[SOLVED] High cpu usage after migrating

[corin.corvus]

Hi,

i migrate a VM to another node. After Migration the VM is on 100% CPU. Why and how i can solve it?
For Workaround i can reset the VM. After i have normal CPU Usage.





After Reset the VM:


Thanks

 

[shanreich]

What kernel version are you running? What CPU models are you using on the respective nodes?

Have you tried using kernel version 6.1 [1] (or 5.19 [2]) and see if the problem persists?

[1] https://forum.proxmox.com/threads/opt-in-linux-6-1-kernel-for-proxmox-ve-7-x-available.119483/
[2] https://forum.proxmox.com/threads/opt-in-linux-5-19-kernel-for-proxmox-ve-7-x-available.115090/

 

[corin.corvus]

What kernel version are you running? What CPU models are you using on the respective nodes?

Have you tried using kernel version 6.1 [1] (or 5.19 [2]) and see if the problem persists?

[1] https://forum.proxmox.com/threads/opt-in-linux-6-1-kernel-for-proxmox-ve-7-x-available.119483/
[2] https://forum.proxmox.com/threads/opt-in-linux-5-19-kernel-for-proxmox-ve-7-x-available.115090/

I never changed the kernal manually.
Kernelversion is on the screenshots:

Node 1:
Linux 5.15.83-1-pve #1 SMP PVE 5.15.83-1 (2022-12-15T00:00Z
AMD Ryzen 5600G

Node 2:
Linux 5.15.74-1-pve #1 SMP PVE 5.15.74-1 (Mon, 14 Nov 2022 20:17:15 +0100
Intel Celeron J4105

 

[shanreich]

Can you try using one of the newer kernels and see if that problem persists?

In general, migration between different CPU models - especially between AMD and Intel is not supported and usually causes many problems.

 

[corin.corvus]

Can you try using one of the newer kernels and see if that problem persists?

In general, migration between different CPU models - especially between AMD and Intel is not supported and usually causes many problems.

It doesnt work on my second cluster with 2 intel cpus too, but at the moment i dont have ram for the test cluster, cause i cant test.

how i can use newer kernels?
now i did updates on the celeron node. now they have the same kernel.

 

[shanreich]

how i can use newer kernels?

you can just upgrade to the newer kernel by running

apt update
apt install pve-kernel-6.1

reboot

 

[corin.corvus]

Node 1:

CPU(s) 12 x AMD Ryzen 5 5600G with Radeon Graphics (1 Socket)

Kernelversion Linux 6.1.0-1-pve #1 SMP PREEMPT_DYNAMIC PVE 6.1.0-1 (Tue, 13 Dec 2022 15:08:53 +0

PVE-Manager-Version pve-manager/7.3-4/d69b70d4

Node 2:

CPU(s) 4 x Intel(R) Celeron(R) J4105 CPU @ 1.50GHz (1 Socket)

Kernelversion Linux 6.1.0-1-pve #1 SMP PREEMPT_DYNAMIC PVE 6.1.0-1 (Tue, 13 Dec 2022 15:08:53 +0

PVE-Manager-Version pve-manager/7.3-4/d69b70d4

Testmigration now with panic kernel



After Reset it works

 

[shanreich]

Since those are different CPU vendors, this is very likely the issue. Even with different models from the same vendor live migration can lead to issues.
Did you try on the second cluster with only Intel CPUs and check if it works there at least?

 

[corin.corvus]

hm... in the past (7.2) it works :/ Can i change the migration mode?
The Node with Ryzen 5600g and the only one not damaged Dell wyse (celeron) worked fine with HA on Proxmox 7.0-7.2. Last month i changed to 7.3 and now it wont work. HA work fine,

On my second Cluster (Test Nodes with damaged Ram 3x Dell Wyse 5070) it didnt work every time (7.1, on 7.2 the ram was damaged).

fyi: Migration from Node 2 to 1 work fine:


Its only Migration from 1 to 2 with the CPU 100% problem.

 

[shanreich]

Maybe you can try downgrading the kernel then to 5.13? You can use the proxmox-boot-tool for this. Additionally, updating the firmware of the BIOS might help.

 

[corin.corvus]

I dont know how, but today the migrations work fine.

 

[corin.corvus]

Hey @shanreich,

one last question. The Migration is running sometimes correct, sometimes with 100% CPU between AMD/Intel and Intel/Intel:




I dont know why, but so often i dont migrate.

The Question:
If i migrate to N-3 i get much IO-Delay (30-70%). If i migrate to N-2 i get no IO-Delay. May the configurates Proxmox Drive the problem? On N-2 i didnt change anything, on N-3 i increased the Swap to 30gb. The Drive is 64gb big (M.2 Sata SSD). This is the only different Config of these Nodes.





Thanks

Add:
Ok no, the Swap or Disk Configuration is not the Problem... hm.

Solved: The Controller on the Mainboard was not correct configured.