A lab server you can break, and the two halves of it

This lesson maps to no SK0-005 objective. CompTIA does not examine "build a lab", and nothing in it will be asked of you as such. It exists because the labs attached to objectives in all four domains ask you to do things to a server that a server in use should never have done to it. Its material still overlaps the exam: the physical half is the hardware of 1.1 and 1.3, the software RAID it suggests is 1.2, the Redfish emulator is the out-of-band management in 1.3, snapshots belong to virtualisation in 2.5, and the warning that a snapshot is not a backup is 4.4.

Why this matters

Server+ is a scenario exam. It will describe a degraded array, a service that will not start after a patch, or a failover that did not happen, and ask what you do next. You can memorise the answers to those questions, but you will recognise them far faster if you have caused each of those faults yourself and watched what the machine did. That needs somewhere to cause them.

This lesson builds that place before any lesson asks you to pull a drive, rebuild an array or break a service. Everything later in the course assumes it exists.

The lesson

Why a server course needs a machine that matters to nobody

Every useful server exercise is destructive. Rebuilding a RAID array erases what was on the member disks. Changing a machine's IP address, DNS server or default gateway can cut it off from the network you are using to reach it. Disabling a service to see what depends on it will break whatever depended on it. Patching and then rolling back is only a lesson if the patch was allowed to do damage first.

None of that is safe on a machine somebody relies on, and that includes your own everyday computer. The rule for this whole course is simple: labs run on a machine whose total loss would cost you nothing but the time to rebuild it. That is not caution for its own sake. A lab you are afraid to break is a lab you will not break, and the faults you do not cause are the ones you will not recognise under exam pressure.

Server+ also covers two different kinds of work, and they need different kinds of lab. Hardware installation and management, the first domain, is physical: drive bays, power supplies, cabling and firmware. Administration, security and most of troubleshooting are software, and almost all of that runs in virtual machines. So the lab has two halves.

The virtual half: a hypervisor, two server VMs, an isolated network and snapshots

The virtual half does most of the work in this course. It needs four things.

A hypervisor runs the virtual machines. On a Windows Pro machine, Hyper-V is built in and can be switched on under Windows Features. VirtualBox is free on Windows, macOS and Linux. If you have a spare machine to dedicate, Proxmox VE installs as a type 1 hypervisor straight onto the hardware; the lesson on hypervisors later in the course explains the difference, and either works here.

Two server virtual machines give you something to administer and something to talk to. Use one Windows Server and one Linux server if you can, because the exam expects you to recognise both. Microsoft publishes a Windows Server evaluation that runs for 180 days, which is long enough for this course, and Ubuntu Server or Rocky Linux are free. Two gigabytes of memory and one or two virtual CPUs each is enough; none of these labs are performance tests.

An isolated network keeps your experiments away from your real one. In Hyper-V this is an internal or private virtual switch; in VirtualBox it is an internal network or a host-only adapter. Do not bridge the lab machines onto your home network. The DHCP lesson will have you run a DHCP server, and a second DHCP server answering on a real network hands out wrong addresses to every device that asks, including other people's.

Snapshots are the reason this half is so forgiving, and they get their own section below.

The physical half: a spare PC, a second disk and what an old machine can still teach

Some of the first domain cannot be virtualised honestly. A virtual machine has no power supply to fail, no fans, no drive bay, and firmware that is really the hypervisor pretending. For those lessons, an old desktop PC teaches more than any simulation.

The machine does not need to be a server or even recent. What makes an old PC useful is what you can do to it:

• Add a second and third disk. Most desktops have spare SATA ports. Two or three small, cheap or salvaged drives are enough to build a software RAID array, fail a member deliberately, and watch the rebuild, which is the core of the storage lessons.
• Watch it POST. Unplugging a memory module or a drive's power cable and reading what the firmware reports is the troubleshooting domain's hardware half in miniature.
• Explore its firmware. Boot order, UEFI against legacy mode, Secure Boot and SATA controller modes are all visible in a desktop's setup screens, and they are the same concepts a server's firmware exposes with more options.
• Read its disks' health. SMART data from a tool such as smartctl shows the same counters a server's storage controller watches for failing drives.

Treat the physical machine with the same rule as the virtual ones: nothing on it you would mind losing. Wipe it before you start.

Mains electricity is the one hazard in this course that a snapshot cannot undo. Never open a power supply, and switch off and unplug the machine before touching anything inside the case.

Emulators for what you cannot own: a Redfish mockup and software RAID

Some server hardware is simply out of reach. Few people own a rack-mount server with a baseboard management controller, or a hardware RAID card with a battery backed cache. For those, emulation gets you close enough to learn the concepts and the commands.

For out-of-band management, the DMTF, the body that maintains the Redfish standard, publishes a Redfish mockup server: a small Python program that serves a realistic Redfish API describing an imaginary server. You can query it with the same HTTPS requests you would send to a real iLO or iDRAC, read its temperatures and power supply status, and see how the data is laid out. The out-of-band management lesson later in this domain uses it. It cannot power a real server on or off, and the lesson says where the emulation stops.

For RAID, software RAID on Linux (mdadm) or Windows (Storage Spaces) demonstrates every level the exam asks about. If you have no spare disks at all, Linux can build an array from loop devices, which are ordinary files pretending to be disks. That is useless for performance and ideal for learning what happens when a member fails, because you can fail one with a single command and rebuild it in seconds.

The honest limit of emulation is hardware behaviour: a controller's cache battery, a drive's actual failure, a power supply's noise. Those you learn by reading logs and specifications, and the exam asks about them in that form.

Taking a snapshot before every lab, and restoring it after

A snapshot records the state of a virtual machine's disk and, optionally, its memory at a moment in time, so you can return to that moment later. It is what makes the virtual half of this lab safe to destroy.

Build the habit now: before every lab, take a snapshot of every machine the lab touches and name it after the lab. When the lab is done, or goes wrong, revert to it. Your two servers then start every lab clean, and a mistake costs a minute rather than an evening.

One distinction matters for the exam as well as the lab. A snapshot is not a backup. A snapshot usually lives on the same storage as the machine it records and depends on the original disk to be useful, so anything that destroys that storage destroys both. Snapshots are for short-term rollback around a change; backups are separate copies you can restore somewhere else. The virtualisation lesson and the backup lessons at the end of the course both return to this, because it is one of the most commonly tested distinctions on the paper.

The physical machine has no snapshots, so treat its setup as disposable instead: keep notes of how you installed it, and be prepared to reinstall.