Glossary

PCIe extension that prevents direct peer-to-peer traffic between devices, ensuring each device sits in its own IOMMU group. The ACS override kernel patch loosens this so devices in the same IOMMU group can be passed to different VMs — useful but a security trade-off.

See also: Add Controller or NVMe to VM

ZFS's in-memory read cache. Holds recently and frequently used data blocks; size scales with available host RAM. Defaults are aggressive — Proxmox post-install lets you cap it.

See also: Post-Install: Storage · ZFS commands

A Linux feature that exposes one part of the filesystem at another path. ProxMenux uses bind mounts to share host directories with LXC containers via pct set -mpN.

See also: LXC Mount Points

Combines multiple physical network interfaces into one logical interface — for redundancy (active-backup), throughput (LACP / 802.3ad) or both. Configured in /etc/network/interfaces.

A virtual layer-2 switch on the host. vmbr0 is the default Proxmox bridge that connects VMs and containers to the physical network. Multiple bridges (vmbr1, vmbr2...) isolate groups of guests.

See also: Bridge analysis & guided repair

Distributed storage system tightly integrated with Proxmox for hyper-converged clusters. Provides object, block and file storage backed by OSDs (Object Storage Daemons). PVE 9 requires Ceph 19.x (Squid).

See also: Upgrade PVE 8 to PVE 9

Short for "container" — Proxmox terminology for an LXC instance. Has a numeric CTID that maps to /etc/pve/lxc/<CTID>.conf. See also LXC.

See also: VM and CT Management

Modern format for APT repository sources, using .sources files with structured key/value blocks (Types, URIs, Suites, Components, Signed-By). Replaces the legacy single-line .list format. Standard from PVE 9 / Debian Trixie onward.

See also: Upgrade PVE 8 to PVE 9

A standalone graphics card on its own PCIe slot (e.g. NVIDIA RTX, AMD Radeon). Contrasts with iGPU. Typically used for VFIO passthrough to a single VM, not for sharing across LXCs.

See also: Add GPU to VM (Passthrough)

ncurses-based tool for drawing menus, prompts and progress bars in a terminal. ProxMenux is built around dialog menus.

Framework that automatically rebuilds out-of-tree kernel modules (NVIDIA, ZFS, custom drivers) when the kernel is upgraded. Avoids modules being left behind after kernel updates.

See also: Install Coral TPU (Host)

Intrusion prevention service that watches log files for repeated authentication failures and bans the offending IPs at the firewall layer. ProxMenux ships pre-configured jails for SSH, the Proxmox UI and the Monitor.

See also: Fail2Ban

ASCII-art system info tool that runs on shell login (the "system summary banner"). Replaces the older neofetch; available as an opt-in option in ProxMenux Post-Install.

See also: Post-Install: Optional

Boot loader for most Linux distributions. Configuration lives in /etc/default/grub + /boot/grub/. Modified for IOMMU enablement, kernel parameters and EFI boot entries. Run update-grub after edits.

See also: GPU Passthrough commands

The classic Debian network configuration system that reads /etc/network/interfaces. ifupdown2 is the rewritten version Proxmox uses by default — supports live reload via ifreload -a.

Graphics processor built into the CPU package (Intel UHD / Iris, AMD Radeon Vega in Ryzen APUs). Typically shared across multiple LXC containers via /dev/dri rather than passed through to a single VM.

See also: Add GPU to LXC

Initial RAM filesystem the kernel uses at boot before the real root filesystem is mounted. Contains drivers needed early (storage, encryption, RAID). Rebuilt with update-initramfs -u -k all after VFIO, IOMMU or storage driver changes.

Hardware feature (Intel VT-d / AMD-Vi) that lets PCIe devices be safely assigned directly to guest VMs. Required for GPU and disk-controller passthrough. Must be enabled both in BIOS / UEFI and in the kernel command line (intel_iommu=on or amd_iommu=on).

See also: Post-Install: Virtualization · Add GPU to VM (Passthrough)

The smallest unit of devices that the IOMMU can isolate from each other. All devices in a group must be passed through together — usually one device, but multi-function cards or shared root-port devices land in the same group. Listed via find /sys/kernel/iommu_groups/ -type l | sort -V.

Network protocol that exposes block storage over IP — a remote disk appears as a local SCSI device. Registered as Proxmox storage via pvesm add iscsi.

See also: Add iSCSI target as Proxmox storage

Linux kernel module that turns the host into a hypervisor. Combined with QEMU for device emulation, it's the foundation Proxmox uses to run VMs.

Linux storage stack that abstracts physical disks into volume groups (VG) of logical volumes (LV). Proxmox uses LVM-thin pools for VM disks by default. Concise tools: pvs, vgs, lvs.

See also: Storage and Disks commands

OS-level virtualization technology — multiple isolated user-space instances share the host's kernel. Lighter and faster than VMs but less isolated. Proxmox manages LXC containers via the pct CLI.

See also: VM and CT Management

Open-source security auditor (by CISOfy) that scans the host and prints a hardening score plus concrete remediation hints. ProxMenux installs the latest from upstream GitHub rather than the (older) Debian package.

See also: Lynis

48-bit identifier burnt into a NIC's firmware. Used by Proxmox Persistent Network Names to pin interface names (eno1, enp3s0) so they survive PCI re-enumeration.

See also: Persistent interface names

Text shown after SSH login. Stored in /etc/motd. ProxMenux Post-Install can add a banner here; the uninstaller restores the backup if you applied it.

Unix-native network file sharing protocol. Registered as Proxmox storage via pvesm add nfs; also usable as an LXC client / server via dedicated ProxMenux flows.

See also: Storage & Share Manager

Linux firewall systems. nftables is the modern engine; iptables is the legacy front-end (still works as a translation layer). Proxmox's pve-firewall emits nftables rules from the cluster firewall config.

See also: Network commands

Free Proxmox repository for users without an enterprise subscription. Less polished QA than the enterprise repo (slightly riskier for production) but free. Set up automatically by the ProxMenux Post-Install.

See also: Post-Install: Basic Settings

Open-source NVIDIA driver bundled with the Linux kernel. Must be blacklisted before installing the proprietary NVIDIA driver, otherwise both load and the install fails.

See also: NVIDIA installation guide

DMTF-standard packaging formats for portable VMs. OVA is a single TAR archive; OVF is a directory with an XML descriptor + external VMDK files. Importable on VMware, VirtualBox and Proxmox.

See also: Export VM to OVA / OVF · Import VM from OVA / OVF

UEFI firmware implementation for VMs. Set with qm set <vmid> -bios ovmf. Required for Windows 11, Secure Boot, and most modern OS installs. Pair with an efidisk0 to persist EFI variables.

See also: Add GPU to VM (Passthrough)

Standalone product from Proxmox for incremental, deduplicated, encrypted VM / CT backups. Recommended over plain vzdump when you have a separate machine to host it on.

CLI tool for managing LXC containers in Proxmox: pct list / start / stop / shutdown / config / enter / exec / push / pull / destroy.

See also: VM and CT Management

FUSE-based virtual filesystem mounted at /etc/pve/. Replicates configuration changes across all cluster nodes in real time. Editing a file under /etc/pve/ propagates automatically.

LXC container where the root user inside the container is root on the host (UIDs not shifted). Required for hardware passthrough (USB devices, GPU sharing) and Samba / NFS server inside the CT. Less secure than unprivileged — use only when needed.

Optional web dashboard installed alongside ProxMenux. Serves real-time host stats (CPU / RAM / disk / network), VM and LXC overview, login + 2FA, on TCP port 8008.

See also: ProxMenux Monitor

CLI for managing the local catalog of CT templates. pveam update refreshes the index; pveam available lists templates; pveam download <storage> <template> downloads one.

The HTTPS service that serves the Proxmox web UI on port 8006 and the web terminal. Restarted automatically during major upgrades — which is why ProxMenux refuses to run the PVE 8 → 9 upgrade from the web terminal.

CLI tool for managing Proxmox storage. pvesm status shows all configured storages; pvesm add <type> registers new ones (NFS, Samba, iSCSI, dir, ZFS, LVM, …).

See also: Storage and Disks commands

Modern QEMU machine type emulating an Intel Q35 chipset with PCIe support. Recommended over the older i440FX for VMs that need PCIe passthrough or modern guest OS support. Set with qm set <vmid> -machine q35.

Default VM disk image format on Proxmox directory storage. Supports thin provisioning, snapshots, and copy-on-write. Convert with qemu-img convert -O qcow2 ....

Generic, open-source machine emulator and virtualizer. Combined with KVM in Proxmox to run VMs at near-native speed.

CLI tool for managing VMs in Proxmox: qm list / start / stop / shutdown / config / set / destroy / importdisk / migrate. Sister command of pct for containers.

See also: VM and CT Management

Command-line tool that talks to ~50 cloud storage providers (Google Drive, Dropbox, OneDrive, Mega, S3, …). Used in the Proxmox cloud-backup guide to mount a personal cloud as a directory the host can write vzdump archives to.

See also: Backup to Personal Cloud

Microsoft's file-sharing protocol family. SMB is the protocol; Samba is the open-source Linux implementation; CIFS is an older SMB dialect. Registered as Proxmox storage via pvesm add cifs.

See also: Storage & Share Manager

Open-source legacy BIOS implementation used by QEMU. Default firmware for new VMs unless you switch to OVMF. Suitable for older OSes and simple Linux installs that don't need UEFI.

Built-in disk diagnostics that report drive health, error counts and self-test results. smartctl (from smartmontools) is the CLI; for NVMe drives, the nvme-cli package adds NVMe-specific commands.

See also: SMART Disk Health & Test · Storage commands → SMART section

Kernel parameter management. Reads /etc/sysctl.conf + /etc/sysctl.d/*.conf at boot. ProxMenux Post-Install ships a curated sysctl tuning profile.

See also: Post-Install: Network

Modern init system used by Debian / Proxmox. systemctl manages services (start / stop / enable / status). journalctl reads the binary log written by systemd-journald.

Codename for Debian 13, the OS base for Proxmox VE 9. Predecessor: bookworm (Debian 12 / PVE 8).

See also: Upgrade PVE 8 to PVE 9

Linux device manager. Reacts to hardware events and creates / removes entries under /dev. Custom rules in /etc/udev/rules.d/ persistently map MAC addresses to interface names, set up NVIDIA device nodes on driver load, etc.

In unprivileged LXC containers, UIDs and GIDs are shifted by +100000 on the host. Container UID 0 = host UID 100000, container UID 1000 = host UID 101000, etc. Affects file ownership on bind mounts.

See also: LXC Mount Points

LXC container where root inside is mapped to an unprivileged UID on the host (UID shift +100000). Safer than privileged but can't do hardware passthrough or run kernel-level services. Default for most use cases.

Linux kernel framework for safe userspace device drivers — used to bind PCIe devices to VMs for direct passthrough. Enabled by loading the vfio, vfio_iommu_type1 and vfio_pci kernel modules, configured via /etc/modprobe.d/vfio.conf.

See also: Post-Install: Virtualization

Family of paravirtualized device drivers (network, storage, balloon, GPU) that run faster than emulated legacy hardware (e.g. e1000 NICs, IDE disks). The guest needs the matching VirtIO driver — kernel-native on Linux, separate ISO install on Windows.

802.1Q tags that segment one physical network into multiple logical networks. Configured per-bridge or per-NIC in /etc/network/interfaces.

A full machine emulation managed by Proxmox via QEMU/KVM. Has a numeric VMID that maps to /etc/pve/qemu-server/<VMID>.conf. Heavier than LXC containers but fully isolated.

See also: VM and CT Management

Proxmox's built-in backup tool for VMs and CTs. Not incremental — every backup is a full snapshot. Configure modes (snapshot / suspend / stop), compression (zstd / pigz / lzo), retention and notification per job.

See also: Backup and Restore commands

Combined filesystem and volume manager known for snapshots, send/receive replication, end-to-end checksums, compression and self-healing. Memory-hungry (see ARC) but extremely robust.

See also: ZFS Management commands