VERR_NO_LOW_MEMORY On 7.1.0

[jeisom]

Not sure what is going on here. I have 50GiB of ram available out of 64GiB with VMs that are set to 16GiB and 24GiB and none launch.

System is Arch Linux, Ryzen 5900X, Plasma Wayland 6.2beta, but 6.1 failed as well.

Could someone point me in the right direction to fix this? VB 7.0 worked fine.

--
Out of memory condition when allocating memory with low physical backing. (VERR_NO_LOW_MEMORY).
Result Code:
NS_ERROR_FAILURE (0x80004005)
Component:
ConsoleWrap
Interface:
IConsole {6ac83d89-6ee7-4e33-8ae6-b257b2e81be8}
--
Log attached.

[jeisom]

Whatever the cause, it is working now. Not sure what package changed. Additionally 7.1.2 is not doing it either.

[nixuser]

This has nothing to do with VirtualBox-7.1, I have been getting this on VirtualBox-7.0.20 for several months now on two very different hosts.

The problem is random. It never seems to happen immediately after a boot, but randomly happens at other times. If you walk away from the system for a few hours it sometimes comes good by itself without rebooting.

On several occasions when it's happened, I have closed VirtualBox and launched several QEMU/KVM VMs via virt-manager no problem. After closing them all, there is a very strong chance (I believe for me so far it has always been the case) VirtualBox will run VMs without this problem.

Host system 1 is Fedora 40 with 32G RAM, host system 2 is Fedora 41 (still beta) with 16G RAM. My VMs are set for 4G. Needless to say the message is bogus, there is always plenty of RAM available.

This issue is driving me nuts.

[Busteros Julio]

Hola, no se el motivo pero se como replicar el problema, copia el archivo vdi a otra carpeta y luego prende la maquina virtual y te aparece de nuevo la falla.
No es la solucion pero es la forma en la que se replica el problema.
Hi, I don't know the reason but I know how to replicate the problem, copy the vdi file to another folder and then turn on the virtual machine and the error appears again.
It is not the solution but it is the way in which the problem is replicated.

[nixuser]

Here is a log of the event from my system.

[llr]

Found the reason - you do not have as much swap as the VM wants to start - more than enough memory.

For some reason it looks to see if your swap has enough free memory to load the VM - it shouldn't care but it does.

[GOF]

Hello

I am starting to face this issue this week as soon as I updated to the current version (had never seen this before): v. 7.1.6r167084
The unfortunate part is that I enabled swap and have plenty (it is basically empty at the moment) and I cannot get past it.

Code: Select all

               total        used        free      shared  buff/cache   available
Mem:            29Gi       7.0Gi       509Mi       1.3Gi        24Gi        22Gi
Swap:          8.0Gi       768Ki       8.0Gi

It seems that I have to reboot my entire machine to be able to start VMs. Additionally if the VM is running for a while, this will happen also. If I clone a VM, this also starts to happen immediately. I am suddenly having a lot of problems with the current version of VirtualBox, with this issue and several other problems all at the same time. It looks like the quality of this software is dropping for some reason.

[simono74]

For anyone with this issue, there is a ticket opened for this and it mentions a workaround to avoid restarting your pc ticket #22185. The workaround is to execute the command: echo 3 | sudo tee /proc/sys/vm/drop_caches
Hopefully they will fix it soon because it is ridiculous that we need to execute this command so frequently. However I am not holding my breath since it doesn't seem like they are even looking at this bug which has been opened for months now. It's ridiculous that VirtualBox is unable to properly determine the amount of available memory...

[klaus]

As the error message says, VirtualBox has trouble getting memory which can be used for "must be below 4G" kind of mapping. I would suspect it's the Linux kernel which does the associated resource management now in a worse way than before.

[GOF]

I gave up on the entire 7.1 branch and had to downgrade all the way to 7.0.24. interestingly I never observed this with 7.1.4, but after updating to 7.1.6 everything broke, and downgrading did not solve the issue. Removing and purging did not resolve issue either. Maybe something else got updated at the same time, so kernel or other package causing this is a possibility. Either way, 7.0 seems to be working reliably with everything else being equal.

[virtualhuman]

I am the poor person who opened bug #22185

Unfortunately Oracle has stopped responding to messages, the *new* version 7.1.6 contained the same broken code that causes this bug.

Indeed, the proposed temporary solution to purge the cache, works for me, but its not a real solution.

Oracle devs are unable to fix their own bug? incompetence? ignorance? who knows...

[MorganJR]

The workaround is to execute the command: echo 3 | sudo tee /proc/sys/vm/drop_caches
Hopefully they will fix it soon because it is ridiculous that we need to execute this command so frequently. However I am not holding my breath since it doesn't seem like they are even looking at this bug which has been opened for months now. It's ridiculous that VirtualBox is unable to properly determine the amount of available memory...
[/quote]

I can confirm that the above-mentioned workaround worked for me.

Thank you.

[hz2025]

I am also seeing this on Versions 7.0.x - I am using an AWS based metal machine i3 instance based on a Debian 12 AWS Marketplace AMI and running many machines in parallel. The metal machine has lots of CPU cores and RAM yet this still happens.

This issue wasn't a problem with a Debian 12 machine running on kernel 6.7.12. A version of the kernel between then till now (6.12/13) is somewhere where this issue has shown.

[KLiF]

Hello All,
I also have been experiencing this OOM issue. I am on 7.1.6_167084. I have done much experimentation including reloading the OS(Fedora Xfce 41). I have found the issue revolves, at least in my case, around the disk/media management. Anytime that I move, clone or change a media such as a disk, when finished and attempt to start up the VM, it will throw the error.
If I clear the caches, as suggested, it will resume normal function. Now I have lots of memory and large swap as well and it has nothing to do with any of the host setup. This is directly related to VBox Disk/Media management.
I have spent many years(35) in IT and one thing I know well from experience and tell all my Devs is, GARBAGE COLLECTION/TRASH CLEANUP! because most Devs rely on the JIT or system to cleanup after a process. In this case, I believe that this is the cause. There needs to be a code change in VBox Media management to clear the HEAP and caches and do some trash cleanup when the selected process finishes.

Do the suggested cache clear for now. I hope that the Oracle VB Devs will read and catch this and implement a fix soon.