On 12/16/2014 10:20 AM, Daniel P. Berrange wrote:
On Fri, Dec 12, 2014 at 04:14:29PM -0500, John Ferlan wrote:
<...snip...>
I'm unclear still on what the difference is between a thin snapshot (with no backing volume) and a thin volume ?
I've asked that question of someone from lvm... The thin shapshot has a 'hidden' backing pool of sorts (seen with lvs -a) # lvs LVM_Test -a ... test LVM_Test swi-a-s--- 4.00m [test_vorigin] 0.00 [test_vorigin] LVM_Test owi-a-s--- 8.00m ... When the thin snapshot is removed, the [test_vorigin] is removed as well. The target file created (/dev/LVM_Test/test) will have the 8M size as seen via the virsh vol-list --details or virsh vol-info commands. Compare that to the thin lv's which have the allocation size listed for capacity until patch 2 is added which asks the pool for it's size (e.g. the --virtualsize value).
FWIW, the original intent was that this provide a volume that is equivalent semantically to a sparse file created on a filesystem. ie the LVM equivalent to 'dd if=/dev/zero of=foo.img seek=1G count=0'
Given Jan's point/concern - I'll post a patch that adds the "--type snapshot" until more time/effort can be devoted to thin pool support. John