On Fri, Aug 09, 2013 at 05:30:35PM +0200, Valerio Pachera wrote: > 2013/8/9 MORITA Kazutaka <morita.kazutaka at gmail.com>: > > I pushed the patch to the master. It's available now. > > I notice now that the patch has been written only 6 hours ago! ;-) > > > Maybe, you misunderstood something I wrote. Basically, it's difficult > > for block device storage to know that which blocks are not used by > > guest file system. > > I found the discussion, but it was with Yuan. > The object was 'vdi and dynamic allocation' > > 2012/12/14 Liu Yuan <namei.unix at gmail.com>: > > This is similar to linux file punch holes to safe disk space. > > Okay, "optimize" is indeed a better wording. let's firstly implement a > > mechanism to optimize the underlying storage allocation via 'collie vdi > > optimize'. This feature is in the plan. > > I don't consider this an important future but I was curious to know if > was already present. Well, with discard/trim support, all-zero objects will be automatically removed from cluster. So this 'optimize' command will duplicate the function and not necessary. There is still another way to discard all-zero objects for now with qemu-img if you don't enable discard support within guest yet. 1 qemu-img convert sheepdog:image local_image # this will auto-trim zeroes for local_image 2 qemu-img convert local_image sheepdog:image # Now zero objects are all trimmed Thanks Yuan |