[PATCH] Fix false alarm on "vol_destroy" about volume is masked.

[Gris Ge]

Problem:
    'vol_destroy' call will failed with error -303 indicating the
    requested volume is masked if that volume was masked to any access group.

Root cause:
    The 'destroy()' method use tpg.luns to check whether certain volume is
    still masked/exported/mapped to any initiator or access group.
    The 'access_group_map_destroy()' does not remove StorageObject and LUN
    instances which cause this false alarm.

Fix:
    Remove the StorageObject and LUN instances if that lun is not masked to
    any initiator in access_group_map_destroy() method.

Tested:
    Tested both for access group volume mapping and initiator volume mapping.

Signed-off-by: Gris Ge <fge(a)redhat.com&gt;
---
 targetd/block.py | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/targetd/block.py b/targetd/block.py
index 888e6e4..2d7d1e1 100644
--- a/targetd/block.py
+++ b/targetd/block.py
@@ -582,4 +582,11 @@ def access_group_map_destroy(req, pool_name, vol_name, ag_name):
         if map_group.tpg_lun == tpg_lun:
             map_group.delete()
 
+    if not len(list(tpg_lun.mapped_luns)):
+        # If LUN is not masked to any access group or initiator
+        # remove LUN instance.
+        lun_so = tpg_lun.storage_object
+        tpg_lun.delete()
+        lun_so.delete()
+
     RTSRoot().save_to_file()
-- 
1.8.3.1