I/O usage monitoring from host

[systemctl]

Hello.
Proxmox 7.1
Is there any way to monitor per-host I/O usage ? In previous versions i could do that by simple "iotop", that showed certain vms and their writes/reads. But now i only see "io_wqe_worker-*" that means nothing for me...

How do u guys monitor which VM uses a lot of disk resources?

Really frustrating for me, any help or suggestions are appreciate.

proxmox 7.1, debian11, qcow2 disks.

 

[LnxBil]

With an extrnal metric server, you're able to look at the I/O per VM.

 

[systemctl]

With an extrnal metric server, you're able to look at the I/O per VM.

So there is no build-in mechanics to monitor it with io_uring enabled by default ?
Cuz qm set ID --DISK OPTIONS,aio=native does the trick, so i think if i need to do it for every vm on cluster...

What external metric can u recommend for this?

 

[LnxBil]

So there is no build-in mechanics to monitor it

You mean besides the build in metrics aqcuisition that just sends all the statistics to an external metric server (e.g. influxdb)?

 

[dcsapak]

qm status ID --verbose

should also show quite some io statistics (per vm disk)

 

[systemctl]

qm status ID --verbose

should also show quite some io statistics (per vm disk)

ye, i know about that, but if i have like 100 vms on cluster and someone of them started using a lot of I/O it's kinda difficult to check them all to find the "criminal". Some time ago i just used iotop, but it's not possible with uring.

I will use external metric + influx + grafana. Ty very much for ur help guys.