I have notice that unraid becomes unresponsive affecting docker and Unraid itself when a VM is stressing the CPU, up to the point to kill unraid and require a hard shutdown.

 

Has happens while doing a benchmark, and after another VM that was freeze and Unraid wasn't able to kill it.

 

This doesn't happens in promox or hyperv so I guess they have some kind of mechanism to ensure hypervisor stability. I think the implementation of KVM in unraid could be done better and there should be a way to manage the resource sharing between docker and VM.

In promox you can share cores among VM but assign preference for a VM.

 

Am I the only one experiencing this? I have had to restore unraid from backup 2 times in 24 hours because of this. I was using 3 VM win 10 and win 2016 and Pfsense.

 

Could promox be integrated in unraid somehow? Or parts of it? @limetech

https://pve.proxmox.com/wiki/Open_Source

 

Edited March 31, 20188 yr by L0rdRaiden

When setting up the VM did you remember to NOT assign all cores to the VM?     You should always leave at least one CPU (and if hyper-threading it’s partner) free for unRAID to use.    Often this is not critical but if you are trying to run CPU intensive tasks in the VM it becomes more important.

Edited March 31, 20188 yr by itimpi

1 hour ago, itimpi said:

When setting up the VM did you remember to NOT assign all cores to the VM?     You should always leave at least one CPU (and if hyper-threading it’s partner) free for unRAID to use.    Often this is not critical but if you are trying to run CPU intensive tasks in the VM it becomes more important.

The first time I had 8/8 cores, second time 4, maybe the second time there was something corrupt in the virtual disk.

 

But still in hyperv I can assign 8 cores benchmark or run a cpu stress test and hyper v doesn't die and the other VM will go slow but won't die.