RAM Overcommitment Issue After VM Suspend-Resume in Workstation 17 Pro

[yalmasri]

Hello,

I’ve encountered a RAM usage concern with VMware Workstation 17 Pro (v17.5.1) on Windows 11, specifically involving AlmaLinux 9.3 VMs. My setup includes 48GB of host RAM, with VMs allocated 16GB each. Initially, VMs consume about 1.5GB upon startup, but after a suspend-resume cycle, they’re locking down the full 16GB allocation without release, dramatically affecting my system's resource availability.

Is there a way to prevent this full RAM reservation by VMs post-resume? Any advice or workaround would be greatly appreciated.

Thanks!

[mhb4ever]

 Edit     >    Preferences   >   Memory

[yalmasri]

Thanks for the reply.

Actually I feel there's misunderstanding to the situation I'm seeing. When the VM is powered on, memory utilization within the VM is very low, say 2GB, and that's actually what's being consumed from my host, so it's expected that suspending/resuming the VM will release/reserve those 2GB to my host. However, what happens is that a 2GB is in fact released upon suspension, but a whole chunk of 16GB is reserved upon resumption (the actual memory configured for the VM). I validated this behavior by reducing the memory allocation to 8GB, and after the suspend-resume cycle, the amount of memory reserved from my system was 8GB, despite the fact that the actual consumption of the VM before this cycle was less than 2GB.

Here's my Memory config for reference:

[mhb4ever]

check 

Fit all virtual machine memory into reserved host RAM

and try 

[scott28tt]

What memory usage does the guest OS show on resume?

[yalmasri]

I temporarily decreased the RAM to 8GB to avoid my system running out of resources.

Here's the memory usage after resume:

[root@k8s-worker-1 ~]# cat /proc/meminfo
MemTotal: 7825224 kB
MemFree: 5057436 kB
MemAvailable: 6277536 kB
Buffers: 4564 kB
Cached: 1419812 kB
SwapCached: 0 kB

Pretty much the same on the other machine (which also was suspended-resumed):

[root@k8s-worker-2 ~]# cat /proc/meminfo
MemTotal: 7825252 kB
MemFree: 5602788 kB
MemAvailable: 6535764 kB
Buffers: 3180 kB
Cached: 1135980 kB
SwapCached: 0 kB

I'm wondering, is this behavior intended and it's designed this way? Can I change it?

[yalmasri]

Anyone can suggest anything to try? This behavior is really annoying.