On Mon, Jul 08, 2013 at 08:59:45AM +0200, Valerio Pachera wrote: > Considering this vdi list > > Name Id Size Used Shared Creation time VDI id Copies Tag > s backup_data 2 1.0 TB 4.7 GB 195 GB 2013-07-04 08:52 c8d129 > 2 lug04 > s backup_data 3 1.0 TB 118 GB 102 GB 2013-07-04 09:59 c8d12a > 2 > s backup_data 4 1.0 TB 115 GB 107 GB 2013-07-05 16:42 c8d12b > 2 lug08 > backup_data 0 1.0 TB 0.0 MB 222 GB 2013-07-08 08:33 c8d12c 2 > > how do I calculate the amount of space used the snapshots? > What I mean is that keeping snapshots should have an impact on the > cluster available size. > Soon or later I'll fill up the cluster if I don't remove snapshots. > I would like to monitor that. > > I tried to delete older snapshot and check if anything was changing in > vdi info, and it didn't > > # collie node info > Id Size Used Avail Use% > 0 2.4 TB 298 GB 2.1 TB 11% > 1 2.4 TB 290 GB 2.2 TB 11% > 2 2.9 TB 354 GB 2.5 TB 11% > 3 2.3 TB 300 GB 2.0 TB 12% > Total 10 TB 1.2 TB 8.8 TB 12% > Total virtual image size 1.3 TB > > # collie vdi delete -s 2 backup_data > > # collie node info > Id Size Used Avail Use% > 0 2.4 TB 298 GB 2.1 TB 11% > 1 2.4 TB 290 GB 2.2 TB 11% > 2 2.9 TB 354 GB 2.5 TB 11% > 3 2.3 TB 300 GB 2.0 TB 12% > Total 10 TB 1.2 TB 8.8 TB 12% > Total virtual image size 1.3 TB Current algorithm only release space after all the snapshots and its clones are deleted. Kazutaka has a patch set in progress to release objects once they are not referenced by any snpashots & clone & working vdi. After this patch set get merged, deletion of snapshots will make effect of reduction of used space. Thanks Yuan |