RBD: Fix total_capacity

Ceph has changed the meaning of the ``bytes_used`` column in the pools
reported by the ``df`` command, which means that in some deployments the
rbd driver is not reporting the expected information ot the schedulers.

The information we should be used for the calculations is returned in
the ``stored`` field in those systems.

This patch uses ``stored`` when present and fallbacks to ``bytes_used``
if not.

Closes-Bug: #1960206
Change-Id: I0ca25789a0b279d82f766091235f24f429405da6

cinder/tests/unit/volume/drivers/test_rbd.py

@@ -1759,6 +1759,51 @@ class RBDTestCase(test.TestCase):
         ])
         self.assertEqual((free_capacity, total_capacity), result)
 
+    @ddt.data(
+        # Normal case, no quota and dynamic total
+        {'free_capacity': 27.0, 'total_capacity': 28.44},
+        # No quota and static total
+        {'dynamic_total': False,
+         'free_capacity': 27.0, 'total_capacity': 59.96},
+        # Quota and dynamic total
+        {'quota_max_bytes': 3221225472, 'max_avail': 1073741824,
+         'free_capacity': 1, 'total_capacity': 2.44},
+        # Quota and static total
+        {'quota_max_bytes': 3221225472, 'max_avail': 1073741824,
+         'dynamic_total': False,
+         'free_capacity': 1, 'total_capacity': 3.00},
+        # Quota and dynamic total when free would be negative
+        {'quota_max_bytes': 1073741824,
+         'free_capacity': 0, 'total_capacity': 1.44},
+    )
+    @ddt.unpack
+    @common_mocks
+    def test_get_pool_nautilus(self, free_capacity, total_capacity,
+                               max_avail=28987613184, quota_max_bytes=0,
+                               dynamic_total=True):
+        client = self.mock_client.return_value
+        client.__enter__.return_value = client
+        client.cluster.mon_command.side_effect = [
+            (0, '{"stats":{"total_bytes":64385286144,'
+             '"total_used_bytes":3289628672,"total_avail_bytes":61095657472},'
+             '"pools":[{"name":"rbd","id":2,"stats":{"kb_used":1510197,'
+             '"stored":1546440971,"bytes_used":4639322913,"max_avail":%s,'
+             '"objects":412}},{"name":"volumes","id":3,"stats":{"kb_used":0,'
+             '"bytes_used":0,"max_avail":28987613184,"objects":0}}]}\n' %
+             max_avail, ''),
+            (0, '{"pool_name":"volumes","pool_id":4,"quota_max_objects":0,'
+             '"quota_max_bytes":%s}\n' % quota_max_bytes, ''),
+        ]
+        with mock.patch.object(self.driver.configuration, 'safe_get',
+                               return_value=dynamic_total):
+            result = self.driver._get_pool_stats()
+        client.cluster.mon_command.assert_has_calls([
+            mock.call('{"prefix":"df", "format":"json"}', b''),
+            mock.call('{"prefix":"osd pool get-quota", "pool": "rbd",'
+                      ' "format":"json"}', b''),
+        ])
+        self.assertEqual((free_capacity, total_capacity), result)
+
     @common_mocks
     def test_get_pool_bytes(self):
         """Test for mon_commands returning bytes instead of strings."""

cinder/volume/drivers/rbd.py

@@ -642,12 +642,15 @@ class RBDDriver(driver.CloneableImageVD, driver.MigrateVD,
 
         total_capacity: float
         free_capacity: float
+
+        # In Nautilus bytes_used was renamed to stored
+        bytes_used = pool_stats.get('stored', pool_stats['bytes_used'])
         quota_outbuf = encodeutils.safe_decode(quota_outbuf)
         bytes_quota = json.loads(quota_outbuf)['quota_max_bytes']
         # With quota the total is the quota limit and free is quota - used
         if bytes_quota:
             total_capacity = bytes_quota
-            free_capacity = max(min(total_capacity - pool_stats['bytes_used'],
+            free_capacity = max(min(total_capacity - bytes_used,
                                     pool_stats['max_avail']),
                                 0)
         # Without quota free is pools max available and total is global size
@@ -657,7 +660,7 @@ class RBDDriver(driver.CloneableImageVD, driver.MigrateVD,
 
         # If we want dynamic total capacity (default behavior)
         if self.configuration.safe_get('report_dynamic_total_capacity'):
-            total_capacity = free_capacity + pool_stats['bytes_used']
+            total_capacity = free_capacity + bytes_used
 
         free_capacity = round((float(free_capacity) / units.Gi), 2)
         total_capacity = round((float(total_capacity) / units.Gi), 2)

releasenotes/notes/rbd-total_capacity-60f10b45e3a8c8ea.yaml

@@ -0,0 +1,8 @@
+---
+fixes:
+  - |
+    RBD driver `bug #1960206
+    <https://bugs.launchpad.net/cinder/+bug/1960206>`_: Fixed
+    ``total_capacity`` reported by the driver to the scheduler on Ceph clusters
+    that have renamed the ``bytes_used`` field to ``stored``. (e.g., `Nautilus
+    <https://docs.ceph.com/en/nautilus/releases/nautilus/#upgrade-compatibility-notes>`_).