Getting an error:
# apt install pve-qemu-kvm_8.1.2-4
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
E: Unable to locate package pve-qemu-kvm_8.1.2-4
PVE 100%CPU on all kvm while vms are idle at 0-5% cpu
- Thread starter Flaxplax
- Start date
-
- Tags
- cpu pve 8.1.3 troubleshooting
# apt install pve-qemu-kvm_8.1.2-4
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
E: Unable to locate package pve-qemu-kvm_8.1.2-4
I just posted about that issue.
I'm running PVE 8.1.3. From 4 VM's running, 2 of them were stuck at 50% CPU usage at the same time the backup job started at 3am, however when I check the CPU usage inside the VM nothing stands out and everything seems to be normal.
Cheers,
Xzi.
I'm running PVE 8.1.3. From 4 VM's running, 2 of them were stuck at 50% CPU usage at the same time the backup job started at 3am, however when I check the CPU usage inside the VM nothing stands out and everything seems to be normal.
Cheers,
Xzi.
I apologize, I should have read this thread first before posting. It looks like my issues are related:
https://forum.proxmox.com/threads/cpu-usage-goes-to-100-after-backup-job-completes.138303/
Is the recommendation to downgrade to pve-qemu-kvm_8.1.2-4 or wait until 8.1.2-6 becomes available?
https://forum.proxmox.com/threads/cpu-usage-goes-to-100-after-backup-job-completes.138303/
Is the recommendation to downgrade to pve-qemu-kvm_8.1.2-4 or wait until 8.1.2-6 becomes available?
Hi there,
i was just debugging the same problem, when i thought why don't give the forum a try.
So my findings.
My symptoms:
After a backup via PBS, the higher CPU usage of every KVM (debian, freebsd) stays at the same level after completion of the Backup as it had have while during the Backup. So, before the backup 2% during the backup up to 25% and even after compeltition i am stuck at the 25%. (per KVM)
A Reboot of the KVM or a Migration does solve the problem until the next Backup-Run.
During The Backup: (perf top/PVE NODE)
INFO: Finished Backup of VM 106 (00:00:03)
INFO: Backup finished at 2023-12-17 21:55:53
INFO: Backup job finished successfully
INFO: notified via target `mail-to-superwichtig`
TASK OK
And then, after the Backup (perf top/PVE NODE):
After Reboot of VM 106 (perf PVE NODE):
At the moment i'am unsure if the problem source is vdso (kvm-clock) or is this the side effect of the many qemu-sys-x86 Shared Object which might "hammer" vdso.
best regards,
fne
i was just debugging the same problem, when i thought why don't give the forum a try.
So my findings.
My symptoms:
After a backup via PBS, the higher CPU usage of every KVM (debian, freebsd) stays at the same level after completion of the Backup as it had have while during the Backup. So, before the backup 2% during the backup up to 25% and even after compeltition i am stuck at the 25%. (per KVM)
A Reboot of the KVM or a Migration does solve the problem until the next Backup-Run.
During The Backup: (perf top/PVE NODE)
INFO: Finished Backup of VM 106 (00:00:03)
INFO: Backup finished at 2023-12-17 21:55:53
INFO: Backup job finished successfully
INFO: notified via target `mail-to-superwichtig`
TASK OK
And then, after the Backup (perf top/PVE NODE):
After Reboot of VM 106 (perf PVE NODE):
At the moment i'am unsure if the problem source is vdso (kvm-clock) or is this the side effect of the many qemu-sys-x86 Shared Object which might "hammer" vdso.
best regards,
fne
Sorry for the Monolog.
I exchanged the qemu-system-x86_64 binary (Version 8.1.2-5) with the last one (8.1.2-4)
The behaviour is now back to normal - during the backup:
After the Backup:
INFO: Finished Backup of VM 106 (00:02:36)
INFO: Backup finished at 2023-12-17 22:35:26
INFO: Backup job finished successfully
INFO: notified via target `mail-to-superwichtig`
TASK OK
best regards,
fne
I exchanged the qemu-system-x86_64 binary (Version 8.1.2-5) with the last one (8.1.2-4)
The behaviour is now back to normal - during the backup:
After the Backup:
INFO: Finished Backup of VM 106 (00:02:36)
INFO: Backup finished at 2023-12-17 22:35:26
INFO: Backup job finished successfully
INFO: notified via target `mail-to-superwichtig`
TASK OK
best regards,
fne
Last edited:
Exactly the samt here, but CPU goes up to 100% on my VMs. So they stuck at 100% but the physical machine cant handle all VMs @ 100%, so its really loud too. I can add, Windows11 VMs also affected.
Reboot solves the issue temporarily. Created a cronjob for a scheduled reboot in the early morning after the backups are finished.
Reboot solves the issue temporarily. Created a cronjob for a scheduled reboot in the early morning after the backups are finished.
what is the correct way to switch back as soon as 8.1.2-6 is out? just run apt remove ve-qemu-kvm_8.1.2-4?
pve-qemu-kvm_8.1.2-6_amd64.deb is latest, but test (http://download.proxmox.com/debian/pve/dists/bookworm/pvetest/....) binary
It will be in a stable version soon, I think
It will be in a stable version soon, I think
Is 8.1.2.5 still in subscription repos? Or was it never in there?
I'm not sure if I can upgrade production cluster
I'm not sure if I can upgrade production cluster
Anyone else still experiencing this issue after updating pve-qemu-kvm_8.1.2-6? Once my weekly backups ran, the same CPU usage problems returned. I did reboot after the update to -6.
