LHC@home

Hi

I had to rebuild my PC, and since then then the LHC runs for 10 sec and then says computation error. the Event Log has no useful info. I've reinstalled VM too. Can you point me to a fix.

I had to rebuild my PC, ... then says computation error. the Event Log has no useful info.

The log clearly points out what's wrong.
LHC VMs require hardware virtualization which is either switched off in your BOIS or you did not completely disable Hyper-V (+ all related components).

VBoxManage.exe: error: Not in a hypervisor partition (HVP=0) (VERR_NEM_NOT_AVAILABLE).
VBoxManage.exe: error: VT-x is disabled in the BIOS for all CPU modes (VERR_VMX_MSR_ALL_VMX_DISABLED)

VBox tasks of all types ending with "Computation Error".
Any ideas or useful links?

---

VBox tasks of all types ending with "Computation Error".
Any ideas or useful links?

In your result:

VBoxManage.exe: error: Not in a hypervisor partition (HVP=0) (VERR_NEM_NOT_AVAILABLE).
VBoxManage.exe: error: AMD-V is disabled in the BIOS (or by the host OS) (VERR_SVM_DISABLED)

With WCG shutting down for a month or more I have moved back to LHC but I am getting similar problems on my 2 laptops which previously worked faultlessly.
The VM is created but instantly fails on start and I get the same VT-x disabled message. I have checked that it is enabled in the Bios and Leomoon gives 2 ticks.
I had updated VBox so will revert to the previous version in case that is to blame.
There have also been a couple of Windows updates so one of them may be the culprit.

The good news is - WCG starting at 28th this month again with a stresstest.
Yes, have also a few days worked to let LHC-Tasks better running.
Your PC show no active VT-X, whatever the reason is.

@Ray: Is <p_vm_extensions_disabled> set to yes >1< in client_state.xml?

I didn't check that as there should have been no reason for that to change but I'll look when I get home and set it to 0 again if it has changed.
Thanks

In your result:
VBoxManage.exe: error: Not in a hypervisor partition (HVP=0) (VERR_NEM_NOT_AVAILABLE).
VBoxManage.exe: error: AMD-V is disabled in the BIOS (or by the host OS) (VERR_SVM_DISABLED)

So, my machine is not supported... Right?

---

No joy.
Virtualisation enabled in Bios and Task Manager and CPU-V both confirm that.
vm_extensions_disabled>0< (unchanged)
No Hyper-V as both hosts are Win10Home but I have made sure
Virtual Machine Platform and
Windows Hypervisor Platform are both unchecked.

I'm currently doing a System Restore on the other host to see if removing the last 2 Windows updates will resolve the problem but it is taking ages. Might have to leave it doing that overnight so I won't know if it's fixed until 2moro. If that does work, I'll get an Update Blocker as neither machine is compatible with Win11 so I'll run then as they are until they die.

The Restore didn't fix it so I don't know what's going on.
Hopefully the new XTrack, Sixtrack upgrade won't be far away and my machines can concentrate on that instead.

The Restore didn't fix it so I don't know what's going on.

@Ray: Could you try 1 CMS-task. CMS is using the newest vboxwrapper. I don't think it helps, but it's better to have tested that.

maeax wrotte:

The good news is - WCG starting at 28th this month again with a stresstest.

maeax, are you sure? They are supposed to be closed down until about Mid-April being in final phase of transition from IBM to Krembil.

maeax wrotte:

The good news is - WCG starting at 28th this month again with a stresstest.

maeax, are you sure? They are supposed to be closed down until about Mid-April being in final phase of transition from IBM to Krembil.

WCG wrote:
Ensuring a smooth and safe transition is our first priority.
As a result, we will need to pause new work units on February 14, to gracefully shut down all computation by February 28.
We will then backup and stress test the new system before proceeding into the final phase of the transition.

Not sure how that stresstest is performed and whether BOINC-users, -hosts and -data are involved or not.

Same result, sadly.
It gets as far as creating the VM and all looks well until it Starts the VM and instantly stops it again and throws the error.
So for some reason, as yet unknown, virtualization is not being seen as available whereas it IS shown as available and active in Task Manager and CPU-V.

Thanks for the PM, Samson.
I suspended the CMS during download so that I could resume it when I had a chance to observe what it would do
but, alas, it didn't go well. I also ran it on its own so there would be no other contention with memory or processor usage.

I skived off work to do that test so I better get back to that and will investigate further this evening.

I am puzzled that both hosts exhibit the same symptoms, having previously been working fine.

Hi Ray,

This is from your logfile here:
https://lhcathome.cern.ch/lhcathome/result.php?resultid=345405352

2022-02-22 12:56:37 (4696): 
Command: VBoxManage -q startvm "boinc_bcabce4bc5f50f29" --type headless
Exit Code: -2147467259
Output:
Waiting for VM "boinc_bcabce4bc5f50f29" to power on...
VBoxManage.exe: error: Not in a hypervisor partition (HVP=0) (VERR_NEM_NOT_AVAILABLE).
VBoxManage.exe: error: VT-x is disabled in the BIOS for all CPU modes (VERR_VMX_MSR_ALL_VMX_DISABLED)
VBoxManage.exe: error: Details: code E_FAIL (0x80004005), component ConsoleWrap, interface IConsole

BOINC correctly copies all it needs to the worker slot and vboxwrapper correctly configures a VM.
The failure happens when vboxwrapper starts the VM via VBoxManage.


Have you searched for the vbox error messages given there?
VERR_NEM_NOT_AVAILABLE
VERR_VMX_MSR_ALL_VMX_DISABLED

There are lots of posts everywhere in the internet like:
https://techsupportwhale.com/not-in-a-hypervisor-partition/
https://stackoverflow.com/questions/33304393/vt-x-is-disabled-in-the-bios-for-both-all-cpu-modes-verr-vmx-msr-all-vmx-disabl

Have you tried to start VirtualBox Manager and created/started a (dummy; diskless) VM using the same user account you use for BOINC?
This sometimes show an error message you don't see when you (or BOINC) start a headless VM.

Another simple reason come to my mind from the past.
Did you re-ïnstalled BOINC lately and maybe did the service install?

I think I'm getting sucked further down the rabbit hole, now.

Virtualization is definitely Enabled in Bios and confirmed in Task Manager and CPU-V but for some unknown reason is not being detected by VBox.
Boinc was last updated in November so not that, and I downgraded VBox to 6.1.30 in case that was the issue, even though earlier tasks had worked with 6.1.32.

In Turn Windows Features on/off, checking Virtual Machine Platform then rebooting to get that to stick, allows an empty VM to boot up after attaching the Guest Additions iso. However, in that configuration Boinc says No Tasks available. Turning that back off, and rebooting, the empty VM fails on start as before, BUT Boinc gets a task, which promptly fails as before. Bizarre behaviour.

Thanks for the input, guys but my brain's getting a bit frazzled and I'm starting to forget what I did or didn't do so I'm going to give up for now and revisit it on Sunday when I will have more time to investigate.

Windows key Settings
Choose Update & Security
Select Recovery
Advanced Section Restart now
Troubleshoot > Advanced options > UEFI Firmware Settings
Click Restart now
Activating Virtualisation (AMD-V - is SVM Parameter or VT-X)
Hoping it works, when you have time at Sunday.

I'm getting the same problem on 6 of my 7 machines with CMS and ATLAS. I have none of the problems suggested earlier. The only things changed on these machines since I last ran LHC ok is Boinc being upgraded, and the OS changed from Windows 10 to Windows 11. I have checked it didn't put HyperV on.

Rosetta Python VB and LHC are running ok on only 1 machine, the only difference I can think of is this is the only one with TPM, "required" for Windows 11. The other 6 I bypassed that check to force Windows 11 to install. Not sure why that would affect VB.

These are my machines:
https://lhcathome.cern.ch/lhcathome/hosts_user.php?sort=rpc_time&rev=0&show_all=0&userid=55945

The only one that works is ID 10772275 (the i5).

Mind you Rosetta had problems before I upgraded to Windows 11, so ignore that. They have different errors anyway.

And ignore the TPM thing too, I was wrong. The i5 running LHC ok does NOT have TPM, it's the Ryzen with TPM. So I have no idea what makes only that one work. Also my Ryzen 3900XT is running it ok (hadn't tried that one before). So the difference is now the two newest PCs run LHC ok. Has there been a change that means it doesn't like older processors?