[libvirt] [PATCH 18/16] snapshot: Consolidate writing snapshot state to disk

[Daniel P. Berrangé]

On Fri, Mar 22, 2019 at 10:49:04AM -0400, John Ferlan wrote:
> 
> 
> On 3/20/19 5:32 PM, Eric Blake wrote:
> > Remove a now-unused parameter from qemuDomainSnapshotWriteMetadata().
> > Then, instead of calling it after every individual change to a given
> > snapshot, call it only once at the end of the API function that
> > resulted in any overall changes.
> > 
> > Signed-off-by: Eric Blake <eblake at redhat.com>
> > ---
> >  src/qemu/qemu_domain.h |  1 -
> >  src/qemu/qemu_domain.c | 13 ++---------
> >  src/qemu/qemu_driver.c | 50 ++++++++++++++----------------------------
> >  3 files changed, 19 insertions(+), 45 deletions(-)
> > 
> 
> I guess proper review of this depends on the decision from 17/16. Still
> if the decision going forward it to have in one large file, then this
> patch appears to be right to me. If there's some pushback that says no
> we have to give the consumer a decision to keep saving single snapshot
> files, then this cannot be done.
> 
> Still in reading this patch, I'm not sure we could keep a hybrid model,
> so that swings the pendulum towards let's go with the one file although
> the RPC limitation is a concern.

No matter what we do for on-disk storage, we must never expose
a bulk query via the API  or RPC.

This shouldn't be affected by the choice of storage format,
though if we never expose bulk format in the APIs, that pushes
me to preferring to mirror that in the storage & stick with
one file per snapshot.


Regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|