My Windows Server 2025 Standard VM randomly becomes completely unresponsive. No response via VNC console, RDP, or any network service. The VM shows as "Running" in XO/XCP-ng Center.
RAM/CPU graphs in XO freeze at the last value. No errors or relevant logs in Windows Event Viewer (daemon/kernel), XCP-ng host logs, or hypervisor logs. Only solution is a forced reboot (or power off/on via XO). After reboot, the VM works normally for hours/days until the next freeze.
This matches known Windows Server 2025 + Xen issues.
Environment
Hypervisor: XCP-ng (version: please add yours, e.g. 8.3.x)
Orchestrator: XO (XOA)
VM Details (from XO API / XO):
{
"id": "adf48a15-0b02-6fee-21d8-895190077680",
"name_label": "<VM name>",
"power_state": "Running",
"virtualizationMode": "hvm",
"os_version": {
"name": "Microsoft Windows Server 2025 Standard",
"uname": "10.0.26100"
},
"CPUs": { "max": 4, "number": 4 },
"memory": { "size": 17179820032 },
"viridian": true,
"pvDriversVersion": "9.1.145-77",
"pvDriversDetected": true,
"managementAgentDetected": true,
"management agent": "9.1.145-77",
"NIC type": "Realtek RTL8139",
"VGA": "cirrus",
"boot": { "firmware": "bios", "order": "cd" },
"other": {
"base_template_name": "Other install media"
}
}
Hardware Virtualization: PVHVM with paravirtualization drivers enabled.
Storage: (please specify: local SR, NFS, iSCSI, etc. + VBD types)
Network: Realtek RTL8139 (emulated)
Steps Already Taken
Viridian extensions are enabled (viridian: true).
Latest XCP-ng Windows PV Drivers + Management Agent 9.1.145-77 installed.
Windows Server 2025 is fully updated.
Checked host logs (xl dmesg, journalctl), guest Event Viewer — nothing at the time of freeze.
VM uses BIOS boot (not UEFI).
Additional Context
The freeze is completely random (no specific workload, time of day, or high load trigger observed).
Reboot always resolves it temporarily.
Similar reports exist for Windows 11 24H2 / Server 2025 on XCP-ng (Viridian synthetic timer, PV drivers, emulated devices).
Questions
Are there any known issues with Realtek RTL8139 NIC on Server 2025? Should I switch to a PV network driver / different emulated NIC?
Any recommendations on VGA (cirrus) or other VM settings?
Should I try forcing a kernel memory dump next time it hangs (xl trigger <domid> nmi)?
Any other diagnostics or logs I should collect?
Thanks in advance!
