redundancy change to the cluster default when rollback VDI

[Saeki Masaki]

When execute VDI rollback ,
redundancy may returns to a cluster default value rather than the VDI defined values.

This behavior, has a risk of disk-full, or data lost.

How to reproduce:

# sheep -v
Sheepdog daemon version 0.9.0_101_g821dc61_dirty

# dog cluster format -c 3
using backend plain store

# dog node list
  Id Host:Port V-Nodes Zone
   0 172.16.2.103:7000 128 0
   1 172.16.2.103:7001 128 1
   2 172.16.2.103:7002 128 2

# dog vdi list
  Name Id Size Used Shared Creation time VDI id Copies Tag Block Size Shift

# dog vdi create -c 2 test 10M
# dog vdi snapshot test

# dog vdi list
  Name Id Size Used Shared Creation time VDI id Copies Tag Block Size Shift
s test 1 10 MB 0.0 MB 0.0 MB 2015-03-02 16:27 7c2b25 2 22
  test 0 10 MB 0.0 MB 0.0 MB 2015-03-02 16:27 7c2b26 2 22

# dog vdi rollback test -s 1
This operation discards any changes made since the previous
snapshot was taken. Continue? [yes/no]: yes

# dog vdi list
  Name Id Size Used Shared Creation time VDI id Copies Tag Block Size Shift
s test 1 10 MB 0.0 MB 0.0 MB 2015-03-02 16:27 7c2b25 2 22
  test 0 10 MB 0.0 MB 0.0 MB 2015-03-02 16:27 7c2b27 3 22 =====>> Copies are not 2

[Saeki Masaki (saeki-masaki)]

I tested in my reproduce environment with version below.

# sheep -v
Sheepdog daemon version 0.9.0_109_gc9c955b

it works good to me