[pbs-devel] [PATCH proxmox-backup] api: add dedicated datastore disk schema

[Christian Ebner]

> On 29.11.2023 09:56 CET Thomas Lamprecht <t.lamprecht at proxmox.com> wrote:
> 
>  
> Am 29/11/2023 um 09:24 schrieb Christian Ebner:
> >> On 28.11.2023 18:49 CET Thomas Lamprecht <t.lamprecht at proxmox.com> wrote:
> >> this is not really helping me, while that might be because its late, I also think
> >> that this needs a different name, as what is a datastore-disk here actually?
> > 
> > The intention was to distinguish between the datastore itself and the disk, on which the datastore might be created upon, as these are 2 independent entities. Most of the time they will be referred to by the same name, as e.g. created by `proxmox-backup-manager disk fs create foo --disk sdd --add-datastore true`.
> > 
> 
> The base intention I get, and I think too that this could/should be
> improved.
> 
> > I choose to refer to this as `datastore disk` as that is what the API methods in src/api2/node/disks/directory.rs are called.
> > 
> > Unfortunately, that does not really match up with ZFS, as there the name refers to the zpool.
> > 
> 
> Why not STORAGE_NAME_SCHEMA then? As this is used as datastore name
> is only for convenience and might not happen if one doesn't ticks the
> "add as datastore" checkbox.
> So having this naming hardwired to datastore or disk, while it isn't
> necessarily related to either type, is IMO just not that ideal.

Ah yes, referring to it as storage name is way better, didn't come to mind yesterday.
This is generic enough to cover all cases.
 
> 
> As description you could then use something like:
> 
> "Name of the Storage. Will be also used for any datastore name, if part of an transaction."

Maybe I will reduce this to a more compact:
"Storage name. Implies the datastore name, if part of a transaction."

Any opinions on that?

> 
> (not really ideal as it's worded a bit overly generic, but just to convey
> the base idea).