OK. I read your "rbd du" results as saying that the clone image "e01f31e94a65cf7e786972b915e07364-1" wrote appoximately 10GB of data, then snapshot "d-1" was created (so the space is associated w/ the snapshot), before another 616MB was written against the "HEAD" revision of the image. If you delete the snapshot "d-1", that 10GB would be associated with the "HEAD" revision (minus any overlap). On Mon, Oct 9, 2017 at 5:44 PM, Josy <josy@xxxxxxxxxxxxxxxxxxxxx> wrote: > Sorry for the confusion. > >>> e01f31e94a65cf7e786972b915e07364-1 is a clone image created from the >>> parent >>> "ostemplates/windows-std-2k8r2-x64-20171004@snap_windows-std-2k8r2-x64-20171004" > > [cephuser@ceph-las-admin-a1 ceph-cluster]$ rbd info > cvm/e01f31e94a65cf7e786972b915e07364-1 > rbd image 'e01f31e94a65cf7e786972b915e07364-1': > size 61440 MB in 15360 objects > order 22 (4096 kB objects) > block_name_prefix: rbd_data.a1b83d1b58ba > format: 2 > features: layering > flags: > create_timestamp: Fri Oct 6 19:07:12 2017 > parent: > ostemplates/windows-std-2k8r2-x64-20171004@snap_windows-std-2k8r2-x64-20171004 > >>> that image named as d-1 is a snapshot of >>> e01f31e94a65cf7e786972b915e07364-1 > > [cephuser@ceph-las-admin-a1 ceph-cluster]$ rbd snap ls > cvm/e01f31e94a65cf7e786972b915e07364-1 > SNAPID NAME SIZE TIMESTAMP > 40 d-1 61440 MB Sun Oct 8 19:07:14 2017 > > > > On 10-10-2017 02:42, Jason Dillaman wrote: >> >> If the clone has written 10GB of data, yes, the clone should show >> 10GB. I am not sure what you are referring to when you say "clone" >> since you only included a single image in your response. The clone is >> the image chained from a parent image snapshot. Is >> "e01f31e94a65cf7e786972b915e07364-1@d-1" the clone's parent? >> >> On Mon, Oct 9, 2017 at 5:14 PM, Josy <josy@xxxxxxxxxxxxxxxxxxxxx> wrote: >>> >>> Thank you for your response! >>> >>> If the cloned VM had written around 10Gbs of data, wouldn't the clone >>> also >>> show that much space? >>> >>> Below is a list of the original image, the clone and new snapshots along >>> with their sizes. >>> >>> The clone is still only a few hundred megabytes, while the snapshot shows >>> "10GB" of disk space. >>> >>> [cephuser@ceph-las-admin-a1 ceph-cluster]$ rbd du >>> cvm/e01f31e94a65cf7e786972b915e07364-1 >>> warning: fast-diff map is not enabled for >>> e01f31e94a65cf7e786972b915e07364-1. operation may be slow. >>> NAME PROVISIONED USED >>> e01f31e94a65cf7e786972b915e07364-1@d-1 61440M 10840M >>> e01f31e94a65cf7e786972b915e07364-1 61440M 616M >>> <TOTAL> 61440M 11456M >>> >>> >>> >>> On 09-10-2017 17:22, Jason Dillaman wrote: >>>> >>>> No -- it means that your clone had written approximately 10GB of space >>>> within the image before you created the first snapshot. If the >>>> "fast-diff" feature is enabled, note that it only calculates usage in >>>> object size chunks (defaults to 4MB) -- which means that even writing >>>> 1 byte to a 4MB object would flag the object as dirty (it's an >>>> upper-bound estimate). >>>> >>>> On Sun, Oct 8, 2017 at 12:01 PM, Josy <josy@xxxxxxxxxxxxxxxxxxxxx> >>>> wrote: >>>>> >>>>> Hello, >>>>> >>>>> I noticed that when we create a snapshot of a clone, the first snapshot >>>>> seems to be quite large. >>>>> >>>>> For example: >>>>> >>>>> Clone VM is taking up 136MBs according to rbd du >>>>> First snapshot: 10GBs >>>>> Second snapshot: 104MBs >>>>> Third snapshot: 57MBs >>>>> >>>>> The clone is a Windows virtual machine, which does take around 10GB if >>>>> it >>>>> wasn't a clone. Does this mean that the first snapshot of a clone needs >>>>> to >>>>> have a full copy of the original cloned image? >>>>> >>>>> Ceph version : Luminous 12.2.1 >>>>> >>>>> >>>>> _______________________________________________ >>>>> ceph-users mailing list >>>>> ceph-users@xxxxxxxxxxxxxx >>>>> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com >>>> >>>> >>>> >>> >> >> > > -- Jason _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |