October 28, 20214 yr For some reason, unraid is eating the ram like nothing and in turn turning off VM when underload. I have only 2 VM on the server and 6-7 docker (yes i know they need ram), but when i took the free -mh, i turned off the docker completely from the setting and then rebooted the unraid server. This is fresh off the boot. available only 433mb. However i have 32gb ram of which 20gb is allocated to Windows VM & 2.5gb is allocated to linux VM. Thats 22.5gb, and yet it says 28gb is used. This remains same weather i set the Windows VM to 20gb or 25gb. So something is clearly wrong here. Diagnostic attached too. PS: in windows vm i am using hugepages. My linux VM goes down the moment server runs out of ram to allocate. Every 2.0s: free -mh Tower: Thu Oct 28 16:11:05 2021 total used free shared buff/cache available Mem: 31Gi 28Gi 644Mi 1.8Gi 1.9Gi 433Mi Swap: 0B 0B 0B Attached pic of htop too. tower-diagnostics-20211028-1615.zip
October 28, 20214 yr Author If i shut down my Windows VM, the ram still remains 99% full. Even after shutting down.
October 28, 20214 yr I think you are legitimately running out of memory. I don't see this as a "runaway issue". When you use hugepages for VMs, that memory is reserved for the VM even when its shut down. That is the one major downside to using hugepages: you cannot dynamically utilize that memory. In addition, beyond the memory assigned to the VMs you have, memory is also utilized by QEMU/KVM for emulating components for those VMs. That utilizes additional memory. So between the 20GB being assigned to your Windows VM via Hugepages, another 2.5GB for your Linux VM, that leaves you with less than 10GB remaining for all of your applications and the emulation required for your VMs as well as any additional overhead for filesystem operations (which can vary). I think you're butting up against the upper limits of your memory usage and should consider reducing your Windows VM to 16GB (along with the hugepages reservation you created). All the best, Jon
October 30, 20214 yr Author On 10/28/2021 at 11:21 PM, jonp said: I think you are legitimately running out of memory. I don't see this as a "runaway issue". When you use hugepages for VMs, that memory is reserved for the VM even when its shut down. That is the one major downside to using hugepages: you cannot dynamically utilize that memory. In addition, beyond the memory assigned to the VMs you have, memory is also utilized by QEMU/KVM for emulating components for those VMs. That utilizes additional memory. So between the 20GB being assigned to your Windows VM via Hugepages, another 2.5GB for your Linux VM, that leaves you with less than 10GB remaining for all of your applications and the emulation required for your VMs as well as any additional overhead for filesystem operations (which can vary). I think you're butting up against the upper limits of your memory usage and should consider reducing your Windows VM to 16GB (along with the hugepages reservation you created). All the best, Jon Thanks for the input Jon, However there is literally no othe process running on my server, i literally turned off docker, and even in htop i can only see qemu only. I did a full shutdown and disconnected the pc for a while and then started back, its much better now, but still missing ram. total used free shared buff/cache available Mem: 31Gi 28Gi 215Mi 685Mi 2.4Gi 1.6Gi Swap: 11Gi 1.4Gi 10Gi 20G with VM and 2.5G static with linux. Thats 22.5G, i still dont know where the rest 5.5G is, i cannot see anything. Even if i consider 2.4G in buff cache, i am still missing 3G of ram. Its much better now as i have 1.6G ram available. Atleast its smooth and operational.
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.