Bug 1034721

Description of problem:
# vgs
  VG   #PV #LV #SN Attr   VSize VFree  
  vg1    3   7   0 wz--n- 2.84g 480.00m
  vg2    1   2   0 wz--n- 1.38g 320.00m

# lvs
  LV        VG   Attr       LSize   Pool      Origin Data%
  lv1       vg1  -wi-a-----  96.00m
  lv2       vg1  -wi-a-----  96.00m
  thin1     vg1  Vwi-a-tz-- 100.00g thinpool           0.05
  thin2     vg1  Vwi-a-tz--  50.00g thinpool2          0.00
  thin22    vg1  Vwi-a-tz-- 100.00g thinpool2          0.00
  thinpool  vg1  twi-a-tz--   1.59g                    3.32
  thinpool2 vg1  twi-a-tz-- 512.00m                    0.00
  thin3     vg2  Vwi-a-tz-- 100.00g thinpool3          0.00
  thinpool3 vg2  twi-a-tz--   1.00g                    0.00

I fire up ipython -i, initialize blivet and say:

> tvg = b.devicetree.getDeviceByName('vg1-thinpool')

> tvg
(instance's string representation follows)
LVMThinPoolDevice instance (0x2dce910) --
name = vg1-thinpool ...

> tvg.freeSpace
-100768.0

As can be seen, tvg.freeSpace is determined as a simple substraction of 100 gigs of its only child 'thin1' and the thinpool's size of 1.59g.


Version-Release number of selected component (if applicable):
python-blivet-0.27-1.fc21.noarch


Steps to Reproduce:
1. See description


Actual results:
The computed free space of a thinpool is wrong.


Expected results:
The computed free space of a thinpool equals its size minus the size of the data (+ metadata?) present on all of its thin volumes.

In this case, size of the thinpool is 1.59g, that's about 1628m. Of that amount, 3.32% is occupied, that is about 54m. So free space of the thinpool should be 1628 - 54 = 1574. Or simply take 96.68% of the thinpool's size.