Hi all,
A while ago we experienced failing back-ups that 'sometimes' failed while using the start/stop back-up method in the vzdump job. The previous work-around was upgrading the host to Proxmox 9.x.
Unfortunately this error has returned after a while, although not as frequently as on Proxmox 8.x
The behavior seems to go as follows:
> Back-up starts, QM calls a shutdown to VMID_X > TAP/FWBR devices to disabled state > systemd reports VMID_X.scope stopped successfully > pvescheduler VMID_X qmp command failed (expected) > systemd cannot move old KVM PID to cgroup /qemu.slice/VMID_X.scope "no such process" > systemd Failed to start VMID_X.scope > qm event cleanup VMID_X
After this the back-up for that VM fails and continues onto the next VM. Oddly enough some VMs go fine, then once this problem happens all VMs that have not been backed-up yet also fail in a sort of domino-effect.
One other thing is that pvestatd keeps reporting 'Use of uninitialized value in multiplication (*) at /usr/share/perl5/PVE/QemuServer.pm line 262(5-6-7)' during the back-up process.
As a workaround the snapshot method does work without issues, atleast for 'the time being'.
We're running the latest PVE-enterprise version for all packages.
Has anyone else experienced this issue before?
A while ago we experienced failing back-ups that 'sometimes' failed while using the start/stop back-up method in the vzdump job. The previous work-around was upgrading the host to Proxmox 9.x.
Unfortunately this error has returned after a while, although not as frequently as on Proxmox 8.x
The behavior seems to go as follows:
> Back-up starts, QM calls a shutdown to VMID_X > TAP/FWBR devices to disabled state > systemd reports VMID_X.scope stopped successfully > pvescheduler VMID_X qmp command failed (expected) > systemd cannot move old KVM PID to cgroup /qemu.slice/VMID_X.scope "no such process" > systemd Failed to start VMID_X.scope > qm event cleanup VMID_X
After this the back-up for that VM fails and continues onto the next VM. Oddly enough some VMs go fine, then once this problem happens all VMs that have not been backed-up yet also fail in a sort of domino-effect.
One other thing is that pvestatd keeps reporting 'Use of uninitialized value in multiplication (*) at /usr/share/perl5/PVE/QemuServer.pm line 262(5-6-7)' during the back-up process.
As a workaround the snapshot method does work without issues, atleast for 'the time being'.
We're running the latest PVE-enterprise version for all packages.
Has anyone else experienced this issue before?
