Windows Server 2022 Datacenter Core (23H2 / Build 25398), AMD X570 and Ryzen 9 3900X with random lockups/freezes

[RamGuy]

Greetings,

I recently made some changes to my home server. It was previously running:

Operating System: Windows Server 2022 LTSC (Build 20348)
CPU: AMD Ryzen 9 3900X
RAM: 4x 16 (64GB) 3200MHz ECC unbuffered DDR4
Motherboard: ASUS Pro WS X570-ACE, BIOS/UEFI version 4101
Network Card: On-board Intel I211-AT + Dedicated Intel X550-T2
Graphics Card: Intel Arc A380
OS Storage: Samsung 980 Pro 1TB NVMe/M.2
Additional Storage: 8x Seagate Ironwolf Pro 14TB in RAID1+0, with 2x Samsung 850 Pro 1TB SATA SSD as cache using Windows Storage Spaces.

Server 2022 running Hyper-V, DHCP, DNS and AD roles. 4x Virtual Machines running in Hyper-V.


It got changed into:

Operating System: Windows Server 2022 Datacenter Core (23H2 / Build 25398)
CPU: AMD Ryzen 9 3900X
RAM: 4x 16 (64GB) 3200MHz ECC unbuffered DDR4
Motherboard: ASUS Pro WS X570-ACE, BIOS/UEFI version 4501
Network Card: On-board Intel I211-AT + Dedicated Intel X550-T2
Graphics Card: Intel Arc A380
OS Storage: Samsung 980 Pro 2TB NVMe/M.2
Additional Storage: 4x Samsung PM9A3 7.6TB NVMe/U.2 in RAID0 using Windows Storage Spaces

Windows Server 2022 runs the same Hyper-V, DHCP, DNS and AD roles. 4x Virtual Machines running in Hyper-V. But this time, it's Server Core (no GUI), which also gives access to a slightly newer version of Windows Server 2022 via 23H2 / 25398.


After the change, my server keeps having lockups/freezes. It's almost certain to happen at least once within 48 hours. There is nothing in Event Viewer pointing at anything other than Event Viewer not logging anything after the lockup/freeze has occurred. I have to power cycle the server for it to start working again. Then it works perfectly fine regardless of the load I put on it until it suddenly freezes again, requiring another forced reboot.


I have installed all the latest AMD chipset drivers, Intel drivers etc.. Consider removing the drivers, as all the hardware has plug 'n' play drivers out of the box. But I did run with dedicated drivers previously without any issues.

Is anyone else having a bad experience running AMD X570, Ryzen, and Windows Server? I had BSOD running Windows Server 2022 when IOMMU was enabled in the BIOS/firmware. But this got fixed. And this time I don't have any BSOD, it just freezes completely. Running without IOMMU is not an option, considering the server will run as a hypervisor.

 

[RamGuy]

It's somewhat challenging to control drivers and hardware status on Windows Server Core without a functioning device manager snap-in. But I installed Windows Admin Center on a VM and according to "Devices" there were several PCI devices in critical state without drivers. Seems like my attempt at utilising pnputil to add the AMD drivers wasn't working as the *.inf files are not exposed when unpacking the AMD drivers.

Luckily enough, upon running the .exe file it still unpacks everything to C:\AMD\, even though the installer will refuse to run on Windows Server. I could point pnputil at C:\AMD\ and get all but one device installed.

It still complains about:

Device description
PCI Device

Device instance ID
PCI\VEN_8086&DEV_4F92&SUBSYS_60041849&REV_00\6&34667467&0&00200019

Status
DN_HAS_PROBLEM, DN_DISABLEABLE, DN_NT_ENUMERATOR, DN_NT_DRIVER

Parent
PCI\VEN_8086&DEV_4FA4&SUBSYS_00008086&REV_00\5&1d36fb12&0&200019


But I can't wrap my head around what this device is. Hopefully, the freezes will stop as a result of me fixing 5-6 PCI devices without drivers that were in a critical state.

 

[RamGuy]

According to hwinfo64, 6&34667467&0&00200019 is Intel DG2 - High Definition Audio Controller. Not sure why it's having issues, as the drivers have been added. But doesn't matter. I don't use this Audio Controller so I've just left it as disabled.

 

[JOSEPHLB]

Server 23H2 is a completely different product than Server 2022..

Under the hood it is the Azure Stack HCI... it is the latest "Annual Channel For Containers'.. defiantly a good chance it doesn't behave the same as the Server 2022 LTSC release.

 

[i386]

OS Storage: Samsung 980 Pro 2TB NVMe/M.2

have you seen this thread :https://forums.servethehome.com/ind...and-990-with-windows-server.39366/post-369391 ?

Edit: fix link (removed question mark at the end)

 

[TLN]

Drive is subject to failure (unless you updated firmware)
I got issues with my Risen 5900, that were fixed by disabling cores. Looks like I fried few cores doing OC. ^_^

 

[vanfront]

Side note: I wouldn't install AD on a hypervisor. I'd use a separate VM to run AD, DHCP, and DNS (and have two such AD VMs). In addition, I wouldn't use Server Core unless forced to in a home lab environment. Not having the GUI tools is just too painful.