Articles

ISOs, guest tools and drivers: what the guest needs beyond its disk

Why a guest needs its own drivers and tools, and what changes when they are missing.

Reading: 4 minServer & Virtualization

Article cover: ISOs, guest tools and drivers: what the guest needs beyond its disk

A guest operating system is installed from a disk image and then runs, yet everything it needs to be managed arrives separately: the hypervisor presents a network card, a disk and a control channel, but each is usable only once the guest owns the matching driver or service. This sheet follows that second half — the image that delivers installers and drivers, the driver a paravirtualized device requires, the agent that lets the host shut the machine down and read its address, the services switched on at both ends, and the first-boot configuration.

The model is two sides, three channels: a device on the host is inert until the guest loads its driver, an ISO until the guest mounts it, an agent channel until a daemon answers on the guest side.

   HYPERVISOR (host side)                GUEST (everything beyond its disk)
   ┌──────────────────────┐              ┌───────────────────────────────┐
   │  virtio device        │   virtio    │  virtio driver                │
   │  (network, block,     │◀───────────▶│  (built into the guest kernel │
   │   console, agent …)   │  queue+irq  │   or installed as a module)   │
   ├──────────────────────┤             ├───────────────────────────────┤
   │  CD-ROM / ISO image   │  install ▶  │  OS installer, virtio-win,    │
   │  (removable media)    │             │  qemu-guest-agent, cloud-init │
   ├──────────────────────┤             ├───────────────────────────────┤
   │  agent channel        │◀───────────▶│  qemu-ga / integration        │
   │  (freeze, shutdown,   │  unix sock  │  services (a daemon inside    │
   │   IP report)          │             │   the guest)                  │
   └──────────────────────┘             └───────────────────────────────┘

Virtio: a device the guest must be taught to drive

The OASIS VIRTIO 1.2 specification describes virtual devices that “by design they look like physical devices to the guest within the virtual machine”, so that the guest can “use standard drivers and discovery mechanisms.” The device sits on the host; the guest supplies the other half. A virtio device is “discovered and identified by a bus-specific method”, and a status flag marks the moment the guest is ready — “the guest OS knows how to drive the device.”

A paravirtualized device is therefore only as good as the driver behind it. Proxmox VE recommends the VirtIO network model and states the precondition plainly: “Like all VirtIO devices, the guest OS should have the proper driver installed.” Where the guest lacks that driver, the emulated E1000 card is the fallback, so an old or unmodified guest boots on the slower emulated device that it can actually see.

The ISO: installers, drivers and tools on removable media

A guest’s disk carries its operating system; the tools arrive on removable media. Proxmox VE treats an ISO as one storage content type among several — “images, cdrom iso images, container templates or container root” directories. A machine is created with an optical drive pointed at an image, for example qm create 300 -ide0 local-lvm:4 -net0 e1000 -cdrom local:iso/proxmox-mailgateway_2.1.iso, and that same drive later carries a driver disk or an agent installer. Linux drivers usually come from the distribution’s packages; Windows drivers do not, so Proxmox’s instructions begin with “First you have to download the virtio-win driver iso”, and that image is used twice — the driver through Device Manager, then the installer in its guest-agent directory.

The guest agent: clean shutdown, consistent freeze, reported address

An agent inside the guest lets the host cooperate with a running operating system. Proxmox VE describes a “communication channel between the host and the guest”; the upstream wiki calls it “a helper daemon, which is installed in the guest.” It exists for two ends — “To properly shutdown the guest, instead of relying on ACPI commands or windows policies”, and “To freeze the guest file system when making a backup/snapshot (on windows, use the volume shadow copy service VSS).”

The freeze is what makes an online backup or a disk-only snapshot consistent: the host tells the guest to sync outstanding writes “via the guest-fsfreeze-freeze and guest-fsfreeze-thaw commands.” Disabling it is possible, at a stated cost: backups “with inconsistent filesystems.”

The agent also reports: “the IP addresses in the VM summary panel are fetched via the guest agent.” On Hyper-V the same consistency role belongs to the guest’s VSS requestor, whose Linux counterpart is “required to create live Linux virtual machine backups.”

Integration services: a switch on both sides

Hyper-V packages the same idea under a name: “Hyper-V Integration Services enhance virtual machine performance and provide convenience features by leveraging two-way communication with the Hyper-V host.” Some services are conveniences, “others are important to the virtual machine’s functionality, such as synthetic device drivers.” The rule is symmetry — a service must be “enabled in both the host and guest in order to function”, and “its corresponding service must be running within the guest in addition to being enabled on the host.” Older hosts deliver the components on a setup disk, which “loads the setup disk in the virtual DVD drive.” The same two-sidedness governs checkpoints: a production checkpoint “uses Volume Shadow Copy Service or File System Freeze on a Linux virtual machine to create a data-consistent backup”, and without it the checkpoint degrades to a memory-state capture.

First boot: cloud-init

A new machine is usually a clone of a template, so its identity is set on first start. cloud-init is “the industry standard multi-distribution method for cross-platform cloud instance initialization”, and its work is defined for that moment: instances are “provisioned during first boot with networking, storage, SSH keys, packages and various other system aspects already configured.” The hypervisor writes the parameters and the guest reads them — Proxmox VE “generates an ISO image to pass the Cloud-Init data to the VM” — and driver, agent and cloud-init share one property: none is on the guest’s disk until something puts it there.

What stops working when they are missing

Several failures trace back to a missing piece.

  • Clean shutdown. The host falls back to an ACPI signal or a policy, and the request times out: “If the guest agent is not running, commands cannot get executed properly and the shutdown command will run into a timeout.”
  • Consistent backup. Without the freeze, a backup or disk-only snapshot can catch a filesystem mid-write, with the guide’s warning of “backups with inconsistent filesystems.”
  • A guest-visible snapshot. An application-consistent checkpoint depends on the freeze or Hyper-V’s VSS requestor; without it the result is crash-consistent.
  • The address after a move. The address the management interface shows is fetched from the agent, so a machine migrated or restarted without a running agent leaves that field empty.

None is a hardware fault, and none is cured by adding memory or cores; each is the price of a guest that never received the other half of its virtual machine.

Level and prerequisites

L4 — the guest-side lifecycle of a virtual machine: the drivers, agents, integration services and first-boot tooling inside the guest that enable features the hypervisor cannot perform alone. Prerequisites: the split between emulated, paravirtualized and passed-through devices and what a guest does or does not see (see the hosts/vCPUs sheet), and the ISO storage content type.

Where to go next

  • Server & Virtualization — the area this sheet belongs to.
  • Named only: PCIe passthrough and IOMMU, SR-IOV, and resource allocation.

References

Nodrius Field Kit — cover

Get the Nodrius Field Kit

Enter your email and we'll send you the complete Field Kit: eight technical resources in one download.

Your email address is sent to Nodrius and used to deliver the Field Kit to you by email. Marketing messages are separate and optional: the checkbox does not affect your download, and you are only added to our updates list if you tick it. Privacy Policy.

Privacy & cookies

This site does not use cookies, analytics or tracking, and it does not profile you. There is nothing technical to switch off, so Accept and Reject change nothing about how the site works: either choice lets you browse everything normally.

The only difference is what your browser remembers: two entries in local storage — your choice here, so this notice is not shown again, and a marker that the Field Kit offer has already been made — so the offer is presented once and not on every page. They contain no identifier, they are not shared with anyone, and the site sets no cookie for either.

What we do with your data
How your email address is used when you ask for a Resource, how long it is kept and which rights you have — in the Privacy Policy.
What the site stores in your browser
Two entries in local storage — your choice on this notice and whether the Field Kit offer has been shown — and no cookies, no session storage, no third-party content: in the Cookie Policy.