PVE provisioned disk space in GUI is missing ....

[Petr Svacina]

Hi, It is possible to include used provisioned space overview in the GUI ?
Because right now I can see only used space in the storage window, but not ALL provisioned ...
And when, thin provisioning is used by default on the CEPH and ZFS storage, admin can be confused and storage can by overspaced, when you do not used commands in the cli, to check the status, like this:

rbd du -p ceph-ssd

<TOTAL> 2.7 TiB 1.3 TiB

So I need the information about 2,7TiB provisioned, when I have 3TiB total space only ....

 

[Petr Svacina]

Hello, can anybody from proxmox answer ? Thanks a lot ...

 

[aaron]

There is already an enhancement request in the bug tracker.

Unfortunately, implementing something like this in a way that you can always trust it, is not easy, as is explained in the bug tracker and why it is still open .

On a ZFS pool or Ceph pool, you could also have other factors at play that can reduce the available size, such as multiple ZFS sub datasets using up space on the same ZFS pool. On Ceph, a full OSD will reduce the available space, multiple pools on the same cluster or namespaces within a pool can also increase the used space even though the pool/namespace you are interested did not change at all.

The sensible way to handle thin provisioned storages and storages where other factors might reduce the available space is to keep a close eye on them and have some monitoring in place that will trigger a warning early enough. The Ceph Manager for example has quite a few modules to integrate with 3rd party monitoring solutions.

 

[Petr Svacina]

Yes: " is to keep a close eye on them" this is exactly what i do NOT want to do . Because if more admins take care about the pool, they always must have on their minds: I must count what is full provisioned, when I want to make a new VM, or expand the disk for a some VM.

And when after login to GUI, we will see the all provisioned disk space on the edge of the pool(s), everybody can handle disk space on the cluster more safely ...

Ok. thanks for the reply.

 

[melroy89]

I hope you could use the guest-agent if needed, if you want to receive the storage values... For instance I'm using LVM as my OS disk.
I think the guest agent should be able to read the usage of /dev/mapper/ubuntu--vg-ubuntu--lv, just like I can do using the df -H command.

 

[Petr Svacina]

This is the way

 

[melroy89]

I noticed I got a bit confused, but this thread seems to talk about "provisioned" disk usage, which is the total disk usage of the VM I guess.

While I'm (also) trying get the VM usage inside the disk by leveraging the guest agent if needed, in order to retrieve the disk usage of the disks inside the VM. https://qemu-project.gitlab.io/qemu/interop/qemu-ga-ref.html#qapidoc-158

If a disk gets too full, we can act in time and extend the virtual disk if needed. A bit simular how you currently show the IP addresses of the VM in the overview page.