Hello,
I am running a small virtualisation environment with two windows and two Linux guests.
I have installed Proxmox VE on top of Debian Jessie stable edition and I was updating it regularly from Jessie to Stretch and recently to Buster, following the update process on the Proxmox wiki page. I'm not using test repository, only the stable one.
You can clearly see when the server was updated. Usually with all 4 VMs on the load average was around 0.50, and now is around 1.00.
The problem is, that there is 0.50 load average more than usual no matter how many VM are running - 1 or 4, after the update. Furthermore, after all VM are shutdown, it takes time for the load average 0.50 to come close to 0.00.
Screenshot of top: the VMs are in idle state, but still there is some CPU activity.
Browsing around the forums I tried the following:
- changing the QemuServer.pm file adding:
push @$cpuFlags , 'hv_synic';
push @$cpuFlags , 'hv_stimer';
- Adding acpi=off in grub
- Enable/Disable KSM service.
- Disable tablet for pointer
I've read that I need to try: perf top, but I don't understand the things it shows me.
I rollbacked all changes I've made, because there was no effect.
After reboot, the load average is 0.00. As soon as a VM is started, 0.50 is added to the load average.
I vaguely remember that I had the same problem since the first installation of the server, but somehow I fixed it googling around. It is embarrassing that I can't remember how exactly
Googling is not helping anymore.
Could you please give me some hints how to trace the excessive load average?
Thank you
I am running a small virtualisation environment with two windows and two Linux guests.
I have installed Proxmox VE on top of Debian Jessie stable edition and I was updating it regularly from Jessie to Stretch and recently to Buster, following the update process on the Proxmox wiki page. I'm not using test repository, only the stable one.
You can clearly see when the server was updated. Usually with all 4 VMs on the load average was around 0.50, and now is around 1.00.
The problem is, that there is 0.50 load average more than usual no matter how many VM are running - 1 or 4, after the update. Furthermore, after all VM are shutdown, it takes time for the load average 0.50 to come close to 0.00.
Screenshot of top: the VMs are in idle state, but still there is some CPU activity.
Browsing around the forums I tried the following:
- changing the QemuServer.pm file adding:
push @$cpuFlags , 'hv_synic';
push @$cpuFlags , 'hv_stimer';
- Adding acpi=off in grub
- Enable/Disable KSM service.
- Disable tablet for pointer
I've read that I need to try: perf top, but I don't understand the things it shows me.
I rollbacked all changes I've made, because there was no effect.
After reboot, the load average is 0.00. As soon as a VM is started, 0.50 is added to the load average.
I vaguely remember that I had the same problem since the first installation of the server, but somehow I fixed it googling around. It is embarrassing that I can't remember how exactly
Googling is not helping anymore.Could you please give me some hints how to trace the excessive load average?
Thank you
Last edited: