The power fails.
Your VMs shut down in order — or live-migrate.

SNMP v1, v2c and v3 (with SHA/MD5 authentication and AES/DES privacy) as well as NUT (Network UPS Tools) over TCP port 3493. Over SNMP the software reads APC PowerNet OIDs, Eaton-specific OIDs (XUPS and MGE), Socomec-specific OIDs (Netvision 5 and 6), Vertiv/Liebert (LIEBERT-GP), Huawei (UPS2000/UPS5000) and CyberPower (CPS-MIB) as well as the RFC 1628 standard (UPS-MIB) as a universal fallback — detection runs automatically. NUT adds devices from Riello, Legrand, Powerware and hundreds of other vendors.

As soon as the UPS switches to battery, the software checks the configured thresholds for battery charge and remaining runtime. Only after several consecutive confirming polls — the safeguard against false triggers — does the emergency shutdown start: every Hyper-V host is processed in the configured order, and on each host every running VM is either stopped with Stop-VM (the guest OS shuts down) or saved with Save-VM (suspend, state written to disk). The host itself then shuts down via shutdown.exe. All active alert channels are notified the moment the outage is detected.

A one-time (perpetual) license — no subscription, no monthly or annual fees. The license covers an unlimited number of Hyper-V hosts and grants permanent use of the version current at the time of purchase. Standard costs 139 €, Pro 199 €, Enterprise 349 €. Purchase is online and activation happens with a key inside the application.

Yes. After installation the software runs for 30 days as a fully functional trial with every feature unlocked — no registration, no credit card, no automatic renewal. After 30 days UPS monitoring and emergency shutdown are locked until a license is activated.

Mainly the licensing model: APC PowerChute Network Shutdown and Eaton Intelligent Power Manager license per server node — three Hyper-V hosts mean three licenses, renewed every year. Here a single one-time license covers any number of hosts.

On top of that come vendor independence (SNMP and NUT rather than one brand), Save-VM as an alternative to shutdown, live migration to a healthy partner host, Teams, SMS and webhook alerts, plus availability and energy reports.

Yes, you can configure any number of Hyper-V hosts. Each host has its own connection settings (IP, port, user, SSL) and a freely chosen shutdown order — hosts with a lower order number shut down first. The process waits for each host to finish before starting the next, rather than shutting all of them down in parallel.

Run Enable-PSRemoting -Force once as administrator on the Hyper-V host. If the software runs in a workgroup or on a different machine, add the host as a TrustedHost:

Set-Item WSMan:\localhost\Client\TrustedHosts -Value '192.168.1.10' -Force

The account used needs membership in the Hyper-V Administrators group. WinRM can optionally run over HTTPS on port 5986. The full walkthrough is in the setup guide.

E-mail via SMTP: enter server, port, TLS/SSL, and username and password in the app. E-mail via Microsoft 365 (app-only): Azure app registration with the Mail.Send application permission, admin consent, and a client secret — tenant ID, client ID, and secret are stored in the app, ideal for running as a service with no signed-in user.

Microsoft Teams (Pro) uses a Teams Workflows webhook: in the channel, open ••• → "Workflows", create the template "Post to a channel when a webhook request is received" and enter the generated URL in the app — no Azure app registration and no interactive user needed. SMS (Pro) via seven.io: create an API key, enter recipient numbers in E.164 format. The webhook channel (Pro) accepts any HTTPS target URL — compatible with Slack, Discord, the classic Teams connector, n8n, Zapier, or your own endpoints, with an optional extra auth header.

Every channel has its own test button right in the app; detailed step-by-step instructions are in the app's built-in FAQ under "Notifications".

New installations send UPS events as standard UPS-MIB (RFC 1628) traps: power loss and the periodic on-battery message as upsTrapOnBattery (with estimated minutes remaining, seconds on battery and the low-battery threshold), all other events as upsTrapAlarmEntryAdded or upsTrapAlarmEntryRemoved carrying the matching upsWellKnownAlarm: LowBattery, ShutdownPending (emergency shutdown triggered or aborted), CommunicationsLost (UPS unreachable or reachable again), BatteryBad, DiagnosticTestFailed and OnBypass. Mains restoration ends the OnBattery alarm with a Removed trap; Added and Removed carry the same upsAlarmId.

For your monitoring system (e.g. PRTG, Zabbix, LibreNMS, Checkmk) to show the traps by name rather than as raw OIDs, the UPS-MIB (RFC 1628, module name UPS-MIB) must be loaded there; most systems ship it already.

Because the monitor server is the sender, the name of the affected UPS is included in upsIdentName; subject and message text are additionally included in sysDescr and sysLocation as before. Messages not related to a UPS (live migration, host availability, updates, test trap) are still sent with the configured trap OID.

Existing installations keep the previous format after the update (all messages with the configured trap OID) so existing receiver rules keep working. You can switch it in the Alert Manager under SNMP Trap → Trap format.

With two hosts on different UPS units the software immediately checks whether the partner node is still on mains power. If it is healthy and its UPS has enough remaining runtime, the running VMs of the affected host are moved there by live migration (Move-VM); the emptied host then shuts itself down.

If the partner has two UPS units, one of them still having mains power is enough. A partner whose UPS is itself on battery is only used if a minimum target runtime is configured, the UPS reaches it and reports more runtime than the failing one. If the partner is affected too, the software does not migrate but gracefully shuts down the hosts and servers of all affected UPS units once the thresholds are reached; UPS units still on mains power and their hosts are left alone. If only one UPS fails, only its hosts are shut down. The machine running the monitoring can also be assigned to a UPS on the Service tab so that it only shuts down when that UPS fails. Live migration is a Pro feature and requires a failover cluster with shared storage or a correctly configured shared-nothing migration.

→ In-depth guide: Hyper-V cluster power outages

UPS redundancy (Pro) concerns a single host or additional server with two independent power feeds — for example a server with two power supplies wired to two separate UPS units. A second, redundant UPS is assigned alongside the primary one. If the primary UPS reports a critical outage, the software first checks whether the second UPS is still healthy. If it is, that host's shutdown is deferred, not cancelled, and rechecked on every subsequent poll — the host is still running on the intact feed after all. Only once both UPS units are critical does it shut down gracefully.

Live migration, by contrast, concerns two different Hyper-V hosts on different UPS units: if a host's power supply fails completely, its VMs move to the healthy partner host. Both features can be combined.

In the host's or server's edit dialog, assign the protecting second UPS in the "Redundant UPS (Pro)" field in addition to the assigned one. As long as at least one of the two reports Online/Charging, the shutdown is deferred instead of executed, and rechecked on every following poll.

Important trap with the default assignment: a Hyper-V host with no UPS assigned is not shut down at all, while a server with no assignment is treated as "belonging to every UPS" and shuts down on any UPS outage — only an explicit assignment restricts it to exactly one UPS.

Timeouts: each UPS runs as its own independent monitoring process with its own poll interval. The status of the other UPS is only considered current as long as it isn't older than three times its own poll interval (minimum 30 seconds) — for an infrequently polled second UPS, its last known status can therefore be several minutes old before it's discarded. Both UPS units should therefore have a similar, fairly short poll interval. You can additionally require a safety margin in the "Min. redundant UPS runtime" field: if the protecting UPS reports less runtime than that, it no longer counts as sufficient despite being Online.

A server's individual early-shutdown threshold respects a configured redundant UPS the same way. All of this is part of the Pro license — without it, a configured second UPS is ignored and affected hosts/servers shut down normally, as if no redundancy were configured.

Yes. The Windows Service can be installed, started and stopped straight from the user interface — no sc.exe required. It starts automatically at boot and monitors the UPS with nobody logged in. The GUI and the service share the same config.json in %ProgramData%\UPSHyperVShutdown; when the GUI detects a running service it locks its own monitoring so the two never shut down at the same time.

Monitoring and shutdown run entirely on the local network: UPS communication over SNMP or NUT, the Hyper-V connection over WinRM. A power outage is therefore handled safely even without internet. Only the license needs outbound access: it is checked online every 30 days, followed by a 7-day grace period. Optionally, the update check and alert channels such as Microsoft 365, Teams, SMS via seven.io or webhook use outbound connections too.

Measurements, history, reports, configuration, credentials, logs and the audit log stay exclusively on your server — no cloud, no vendor account, no telemetry. Only these go out: the license check with Polar (license key, anonymized hardware fingerprint, IP address), the update check with GitHub (only your IP address; alternatively your own update source via UNC path or HTTPS) and alerts over the channels you set up yourself. A support bundle is only created when you export and share it yourself.

Yes, two independent ways — both an Enterprise feature. A lightweight HTTP endpoint reports current health: /health as JSON and /health/simple as plain-text OK/WARN/CRIT for classic monitoring systems such as Zabbix, PRTG or Nagios, plus /metrics in Prometheus text format for Prometheus/Grafana and Zabbix's native Prometheus scraping, and /prtg in the native JSON schema of the PRTG "HTTP Data Advanced" sensor — with a matching HTTP status code, bound to localhost by default, and optionally secured with a bearer token.

In addition, the software writes its most important events to the Windows Application Event Log with fixed, permanently documented event IDs. Microsoft Sentinel and Splunk already collect Windows events in most environments — EventID and category are available there as their own directly filterable fields, no need to parse the free-text message.

Requires an Enterprise license — without one the service won't start the endpoint, even if it's enabled. In the desktop app, under General Settings → "📡 Monitoring" tab, enable it (default port 9099, changeable right there). For Prometheus, add a job to your scrape_configs targeting <host>:9099 that scrapes /metrics (a 30s scrape interval is enough). For PRTG, create a sensor of type "HTTP Data Advanced" with the URL http://<host>:9099/prtg — PRTG automatically creates the channels (battery charge, runtime, overall health, and the rest), no manual mapping needed.

The endpoint is bound to localhost by default; to reach it from another machine, set the bind address in the same tab to a specific IP and open the Windows Firewall for the monitoring source. Optionally secure it with a bearer token — the "Generate token" button in the same tab creates it automatically and shows it once to copy, no manual editing of config.json needed.

By default at https://<server IP>:9443/ (HTTPS, the port is configurable via WebManagementPort). On a full installation (GUI + service), enable it in the desktop application under General Settings → Remote Web Interface and set up a user account there — port, bind address, session timeout and IP whitelist are configurable in the same place.

For the "Service only" installation type (Server Core), also select the "Enable remote web interface" task in the installer; on the first service start the web interface runs in setup mode, and the first account is created with a one-time setup token (see the next question). The initial self-signed certificate triggers a browser certificate warning (the fingerprint is written to the service log) — a custom certificate can be uploaded later directly in the web interface.

Requires an Enterprise license or the active 30-day trial.

Only needed for the "Service only" installation type (e.g. Server Core) — on a full installation the account is created directly in the desktop application. So that not just anyone on the network can create the first account, the setup page requires a one-time setup token. On start the service writes it to C:\ProgramData\UPSHyperVShutdown\web-setup-token.txt, which only administrators can read.

Read it as an administrator in PowerShell — on the server:

Get-Content "C:\ProgramData\UPSHyperVShutdown\web-setup-token.txt"

or from the management PC (adjust the server name):

Invoke-Command -ComputerName SERVER01 -ScriptBlock { Get-Content "C:\ProgramData\UPSHyperVShutdown\web-setup-token.txt" }

Then open https://SERVER01:9443/ on the management PC, accept the warning for the self-signed certificate and enter the token, a user name and a password (at least 12 characters). After a successful setup the service deletes the token file; from then on you sign in with user name and password.

File missing or token rejected: restart the service (Restart-Service UpsHyperVShutdown) — this creates a new token — and read the file again. After 5 wrong attempts the IP address is locked out for 5 minutes. Details are in the service log under C:\ProgramData\UPSHyperVShutdown.

Standard covers the complete core: UPS monitoring via SNMP v1/v2c/v3 and NUT, emergency shutdown for any number of Hyper-V hosts and VMs, additional physical servers, e-mail alerts over SMTP and Microsoft 365, SNMP trap, the Windows Service and the auto-updater.

Pro adds multiple UPS units with per-host assignment, UPS redundancy, live migration instead of shutdown, Microsoft Teams, SMS and webhook alerts, battery health trend, availability reports, energy and electricity cost analysis, CSV/PDF export, the monthly manager report by e-mail, Tamper Protection (digital signature of the configuration file), and the Topology tab (graphical overview with protection status and PDF emergency-plan export).

Enterprise additionally adds the remote web interface (complete browser-based operation, especially for Windows Server Core without a local GUI) and monitoring integration (a health endpoint with JSON, Prometheus metrics and native PRTG sensor format, plus fixed event IDs in the Windows Event Log for SIEM integration) and UPS monitoring: monitoring-only mode for the whole installation or per UPS, warning thresholds for battery, load, temperature and input voltage, and the UPS self-test on demand or on a schedule. The 30-day trial unlocks every Enterprise feature.