shadow_copy2_realpath() snapshot path component removal

[David Disseldorp]

On Thu, 13 Feb 2014 14:36:14 -0800
Jeremy Allison <jra at samba.org> wrote:

> > > I expect that's just a side-effect of VSS doing the COW snapshot at the block layer.
> > > I.e. The filesystem/SD permit write access, but the IOs are rejected at the device layer.  
> > 
> > Of course it is, but that is not the point. Users get very unhappy
> > when it does not behave the way it behaves on Windows.  
> 
> New test case please inside smbtorture4 ! That way we can at least know when
> we're working :-).

Sadly, testing of snapshot IO against a Windows server is difficult to
automate. The only mechanism to trigger a snapshot remotely via SMB is
using FSRVP, and AFAICT on Windows FSRVP created snapshots are not
exposed for access via previous file version (@GMT/timewarp token)
requests.

Cheers, David