[pve-devel] [pve-manager] Adding real disk usage information (discussion)

Bruce Wainer brwainer at gmail.com
Wed Apr 21 03:37:08 CEST 2021


Dominik,
Thank you for the insight. There is certainly complexity I did not
consider, even if I were to look at the narrow case of local ZFS storage.
Regardless, this would be helpful to me and if I make anything then I will
submit it. I already have signed the CLA and have code accepted in
pve-zsync.
Thank you,
Bruce

On Mon, Apr 19, 2021 at 2:55 AM Dominik Csapak <d.csapak at proxmox.com> wrote:

> On 4/16/21 22:18, Bruce Wainer wrote:
> > Hello,
> >
>
> Hi,
>
> > I am interested in seeing real disk usage information for VM Disks and CT
> > Volumes, on storage types that have thin provisioning and/or snapshots.
> > Specifically I would like to see "Current Disk Usage (Thin)" and either
> > "Snapshot Usage" or "Total Disk Usage". I only use local ZFS on servers
> at
> > this time, but I'm sure the GUI side would be best made flexible.
>
> while this sounds sensible, this will get hard very fast.
> For example, take a LVM-Thin storage.
>
> I have a template which has an LV which uses some space.
> This can have X linked clones, where each clone can have Y snapshots.
>
> since lvmthin lvs/snapshots/etc. are only very loosely coupled.
> It is very hard to attribute the correct number to any
> of those vms/templates. (e.g. do you want to calculate the
> template storage again for each vm? only once? what if
> you cloned a vm from a snapshot?)
>
> It gets even harder on storage that can deduplicate (e.g. ZFS) or
> where the 'real' usage is dynamically inflated by some form of replica
> (e.g. Ceph).
>
> So, while this sounds nice, and we would probably not oppose a clean
> solution, this is not a trivial problem to solve.
>
> >
> > Is someone interested in helping with this? Where would I start,
> especially
> > on the GUI part, if I were to develop this myself?
>
> anyway, to answer this question, the storage plugins in the backend can
> be found in the pve-storage git repo[0]
>
> the point where the status api calls of the vms/cts are called live
> in qemu-server[1] and pve-container[2] respectively
> (the api part is in PVE/API2/)
>
> you can find the gui part in pve-manger[3] in www/manager6
>
> also if you want to send patches, please read the developer
> documentation [4] especially the bit about the CLA
>
> if you have any more question, please ask :)
>
> hope this helps
> kind regards
>
> 0:
>
> https://git.proxmox.com/?p=pve-storage.git;a=tree;f=PVE/Storage;h=fd53af5e74407deda65785b164fb61a4f644a6e0;hb=refs/heads/master
> 1: https://git.proxmox.com/?p=qemu-server.git;a=summary
> 2: https://git.proxmox.com/?p=pve-container.git;a=summary
> 3:
>
> https://git.proxmox.com/?p=pve-manager.git;a=tree;f=www/manager6;hb=refs/heads/master
> 4: https://pve.proxmox.com/wiki/Developer_Documentation
>
> >
> > Thank you,
> > Bruce Wainer
> > _______________________________________________
> > pve-devel mailing list
> > pve-devel at lists.proxmox.com
> > https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
> >
> >
>
>
>



More information about the pve-devel mailing list