All my invalids are very short lived tasks and as damotbe said in another post they correspond to bad molecules.
Tullio

---

bonjour j'utilise windows 10 21h1 cpu i7 9750h, 32gigas ram et oracle vo 6.128
j ai utilisé la 6.122 cette nuit et le probleme est identique.
merci a tous pour votre aide

It is a different problem, but now you are not having many errors for "VERR_NEM_INIT_FAILED (VERR_NEM_VM_CREATE_FAILED)"

Some tasks have worked and validated so you have all that is necessary for it to work.
How many are you trying to run at the same time ?
I would stop new tasks from running, once those running have failed or completed I would then go into VirtualBox and remove any orphaned VMs to clear the environment.
For good measure I would reboot.
Then I would allow just a single task to run and see what happens, there are many ways to do this, if you don't know ask.
If that still fails what is the error message in the Stderr ouput ?

Running VMs is hard work for the disk io especially if you are spinning up quite a few of them at the same time, even on an SSD. If the computer has a high CPU usage this can also cause problems for VMs.

ok merci
j ai réglé quchem en taches illimitées mais avec 2 cpu.
j ai un pc avec un i9 10900 et je n'ai pas ce problème et l'installation Windows est identique.
on va attendre et on verra comme ça pour quchem.
j'ai déja nettoyé les machines oracle invalides.

All my invalids are very short lived tasks and as damotbe said in another post they correspond to bad molecules.
Tullio

I don't get any short lived tasks.
I am running native linux but if it was a bad molecule wouldn't I get them too ?

If they are bad molecules why can others hosts get valid results with them ?
For example this task you ran: https://quchempedia.univ-angers.fr/athome/workunit.php?wuid=2954031
You got a Validate error in less than a minute, initially it said "ERROR: Vboxwrapper lost communication with VirtualBox, rescheduling task for a later time" and rescheduled it to start again. It re-started about an hour later but as soon as it started the VM it shut it down. It does not look like any actual work was done to a molecule, good or bad.
Two other hosts successfully ran that task to completion and got credit.
The two other failures (on native linux) are due to bad hosts, I reported 8613 previously but it is still churning out bad results, 9415 I didn't report for blacklisting as it wasn't getting new work.

I would say the bad molecule tasks are those that run to completion by multiple hosts but get a status of "Completed, validation inconclusive" and never validate, eventually having too many attempts.

Your short lived tasks look like your host/environment can't cope with what it is being asked to do. Some get to work but more of them fail to start properly.

I am number 30 in the RAC ranking list. Before the 30 September stop due to an expired security certificate I was even better and I am now climbing back again. Thie means my results are good even if my CPU, an Intel i5 9400f is vastly inferior to the majority of Linux hosts.
Tullio

---

si j'installais quchem avec boinc sous linux avec une machine virtuelle sous linux mint.
quchem fonctionne en natif sous linux.
je mets 16 gigas de ram taches illimitées cpu illimités
qu 'en pensez vous.
merci

i have a Linux Virtual Machine on a Windows 10 host running SuSE Tumbleweed, which is a development version with kernel 5.14.14. But it is updated very frequently and I have to reboot it, so I am not running any project using VirftualBox on it, only Einstein@home.
Tullio

---

I am number 30 in the RAC ranking list. Before the 30 September stop due to an expired security certificate I was even better and I am now climbing back again. Thie means my results are good even if my CPU, an Intel i5 9400f is vastly inferior to the majority of Linux hosts.
Tullio

If you were the only person in a race you would finish first.
Does not make you a good runner.

There is not a lot of work being done here.
You have the hosts that should be blacklisted but aren't, producing thousands of invalid results.
You have those running VBox and producing many short running invalid results, difficult to quantify how many of these, but a lot.
You have the bad molecules that run for a long time that have to run 8 times but get no credit.

Chances of getting good value for the time and cost of running this project are low.
I'd like to get to 50M but currently babysitting a single host running a few tasks now and then to reduce wasted time/effort.

si j'installais quchem avec boinc sous linux avec une machine virtuelle sous linux mint.
quchem fonctionne en natif sous linux.
je mets 16 gigas de ram taches illimitées cpu illimités
qu 'en pensez vous.
merci

I think the translation is telling me you want to run unlimited cores in the VM ?
You need to leave some spare CPU capacity for the OS and running the VM.

I haven't made note of how much memory tasks are using natively but if you only run say 10 cores in the VM then 16 Gb would be enough RAM.

I am runing 6 BOINC projects on 4 different CPUs,3 using Windows 10 and 11, one Linux and a Linux Virtual Machinwe. I am an old UNIX user in my professional life, and I was surprised to see that on QuChem most, if not all, Windows tasks using VirtualBox are faster than their Linux wingmen. So I stuck to Windows on ths project.
Tullio

---