Fixing khugepaged CPU usage VMware Workstation

vmware-workstation-khugepaged-fix.md

If you run VMware Workstation 11 or above you may encounter high CPU usage from process khugepaged on Ubuntu 15.04+

The fix is to disable transparent hugepages. It seems Ubuntu has it enabled by default.

You can check the current status on your system by running:

cat /sys/kernel/mm/transparent_hugepage/enabled

cat /sys/kernel/mm/transparent_hugepage/defrag

Fedora outputs: always [madvise] never but Ubuntu outputs: [always] madvise never

Fedora seems to not be effected but I havn't tested it myself.

So I suggest not using madvise and just disable it totally.

To disable it run the following commands as root:

echo never > /sys/kernel/mm/transparent_hugepage/enabled

echo never > /sys/kernel/mm/transparent_hugepage/defrag

That will only disable it for the current session.

To have it persistant across reboots I suggest adding this to your rc.local:

# Fix for VMware Workstation 11+ khugepaged.
echo never > /sys/kernel/mm/transparent_hugepage/enabled
echo never > /sys/kernel/mm/transparent_hugepage/defrag

Ensure this goes above the line:

exit 0

More info and references:

• https://communities.vmware.com/thread/505923
• https://docs.mongodb.com/manual/tutorial/transparent-huge-pages/

@neves-0 is right, that’s the most efficient way, although I personally have sometimes additional logic around these things and keep them as sysctl calls in a script for that reason.

@slapbox you need to ask Linux folks why systemd :-)..

Testing went well with the Arch linux-zen kernel; thanks again.

For Arch users, create a .conf file in /etc/sysctl.d/.

Setting vm.compaction_proactiveness to be 0 doesn't solve the problem for me...

Same for me.

None of the above works for me. Did the hugepage_transparent things, the compaction_proactiveness, both directly, in etc/, and in vmware *.vmx file. Checking the various variable from /proc and /sys returns the expected value. Still, kcompactd0 kicks in. It is slightly less severe than it uses to be, but still it freezes the guest to death.

I'm thinking about recompiling the kernel and disabling that kcompactd0 thing altogether.

Host Debian 5.16.18-1 (2022-03-29) x86_64 GNU/Linux, 16GB RAM.
VMWare 16.2.3 build-19376536
Guest Windows 10 (64-bit), 8GB RAM allocated.

When I first set only:
vm.compaction_proactiveness=0
back in December it appeared a complete fix, just like a number of previous (but different) fixes did.

However I now see some regular pauses, but much less serious ones. I might lock up occasionally for 10-30 seconds, but generally the problem clears after a few iterations and the system is fine again. It's an annoyance, for sure, but it doesn't block my work.

In case it lends any light on the problem, I am using a 12-core AMD with 64 GB of memory on Ubuntu 20.04. I have seen this problem in Windows VMs with 16 GB of virtual RAM, and 4 or 6 virtual CPUs.

I fear that this fix, like all the others, will gradually become less effective, finally making VMWare Workstation unusable. Previously this problem became so bad I temporarily switched to VirtualBox, and never saw the problem there.

i can confirm what gene-olson says.
i started get freezes even with compaction_proactiveness is 0.

Comfirm that win10 guest freezes and kcompactd0 kicks in with compaction_proactiveness=0.

Tested with:
Ubuntu 22.04 kernel 5.15.35 / 5.16.20
VMWare 16.2.3 / 16.2.0 / 16.1.2
Win10 Guest, 4GB / 6GB memory allocated

i have lowered the memory for 3d graphics and its better.
my config is:
ryzen 5 5600g / 32 gb ram
guest is win10 with 6 cpu and 16 gb ram.
lowered graphics to 256mb.

i have lowered the memory for 3d graphics and its better. my config is: ryzen 5 5600g / 32 gb ram guest is win10 with 6 cpu and 16 gb ram. lowered graphics to 256mb.

same here, reduced to 1Gb and its fine now

I've been dealing with this issue for months on Debian 11. None of the mentioned methods were the solution for me. And today i solved my problem. What was described here helped to solve my problem.

These three lines as root should fully disable kernel memory defragmentation:

echo never > /sys/kernel/mm/transparent_hugepage/defrag
sysctl -w vm.compaction_proactiveness=0
sysctl -w vm.extfrag_threshold=1000

Note that this will greatly increase memory fragmentation and therefore memory pressure as compation is fully disabled and should be reverted when VMWare is not in use.

How to revert:

sysctl -w vm.compaction_proactiveness=20
sysctl -w vm.extfrag_threshold=500
echo always > /sys/kernel/mm/transparent_hugepage/defrag
sysctl -w vm.compact_memory=1

I tried the above options by @msizanoen1. Indeed the VM is now usable. However, it is still slow and I see that kswapd0 now occasionally will peg 1 CPU at 100%, despite the fact that I have no swap enabled. This process is only running when I run a VMware VM, which suggests this is somehow linked...

I tried the above options by @msizanoen1. Indeed the VM is not usable. However, it is still slow and I see that kswapd0 now occasionally will peg 1 CPU at 100%, despite the fact that I have no swap enabled. This process is only running when I run a VMware VM, which suggests this is somehow linked...

It's likely that not using swap was the cause for kswapd consuming 100%. Generally it's not recommended to run a Linux system without some kind of swap, and this might be especially true when running with memory defragmentation disabled.

Try setting /proc/sys/vm/compaction_proactiveness to 1. The thing is, you need to compact eventually. What you want to avoid is a war between VMware and kernel. This will obvisouly happen if you compact too eagerly, which is the default. But if you compact too lazily, for example by not being proactive at all, you will run into a situation where you must do it reactively, with the same bad result. This is as if everyone has kept missing the proactiveness part. Setting it to 1 has now worked for me for some time, at least weeks, if not months.

Try setting /proc/sys/vm/compaction_proactiveness to 1. The thing is, you need to compact eventually. What you want to avoid is a war between VMware and kernel. This will obvisouly happen if you compact too eagerly, which is the default. But if you compact too lazily, for example by not being proactive at all, you will run into a situation where you must do it reactively, with the same bad result. This is as if everyone has kept missing the proactiveness part. Setting it to 1 has now worked for me for some time, at least weeks, if not months.

AFAIK (and through my own testing and reading of the kernel source) setting vm.extfrag_threshold=1000 and disabling transparent hugepage defrag will prevent the kernel from ever compacting memory, reactively or not, and will cause it to fall back to swapping pages out of memory and/or invoking the OOM killer instead.

vm.compaction_proactiveness and /sys/kernel/mm/transparent_hugepage/defrag controls different aspects of proactive memory compaction, while vm.extfrag_threshold controls reactive memory compaction (e.g. when the kernel needs to allocate a large chunk of continuous memory). Setting vm.extfrag_threshold to 1000 disables reactive memory compaction.