage
3. pve_list_vms — all VMs across all nodes (set include_containers: true to also see LXCs)

VM lifecycle

• Start: pve_start_vm
• Graceful shutdown: pve_stop_vm (uses ACPI — guest OS shuts down cleanly)
• Force stop: pve_kill_vm (hard power off — only if graceful fails)
• Reboot: pve_reboot_vm

Creating a VM from a template

1. Find a suitable template with pve_list_vms (templates show template: 1 in config)
2. Get next free ID: pve_next_vmid
3. Clone: pve_clone_vm with full: true and target storage
4. Update config if needed: pve_update_vm_config (set name, memory, CPU, onboot)
5. Start: pve_start_vm

Snapshots

• List: pve_list_snapshots
• Create: pve_create_snapshot (name must have no spaces; include description)
• Rollback: pve_rollback_snapshot (VM should be stopped first for clean rollback)
• Delete: pve_delete_snapshot

Migration

Use pve_migrate_vm to move a VM to another node. Set online: true for live migration (VM stays running — requires shared storage or same-storage setup).

Storage

• pve_list_storage — pools with free/used space
• pve_list_isos — ISO images and container templates on a storage

Best Practices

• Always snapshot before major changes (OS upgrades, config changes)
• Use onboot: true on production VMs so they auto-start after node reboot
• For Kubernetes nodes, provision with at least 2 CPUs and 2 GB RAM per node (see kubernetes-cluster skill)
• Use descriptive VM names and descriptions — they help when managing many VMs

Reference

See references/vm-config-fields.md for full VM configuration field reference.