Hi all. I can't understand what's happening here... maybe someone can
offer an opinion.

I've got virtualbox installed, and it happily runs *nix of various
flavours. No issues at all, and memory usage seems reasonable.

However, I created a new vm and installed windows10 on it. I allowed
10Gb max memory for the vm. When I run this, memory usage for the host
system jumps to 11Gb, and the vm is pretty soon unresponsive, and is
virtualbox itself - I have to resort to kill -9 to get rid of the
processes. That's bad - but the host's memory isn't being released, and
system monitor shows 11Gb in constant use even with no vbox active.

Starting a new copy of virtualbox promptly chews up all the remaining
memory (16Gb in the host) a bit of swap, and the host then just hangs:
power switch time.

OK, windows is a hog, but I tried this a couple of years ago (on a lower
spec machine) and it did stagger along; no hanging.

What I really don't get is the "lost" 11Gb of memory, with no processes,
seemingly, using it.

Any thoughts please?

v'box 6.1, btw; mint 21.2.

--
Mike Scott
Harlow, England

On 1/18/24 07:16 AM, Mike Scott wrote:
> Hi all. I can't understand what's happening here... maybe someone can
> offer an opinion.
>
> I've got virtualbox installed, and it happily runs *nix of various
> flavours. No issues at all, and memory usage seems reasonable.
>
> However, I created a new vm and installed windows10 on it. I allowed
> 10Gb max memory for the vm. When I run this, memory usage for the host
> system jumps to 11Gb, and the vm is pretty soon unresponsive, and is
> virtualbox itself - I have to resort to kill -9 to get rid of the
> processes. That's bad - but the host's memory isn't being released, and
> system monitor shows 11Gb in constant use even with no vbox active.
>
> Starting a new copy of virtualbox promptly chews up all the remaining
> memory (16Gb in the host) a bit of swap, and the host then just hangs:
> power switch time.
>
> OK, windows is a hog, but I tried this a couple of years ago (on a lower
> spec machine) and it did stagger along; no hanging.
>
> What I really don't get is the "lost" 11Gb of memory, with no processes,
> seemingly, using it.
>
> Any thoughts please?
>
> v'box 6.1, btw; mint 21.2.
>
>
Same setup and I run win10 and 11 with, well little problem. 11 is more
responsive than 10. 10 feels sluggish.

I have 16G and quad core cpu. I give 8G to the guest, and 4 cores.
I don't want to spend the time, but I've heard shifting the cores and
memory one way or another helps, but I don't want to spend the time,
it's of low importance to me.

I also have the vdi set to dynamic and it resides on an SSD. Luckily I
don't use them much so the wear isn't an issue.
--
Linux Mint 21.2 Cinnamon
Al

On 1/18/2024 7:16 AM, Mike Scott wrote:
> Hi all. I can't understand what's happening here... maybe someone can offer an opinion.
>
> I've got virtualbox installed, and it happily runs *nix of various flavours. No issues at all, and memory usage seems reasonable.
>
> However, I created a new vm and installed windows10 on it. I allowed 10Gb max memory for the vm. When I run this, memory usage for the host system jumps to 11Gb, and the vm is pretty soon unresponsive, and is virtualbox itself - I have to resort to kill -9 to get rid of the processes. That's bad - but the host's memory isn't being released, and system monitor shows 11Gb in constant use even with no vbox active.
>
> Starting a new copy of virtualbox promptly chews up all the remaining memory (16Gb in the host) a bit of swap, and the host then just hangs: power switch time.
>
> OK, windows is a hog, but I tried this a couple of years ago (on a lower spec machine) and it did stagger along; no hanging.
>
> What I really don't get is the "lost" 11Gb of memory, with no processes, seemingly, using it.
>
> Any thoughts please?
>
> v'box 6.1, btw; mint 21.2.

In the past, life was pretty simple.

"When a process exits, its resources are harvested..."

There were few exceptions to the rule.

Maybe something like /dev/shm (shared memory) usage between two
processes would be persistent. Shared memory could be used for
communication, and it is possible it would survive one of the
processes dying.

However, in 2024, we have virtualization, and only the power meter
connected to your mains plug, has any idea what is going on inside
the computer.

For example, with Windows booted native, today the Task Manager does
not display all activity. It does not show the Memory Compressor,
but Process Explorer shows it. If a certain field in the metadata of
an executable is blank, it does not show in Task Manager.

More worrying though, is Windows by itself, is virtualized. It isn't
anointed any more. With the inverted Hypervisor, it is just a child like
all your other virtualization activities. Task Manager simply does not
show all the activity on the machine as a result. While a previous-generation
picture of the inverted hypervisot is available, the diagram has
not been updated for Win10/Win11. And I don't know of any utility at all,
to display "who is home right now".

Now, Linux does not use an inverted hypervisor. It uses a regular one,
where the boot OS is anointed, and any children are "beneath it".
This means, we "expect better behavior of our kit", when it is run that
way.

Since /dev/shm is an unlikely candidate, we can dismiss it for the moment.
But it would be a good idea to examine the .vbox file and see if there
are any "special additions" implying something exotic has been added.
Some of these features could be added with the invocation of
"vboxmanage" from the shell.

You can see an example at the bottom of this page.

https://www.virtualbox.org/manual/ch10.html

Turning on Nested Paging might be a good idea, if the processor has it.
In the best vase, I saw a Linux Guest one day, not use all of its
memory allocation, because it was asking for memory only when it
was absolute needed.

Some symptoms a person might see, may be a side effect of the CPU vintage
and the usage of fall-back code to emulate missing bits. For example,
at one time things like VT-x were a "must have", or at least some
software people told us that. And we ran out and bought a
different CPU. Then, a year later, we discover that "heh! you don't
absolutely need that, with a 10% drop in performance, VT-x calls
can be handled in software". And then you're madder than hell, because
you got tricked into buying another processor.

I have *never* seen your symptoms, and all that means is, I haven't
added some feature that you are using. It's not unusual for people
to have different results, just from some innocuous setting they
don't remember fiddling with :-)

I remember my very first experience with VirtualBox, changing the
chipset from 440BX to ICH something-or-other, and I could not get
anything to work. I was seeing a black screen, and I could not
figure out why. What I learned from that first experience was
"don't touch the controls, dumbass" :-) After that, I tended to
treat it like bomb-squad material.

But maybe a scan through the .vbox file, will tweak some recollection.

Windows as a guest (any version), does not need more than 4GB.
W11 uses 2.6GB. W10 uses 0.35GB. W7 uses 0.35GB. The extra space
on W11 is for the sandbox image stored in memory. If for solid technical
reasons, the sandbox can never run on W11, the memory consumption drops
when the sandbox image is not included. Thus, 4GB should be sufficient
for an adventure. Any adventure. You should be able to finish a
Windows Update with that much present. (And that advice applies
to an x64 OS. W10 has x86 OS as well, and above about 3GB, it might
not be using the excess RAM. There is a way to use memory outside
the "memory license", but Microsoft sent lawyers to put a stop
to that. The feature in someones code, disappeared on the very
next release :-) The memory license for x86 is for Ring 3, and is not
enforced on Ring 0.)

I used to use 8GB on WinXP x86... just not all for programs. I did
try setting up the pagefile.sys onto the excess RAM, and that
worked for a couple days (buttery smooth paging! in WinXP!) but it
had side effects and I had to put my toy away.

Paul

> Hi all. I can't understand what's happening here... maybe someone can
> offer an opinion.
>
> I've got virtualbox installed, and it happily runs *nix of various
> flavours. No issues at all, and memory usage seems reasonable.
>
> However, I created a new vm and installed windows10 on it. I allowed
> 10Gb max memory for the vm. When I run this, memory usage for the host
> system jumps to 11Gb, and the vm is pretty soon unresponsive, and is
> virtualbox itself - I have to resort to kill -9 to get rid of the
> processes. That's bad - but the host's memory isn't being released, and
> system monitor shows 11Gb in constant use even with no vbox active.
>
> Starting a new copy of virtualbox promptly chews up all the remaining
> memory (16Gb in the host) a bit of swap, and the host then just hangs:
> power switch time.
>
> OK, windows is a hog, but I tried this a couple of years ago (on a lower
> spec machine) and it did stagger along; no hanging.
>
> What I really don't get is the "lost" 11Gb of memory, with no processes,
> seemingly, using it.
>
> Any thoughts please?
>
> v'box 6.1, btw; mint 21.2.
>
>
I have a similar setup, and just like Al, I only gave 8 GB to the VM,
and a few cores, and it's quite stable. Total memory used at this moment
is 13.2 GB, with Firefox, Thunderbird, and that VM, and Gramps.

I always use the latest VirtualBox from VirtualBox.org, so that is 7.0,
on Mint 21.3, upgraded from 21.2.

On 1/18/24 13:16, Mike Scott wrote:
> Hi all. I can't understand what's happening here... maybe someone can
> offer an opinion.
>
> I've got virtualbox installed, and it happily runs *nix of various
> flavours. No issues at all, and memory usage seems reasonable.
>
> However, I created a new vm and installed windows10 on it. I allowed
> 10Gb max memory for the vm. When I run this, memory usage for the host
> system jumps to 11Gb, and the vm is pretty soon unresponsive, and is
> virtualbox itself - I have to resort to kill -9 to get rid of the
> processes. That's bad - but the host's memory isn't being released, and
> system monitor shows 11Gb in constant use even with no vbox active.
>
> Starting a new copy of virtualbox promptly chews up all the remaining
> memory (16Gb in the host) a bit of swap, and the host then just hangs:
> power switch time.
>
> OK, windows is a hog, but I tried this a couple of years ago (on a lower
> spec machine) and it did stagger along; no hanging.
>
> What I really don't get is the "lost" 11Gb of memory, with no processes,
> seemingly, using it.
>
> Any thoughts please?
>
> v'box 6.1, btw; mint 21.2.

As I suggested to Big Al in the Wine tread, ditch it and use proxmox.

--
-------------

The moment any organization established from pure noble intentions get
some influence, it will be corrupted from both inside and outside.
Then we have organizations established from pure evil.

Edmund

On 1/18/2024 7:16 AM, Mike Scott wrote:
> Hi all. I can't understand what's happening here... maybe someone can offer an opinion.
>
> I've got virtualbox installed, and it happily runs *nix of various flavours. No issues at all, and memory usage seems reasonable.
>
> However, I created a new vm and installed windows10 on it. I allowed 10Gb max memory for the vm. When I run this, memory usage for the host system jumps to 11Gb, and the vm is pretty soon unresponsive, and is virtualbox itself - I have to resort to kill -9 to get rid of the processes. That's bad - but the host's memory isn't being released, and system monitor shows 11Gb in constant use even with no vbox active.
>
> Starting a new copy of virtualbox promptly chews up all the remaining memory (16Gb in the host) a bit of swap, and the host then just hangs: power switch time.
>
> OK, windows is a hog, but I tried this a couple of years ago (on a lower spec machine) and it did stagger along; no hanging.
>
> What I really don't get is the "lost" 11Gb of memory, with no processes, seemingly, using it.
>
> Any thoughts please?
>
> v'box 6.1, btw; mint 21.2.
>
>

LM21.2 Vbox 6.1.38 5280 MB, 6 Cores, Enable PAE/NX, Enable Nested Paging

~/VirtualBox VMs/Win10-64/Logs/Vbox.log 247.1 KB <=== See if you paused it

~/VirtualBox VMs/Win10-64/WIN10-64.vhd 23.2GB

I would check the Vbox.log and see what happened
as far as VirtualBox is concerned.

My example isn't quite the same as yours, but it's behaving.

The "virtual" it shows in "top" is not all resident.
The "RES" field is exactly equal to the allocation
defined in the VirtualBox settings for the VM.

Paul

On 18/01/24 13:16, Mike Scott wrote:
> Hi all. I can't understand what's happening here... maybe
> someone can offer an opinion.
>
> I've got virtualbox installed, and it happily runs *nix of
> various flavours. No issues at all, and memory usage seems
> reasonable.
>
> However, I created a new vm and installed windows10 on it. I
> allowed 10Gb max memory for the vm. When I run this, memory
> usage for the host system jumps to 11Gb, and the vm is
> pretty soon unresponsive, and is virtualbox itself - I have
> to resort to kill -9 to get rid of the processes. That's bad
> - but the host's memory isn't being released, and system
> monitor shows 11Gb in constant use even with no vbox active.
>
> Starting a new copy of virtualbox promptly chews up all the
> remaining memory (16Gb in the host) a bit of swap, and the
> host then just hangs: power switch time.
>
> OK, windows is a hog, but I tried this a couple of years ago
> (on a lower spec machine) and it did stagger along; no hanging.
>
> What I really don't get is the "lost" 11Gb of memory, with
> no processes, seemingly, using it.
>
> Any thoughts please?
>
> v'box 6.1, btw; mint 21.2.
>
>

Under Debian Bookworm and VirtualBox I came across much the
same problem.

I had to launch xfce4-taskmanager in ROOT mode, and not only
did I have to shut down Vbox related tasks, but also some
libvirt modules loaded (once assessed they were "child" of
the Vbox process : showing the entire cmdline may be
clafifying) that somehow survived the shut down of the
parent process.
They might behave like daemons self-reloading or else, I
dunno, but some other modules remained alive once the GUI
and the VM had been killed by hand. Only when ALL of them
was hunted for and killed, was the ram released.

I am no longer using VBox ... for unknown reasons fails to
load a bunch of saved VM and I was unable to diagnose why.

--
1) Resistere, resistere, resistere.
2) Se tutti pagano le tasse, le tasse le pagano tutti
MarioCPPP

On 18/01/24 17:32, Paul wrote:
> On 1/18/2024 7:16 AM, Mike Scott wrote:
>> Hi all. I can't understand what's happening here... maybe someone can offer an opinion.
>>
>> I've got virtualbox installed, and it happily runs *nix of various flavours. No issues at all, and memory usage seems reasonable.
>>
>> However, I created a new vm and installed windows10 on it. I allowed 10Gb max memory for the vm. When I run this, memory usage for the host system jumps to 11Gb, and the vm is pretty soon unresponsive, and is virtualbox itself - I have to resort to kill -9 to get rid of the processes. That's bad - but the host's memory isn't being released, and system monitor shows 11Gb in constant use even with no vbox active.
>>
>> Starting a new copy of virtualbox promptly chews up all the remaining memory (16Gb in the host) a bit of swap, and the host then just hangs: power switch time.
>>
>> OK, windows is a hog, but I tried this a couple of years ago (on a lower spec machine) and it did stagger along; no hanging.
>>
>> What I really don't get is the "lost" 11Gb of memory, with no processes, seemingly, using it.
>>
>> Any thoughts please?
>>
>> v'box 6.1, btw; mint 21.2.
>
> In the past, life was pretty simple.
>
> "When a process exits, its resources are harvested..."
>
> There were few exceptions to the rule.
>
> Maybe something like /dev/shm (shared memory) usage between two
> processes would be persistent. Shared memory could be used for
> communication, and it is possible it would survive one of the
> processes dying.
>
> However, in 2024, we have virtualization, and only the power meter
> connected to your mains plug, has any idea what is going on inside
> the computer.
>
> For example, with Windows booted native, today the Task Manager does
> not display all activity. It does not show the Memory Compressor,
> but Process Explorer shows it. If a certain field in the metadata of
> an executable is blank, it does not show in Task Manager.
>
> More worrying though, is Windows by itself, is virtualized. It isn't
> anointed any more. With the inverted Hypervisor, it is just a child like
> all your other virtualization activities. Task Manager simply does not
> show all the activity on the machine as a result.

some libvirt related processes are hidden at a user-level
and show up only if the taskmanager is invoked as root.
Also, even if listed, as a plain user it would be hard to
kill certain modules. I am talking about the GUI, not
terminal kill, which is more explicit in this respect.

> While a previous-generation
> picture of the inverted hypervisot is available, the diagram has
> not been updated for Win10/Win11. And I don't know of any utility at all,
> to display "who is home right now".
>
> Now, Linux does not use an inverted hypervisor. It uses a regular one,
> where the boot OS is anointed, and any children are "beneath it".
> This means, we "expect better behavior of our kit", when it is run that
> way.
>
> Since /dev/shm is an unlikely candidate, we can dismiss it for the moment.
> But it would be a good idea to examine the .vbox file and see if there
> are any "special additions" implying something exotic has been added.
> Some of these features could be added with the invocation of
> "vboxmanage" from the shell.

it also launches sub processes libvirt related that somewhat
"detaches" from the GUI.

>
> You can see an example at the bottom of this page.
>
> https://www.virtualbox.org/manual/ch10.html
>
> Turning on Nested Paging might be a good idea, if the processor has it.
> In the best vase, I saw a Linux Guest one day, not use all of its
> memory allocation, because it was asking for memory only when it
> was absolute needed.
>
> Some symptoms a person might see, may be a side effect of the CPU vintage
> and the usage of fall-back code to emulate missing bits. For example,
> at one time things like VT-x were a "must have", or at least some
> software people told us that. And we ran out and bought a
> different CPU. Then, a year later, we discover that "heh! you don't
> absolutely need that, with a 10% drop in performance, VT-x calls
> can be handled in software". And then you're madder than hell, because
> you got tricked into buying another processor.
>
> I have *never* seen your symptoms, and all that means is, I haven't
> added some feature that you are using. It's not unusual for people
> to have different results, just from some innocuous setting they
> don't remember fiddling with :-)
>
> I remember my very first experience with VirtualBox, changing the
> chipset from 440BX to ICH something-or-other, and I could not get
> anything to work. I was seeing a black screen, and I could not
> figure out why. What I learned from that first experience was
> "don't touch the controls, dumbass" :-) After that, I tended to
> treat it like bomb-squad material.
>
> But maybe a scan through the .vbox file, will tweak some recollection.
>
> Windows as a guest (any version), does not need more than 4GB.
> W11 uses 2.6GB. W10 uses 0.35GB. W7 uses 0.35GB. The extra space
> on W11 is for the sandbox image stored in memory. If for solid technical
> reasons, the sandbox can never run on W11, the memory consumption drops
> when the sandbox image is not included. Thus, 4GB should be sufficient
> for an adventure. Any adventure. You should be able to finish a
> Windows Update with that much present. (And that advice applies
> to an x64 OS. W10 has x86 OS as well, and above about 3GB, it might
> not be using the excess RAM. There is a way to use memory outside
> the "memory license", but Microsoft sent lawyers to put a stop
> to that. The feature in someones code, disappeared on the very
> next release :-) The memory license for x86 is for Ring 3, and is not
> enforced on Ring 0.)
>
> I used to use 8GB on WinXP x86... just not all for programs. I did
> try setting up the pagefile.sys onto the excess RAM, and that
> worked for a couple days (buttery smooth paging! in WinXP!) but it
> had side effects and I had to put my toy away.
>
> Paul

--
1) Resistere, resistere, resistere.
2) Se tutti pagano le tasse, le tasse le pagano tutti
MarioCPPP

On 1/20/2024 6:55 PM, MarioCCCP wrote:
> On 18/01/24 13:16, Mike Scott wrote:
>> Hi all. I can't understand what's happening here... maybe someone can offer an opinion.
>>
>> I've got virtualbox installed, and it happily runs *nix of various flavours. No issues at all, and memory usage seems reasonable.
>>
>> However, I created a new vm and installed windows10 on it. I allowed 10Gb max memory for the vm. When I run this, memory usage for the host system jumps to 11Gb, and the vm is pretty soon unresponsive, and is virtualbox itself - I have to resort to kill -9 to get rid of the processes. That's bad - but the host's memory isn't being released, and system monitor shows 11Gb in constant use even with no vbox active.
>>
>> Starting a new copy of virtualbox promptly chews up all the remaining memory (16Gb in the host) a bit of swap, and the host then just hangs: power switch time.
>>
>> OK, windows is a hog, but I tried this a couple of years ago (on a lower spec machine) and it did stagger along; no hanging.
>>
>> What I really don't get is the "lost" 11Gb of memory, with no processes, seemingly, using it.
>>
>> Any thoughts please?
>>
>> v'box 6.1, btw; mint 21.2.
>>
>>
>
> Under Debian Bookworm and VirtualBox I came across much the same problem.
>
> I had to launch xfce4-taskmanager in ROOT mode, and not only did I have to shut down Vbox related tasks, but also some libvirt modules loaded (once assessed they were "child" of the Vbox process : showing the entire cmdline may be clafifying) that somehow survived the shut down of the parent process.
> They might behave like daemons self-reloading or else, I dunno, but some other modules remained alive once the GUI and the VM had been killed by hand. Only when ALL of them was hunted for and killed, was the ram released.
>
> I am no longer using VBox ... for unknown reasons fails to load a bunch of saved VM and I was unable to diagnose why.

Now that you mention it, that is expected behavior.

The console for VirtualBox, you can run multiple copies of that application
as a starting point.

cinnamon
VirtualBox # The console is a child of Cinnamon desktop
VirtualBox # The console is a child of Cinnamon desktop
VirtualBox # The console is a child of Cinnamon desktop

You can also exit all the console windows, and leave just the VM (virtual machine) window running.

This means the virtual machine *is* a detached process. Killing the console,
does not affect the virtual machine. You need to know the name so you can
spot the thing, if the window for VirtualBoxVM were to disappear.

Then, we need to check with "ps" or some other application,
whether this "libvirt" you refer to, plays a part in the
detached process or not.

INIT
VBoxSVC # parent
VirtualBoxVM # child <=== my test copy of Windows 10 is in here right now

cinnamon
VirtualBox # The console is a child of Cinnamon desktop

libvirt is likely part of qemu kvm or something ? I have copies
of it, but I don't know which OS install has those.

Thanks for the prompt! Maybe Mikes issue will be easier to understand now,
depending on the details of what exactly happened (visual window
for VirtualBoxVM disappeared for some reason?). You should be able to kill
a VM, from the VirtualBox console window (start one anytime, for a look).

Paul

On 21/01/24 03:19, Paul wrote:
> On 1/20/2024 6:55 PM, MarioCCCP wrote:
>> On 18/01/24 13:16, Mike Scott wrote:
>>> Hi all. I can't understand what's happening here... maybe someone can offer an opinion.
>>>
>>> I've got virtualbox installed, and it happily runs *nix of various flavours. No issues at all, and memory usage seems reasonable.
>>>
>>> However, I created a new vm and installed windows10 on it. I allowed 10Gb max memory for the vm. When I run this, memory usage for the host system jumps to 11Gb, and the vm is pretty soon unresponsive, and is virtualbox itself - I have to resort to kill -9 to get rid of the processes. That's bad - but the host's memory isn't being released, and system monitor shows 11Gb in constant use even with no vbox active.
>>>
>>> Starting a new copy of virtualbox promptly chews up all the remaining memory (16Gb in the host) a bit of swap, and the host then just hangs: power switch time.
>>>
>>> OK, windows is a hog, but I tried this a couple of years ago (on a lower spec machine) and it did stagger along; no hanging.
>>>
>>> What I really don't get is the "lost" 11Gb of memory, with no processes, seemingly, using it.
>>>
>>> Any thoughts please?
>>>
>>> v'box 6.1, btw; mint 21.2.
>>>
>>>
>>
>> Under Debian Bookworm and VirtualBox I came across much the same problem.
>>
>> I had to launch xfce4-taskmanager in ROOT mode, and not only did I have to shut down Vbox related tasks, but also some libvirt modules loaded (once assessed they were "child" of the Vbox process : showing the entire cmdline may be clafifying) that somehow survived the shut down of the parent process.
>> They might behave like daemons self-reloading or else, I dunno, but some other modules remained alive once the GUI and the VM had been killed by hand. Only when ALL of them was hunted for and killed, was the ram released.
>>
>> I am no longer using VBox ... for unknown reasons fails to load a bunch of saved VM and I was unable to diagnose why.
>
> Now that you mention it, that is expected behavior.

Agreed.
But now that you speak about, maybe also some VBoxDrv or
similar was still running in the background. And closing the
window of a frozen VM, did not shutdown such background
"drivers" ... Sorry if I can't recall the exact name.

>
> The console for VirtualBox, you can run multiple copies of that application
> as a starting point.
>
> cinnamon
> VirtualBox # The console is a child of Cinnamon desktop
> VirtualBox # The console is a child of Cinnamon desktop
> VirtualBox # The console is a child of Cinnamon desktop
>
> You can also exit all the console windows, and leave just the VM (virtual machine) window running.
>
> This means the virtual machine *is* a detached process. Killing the console,

and when it malfunction, even the window of the VM may not
kill the underlying drivers or other services.

> does not affect the virtual machine. You need to know the name so you can
> spot the thing, if the window for VirtualBoxVM were to disappear.
>
> Then, we need to check with "ps" or some other application,
> whether this "libvirt" you refer to, plays a part in the
> detached process or not.
>
> INIT
> VBoxSVC # parent
> VirtualBoxVM # child <=== my test copy of Windows 10 is in here right now
>
> cinnamon
> VirtualBox # The console is a child of Cinnamon desktop
>
> libvirt is likely part of qemu kvm or something ? I have copies
> of it, but I don't know which OS install has those.

Maybe. I have also Qemu installed, and it could possibly
(not sure) a shared library.

>
> Thanks for the prompt! Maybe Mikes issue will be easier to understand now,
> depending on the details of what exactly happened (visual window
> for VirtualBoxVM disappeared for some reason?). You should be able to kill
> a VM, from the VirtualBox console window (start one anytime, for a look).
>
> Paul
>
>

--
1) Resistere, resistere, resistere.
2) Se tutti pagano le tasse, le tasse le pagano tutti
MarioCPPP

On 21/01/2024 15:33, MarioCCCP wrote:
> On 21/01/24 03:19, Paul wrote:
>> On 1/20/2024 6:55 PM, MarioCCCP wrote:
>>> On 18/01/24 13:16, Mike Scott wrote:
>>>> Hi all. I can't understand what's happening here... maybe someone
>>>> can offer an opinion.
>>>>
>>>> I've got virtualbox installed, and it happily runs *nix of various
>>>> flavours. No issues at all, and memory usage seems reasonable.
>>>>
>>>> However, I created a new vm and installed windows10 on it. I allowed
>>>> 10Gb max memory for the vm. When I run this, memory usage for the
>>>> host system jumps to 11Gb, and the vm is pretty soon unresponsive,
>>>> and is virtualbox itself - I have to resort to kill -9 to get rid of
>>>> the processes. That's bad - but the host's memory isn't being
>>>> released, and system monitor shows 11Gb in constant use even with no
>>>> vbox active.
>>>>
>>>> Starting a new copy of virtualbox promptly chews up all the
>>>> remaining memory (16Gb in the host) a bit of swap, and the host then
>>>> just hangs: power switch time.
......

>> Then, we need to check with "ps" or some other application,
>> whether this "libvirt" you refer to, plays a part in the
>> detached process or not.
>>
>>    INIT
>>      VBoxSVC           # parent
>>         VirtualBoxVM   # child  <=== my test copy of Windows 10 is in
>> here right now
>>
>>    cinnamon
>>       VirtualBox       # The console is a child of Cinnamon desktop
>>
>> libvirt is likely part of qemu kvm or something ? I have copies
>> of it, but I don't know which OS install has those.
>
> Maybe. I have also Qemu installed, and it could possibly (not sure) a
> shared library.
>

This is odd - its behaviour has changed dramatically since I posted my
query. But there has been a new version (now at 6.1.48, was 6.1.38)
installed by the upgrade manager in the interrim, so I'll have to assume
there was a bug.

Now, with the vbox manager running, usage is around 1.8Gb. On starting
virtual w10 this rises to about 10Gb (the vm is set to 8). Exiting and
saving nstate now drops back to 1.8. Restarting the saved vm, memory
usage rises to rather less than 5Gb. I don't now get windows vm's that
won't go away, and I've not had all of the host memory chewed up.

So, yes, presumably has to have been an outright bug in 6.1.38. But
thanks to all for offering comments.

--
Mike Scott
Harlow, England