Incomplete disk IO stats after upgrading to PVE8

[keeka]

Following the upgrade to PVE8, I notice in the web UI two containers that show zero Disk IO. Prior to the upgrade, these displayed disk read/write stats.
All other containers and VMs are displaying levels of disk IO as before.

The two CTs concerned are both unpriviledged Debian 11 containers hosting docker and both appear to be working, as before the upgrade.

I wondered if anyone else has noticed this?

 

[keeka]

I just created a new debian11 container and generated some IO. It appears in the summary page charts.
I then installed docker, started a simple container:
The pve host search page shows disk read, but zero disk write. Where as the guest summary Disk IO shows zero read and zero write. So, also some apparent inconsistency there too.

 

[exellated]

Having practically the same issue on a fresh installation of PVE 8.1.
I'm using containers with Ubuntu 20.04 template and after installing docker and docker-compose in the container seems the "Disk IO" statistics becomes broken (zero diskread/diskwrite data on "Statistics" page of the container). I have another node running PVE 7.1 and there is no such problem here.

 

[lucius_the]

Having the same issue. Upgraded from PVE7 to PVE8 (currently on 8.1.4).
LXC disk statistics in the GUI show 0/0 from the moment I upgraded from PVE 7.

Any chance to fix this ?

 

[keeka]

After an apt dist-upgrade and host reboot yesterday evening, my CT docker guests' disk graphs have sprung to life and are reporting IO again.

The upgraded packages:

libpve-common-perl:all 8.2.2
qemu-server:amd64 8.2.4
pve-docs:all 8.2.3
proxmox-kernel-6.8.12-1-pve-signed:amd64 6.8.12-1
pve-manager:amd64 8.2.4
man-db:amd64 2.11.2-2
pve-ha-manager:amd64 4.0.5

I am guessing this is down to changes in either the new kernel or libpve. I wonder, what changed to cause this?