Can't install Windows 2016 and 2019 on my VM Version 6.1.34 r150636 (Qt5.6.2)

[HollyNav]

I'm new to VM, need support and advise from all folks here.

My host spec
- Lenovo X13 notebook (AMD Ryzen 5)
- RAM 32 GB
- SSD 512 GB

I established my new VM for testing purpose, downloaded ISO file from Microsoft site for both Windows Server 2016 and 2019 version. Here is my guest OS allocation details :
- Allocate processor : 4 core (1st tries), 8 core (2nd tries)
- Allocate memory : 4 GB (1st tries), 8 GB (2nd tries)
- Storage : SSD 50 GB
- Extended feature - Enable I/O APIC
- Network : NAT Network

I noticed that with more resources allocate to my VM, it had slightly better response eventually failed to move forward.

After first start VM preparation, i noticed high CPU usage on session info - 99% CPU load at all time. I left it half day but no progress for my installation. I can't install both Windows Server 2016 and 2019. My screen was frozen at Windows logo and message at the bottom 'Setup is Starting'. After 1 day, i found that VM thrown error on my screen "Windows Setup could not configure Windows to run on this computer's hardware".

Need advice from all folks here to help me out. I need to do lab for my project setting. thank you

[scottgus1]

I don't see the host OS type, though I surmise it's Windows from the title bar in the screenshot. If the host is Windows, right-click the guest in the main Virtualbox window's guest list, choose Show Log.

Search the far left tab's log for this text:

Attempting fall back to NEM

If you find it, Hyper-V is enabled. You have two options:

1. Try the 6.1.35 test build. Save the test build installer until you update again later. You'll need it if your host's MSI installation database goes south (they do) and the installer will not be available for download after the next even-number version gets released.

2. Turn off Hyper-V. See HMR3Init: Attempting fall back to NEM (Hyper-V is active).

If you don't find that text, save the far left tab's log, zip the log file, and post the zip file, using the forum's Upload Attachment tab.

[fth0]

f you find it, Hyper-V is enabled.

It's even easier to admire the green turtle in the screenshot.

[scottgus1]

Ah, yes, the green turtle! Should look at those screenshots better....

OK, so, Hollynav, your host Hyper-V is enabled. Option 1 or 2 are the choices available. Fth0 has confirmed that 6.1.35 fixes some important Hyper-V glitches.

[HollyNav]

Ah, yes, the green turtle! Should look at those screenshots better....

OK, so, Hollynav, your host Hyper-V is enabled. Option 1 or 2 are the choices available. Fth0 has confirmed that 6.1.35 fixes some important Hyper-V glitches.

Thanks Scottgus1,

I double check my environment, my Hyper-V was unchecked. Now downloading new ISO file and start new VM again, but found no response after pressed start. I checked Session Information, CPU Guest Load is only 50%. I have no issue when setting new Server 2012 R2 VM.

[HollyNav]

now I can create new VM with 2016 but half-day installation is in progress. I don't know why it consume such hours like this.
I dedicated 4 CPU core and 8 GB or RAM on my Ryzen7 notebook (Lenovo X13). At first, I thought it was about type of CPU - AMD and Intel but I can build Ubuntu, CentOS and Windows 2012R2 VMs with no issue.
Found issue only when build new VM with Windows Server 2016 and 2019.

[BillG]

Do you still have the green turtle? If you do, you still have the Windows hypervisor loaded. This is discussed in the reference Scott gave you earlier.

I am a bit surprised that you can install server 2019 and not 2016. If you go to Settings of the server 2016 vm, what does it display in the Version box?

Version.png (18.61 KiB) Viewed 947 times

[mpack]

I double check my environment, my Hyper-V was unchecked.

Which is why unchecking that checkbox is not among of the list of things you need to do to truly get rid of Hyper-v. Follow the link that Scott gave you.

[BillG]

Indeed. I have a test machine which always has Hyper-V installed. It causes no problems at all if the hypervisor is not loaded. It is currently running test build 6.1.35 and I can easily test the difference in performance running with or without the Windows hypervisor.

Since last posting here I have installed Server 2016 on it and it runs as expected and boots in 10-12 seconds with or without the hypervisor.

[fth0]

I have a test machine which always has Hyper-V installed. It causes no problems at all if the hypervisor is not loaded. It is currently running test build 6.1.35 and I can easily test the difference in performance running with or without the Windows hypervisor.

In my own setup with VirtualBox 6.1.35r132011, the performance ratio of the VM starting phase (~1st minute) is roughly 2:3 (VirtualBox using NEM vs. VT-x). Just out of curiosity, what is your experience?

[BillG]

In my own setup with VirtualBox 6.1.35r132011, the performance ratio of the VM starting phase (~1st minute) is roughly 2:3 (VirtualBox using NEM vs. VT-x). Just out of curiosity, what is your experience?

I don't see much difference at that point. I will attach the logs from a Windows 10 guest with and without the green turtle.

I do notice some jobs seem much slower. I haven't pinned down the cause, but I suspect they are processor hungry ones.

VMLogs.zip

(62.71 KiB) Downloaded 4 times

[fth0]

00:00:06.498761 VMMDev: Guest Log: BIOS: Booting from Hard Disk...
00:00:09.863635 Display::i_handleDisplayResize: uScreenId=0 pvVRAM=000000000c3b0000 w=1487 h=916 bpp=32 cbLine=0x173C flags=0x1 origin=0,0
00:00:32.222147 VMMDev: Guest Additions capability report: (0x1 -> 0x5) seamless: yes, hostWindowMapping: no, graphics: yes

00:00:06.542822 VMMDev: Guest Log: BIOS: Booting from Hard Disk...
00:00:15.276960 Display::i_handleDisplayResize: uScreenId=0 pvVRAM=000000000bac0000 w=1487 h=916 bpp=32 cbLine=0x173C flags=0x1 origin=0,0
00:00:52.670203 VMMDev: Guest Additions capability report: (0x1 -> 0x5) seamless: yes, hostWindowMapping: no, graphics: yes

Thanks for the log files. You can see that the performance ratio is similar to my assertion: 32:52 or 26:46 are between 1:2 and 2:3.

In the end, the exact measurements do not matter IMHO. They just show that NEM is significantly slower, but is usable to some degree.