RE: [PATCH v3 4/4] ceph: adding CEPH_SUBVOLUME_ID_NONE

[Viacheslav Dubeyko <Slava.Dubeyko@ibm.com>]

On Thu, 2025-12-04 at 10:18 +0200, Alex Markuze wrote:
> There is no separate test needed. The client only differs in that the
> mount path is for a subvolume.
> Regardless its outside the scope of this patchset
> 

If we implement any new functionality, then unit-test or special test in
xfstests must be introduced for any new functionality. So, I think it's very
relevant to the patchset.

Also, I believe that fourth patch should be merged into second and third ones. I
don't see the point of introducing not completely correct code and then fix it
by subsequent patch. It looks really wrong to me.

Thanks,
Slava. 

> On Thu, Dec 4, 2025 at 12:55 AM Viacheslav Dubeyko
> <Slava.Dubeyko@ibm.com> wrote:
> > 
> > On Wed, 2025-12-03 at 23:22 +0200, Alex Markuze wrote:
> > > The latest ceph code supports subvolume metrics.
> > > The test is simple:
> > > 1. Deploy a ceph cluster
> > > 2. Create and mount a subvolume
> > > 3. run some I/O
> > > 4. I used debugfs to see that subvolume metrics were collected on the
> > > client side and checked for subvolume metrics being reported on the
> > > mds.
> > > 
> > > Nothing more to it.
> > > 
> > 
> > So, if it is simple, then what's about of adding another Ceph's test into
> > xfstests suite? Maybe, you can consider unit-test too. I've already introduced
> > initial patch with Kunit-based unit-test. We should have some test-case that
> > anyone can run and test this code.
> > 
> > Thanks,
> > Slava.
> >