Over 16 years of working on Windows infrastructure (for public companies, billion dollar enterprises, and small businesses), the same Hyper-V problems kept showing up. Not exotic ones. The same dozen misconfigurations, over and over, quietly waiting to become outages: checkpoints nobody deleted, iSCSI targets that would not survive a reboot, clusters that looked highly available but were not.
So the first thing I do on any Hyper-V environment is run an assessment script I have been refining for years. As of this month it is free and open source on GitHub, and this post explains what it checks, why it is strictly read-only, and how to run it on your own servers.
What it checks (the greatest hits)
The script maps a Hyper-V host (or an entire failover cluster) end to end: every VM’s CPU, memory, and disks (provisioned versus actually used), checkpoints, integration services, replication health, storage paths, network acceleration settings, patch recency, and about forty best practice checks on top of the raw inventory.
The check I care most about is N+1 failover capacity: if one cluster node dies right now, can the smallest surviving node actually absorb every running VM? Plenty of clusters pass a glance and fail this math, and the owners find out which VMs will not restart during the outage. The script does the arithmetic and prints a PASS or FAIL with the numbers shown.
Other favorites from the field: iSCSI storage with zero persistent targets (meaning the SAN does not reconnect after a reboot: the outage you discover at 6 a.m.), checkpoints on domain controllers (a USN-rollback hazard that can silently break Active Directory replication), VMQ enabled on 1-gigabit network cards (the classic “VMs are slow and nobody knows why”), and memory demand exceeding assigned memory: a virtual machine quietly thrashing its page file while users blame “the server being old.”
Every finding explains itself
Most audit tools dump a spreadsheet and leave the interpretation to you. This one writes an HTML report where every finding has three parts: what we saw, why it matters, and what to do. The goal is that a business owner, not just an engineer, can read the report and understand what is actually at risk.
You also get a full console transcript for the raw detail, and an optional JSON export if you want to feed the results into other tooling.
Why it is strictly read-only
The script changes nothing. No settings, no restarts, no “auto-fix.” It uses only read commands, and we verify that claim by parsing the code itself. There is not a single command in the file that changes state on your infrastructure. That is a deliberate design decision, not a missing feature: it means you can run it on a production host at 2 p.m. on a Tuesday without a change window, and it means every fix it recommends gets made by a human who understands the environment.
That is also our general philosophy for managed IT: tools should observe and report; people should decide and change.
How to run it on your own server
Grab it from github.com/kythel/hyperv-audit. On the Hyper-V host, open PowerShell as Administrator and run .\Invoke-HyperVAudit.ps1. It auto-detects whether the machine is standalone or clustered, reaches the other cluster nodes over the same read-only channel, and drops the HTML report and transcript on your desktop. The whole run typically takes a minute or two.
The source is one file on purpose: audit it yourself before running it. MIT licensed: use it, fork it, run it for your own clients.
If the report finds something scary
Some findings are ten minute fixes. Some (an N+1 failure, a degraded disk, replication in a critical state) deserve a careful hand. If you are in Palm Beach or Broward County and want a second opinion on your report, send it over: reading it costs you nothing, and you will get a straight answer about what actually needs fixing and what can wait. That conversation starts on our IT support page, or just call.
And if you would rather never think about any of this: catching these problems before they become outages is literally what our managed IT plans exist for.
Ran the audit and want a second opinion on the report? Send it over. Reading it is free, and you will get a straight answer about what matters.