Adding optional service_check_rw parameter to check for read-only filesystem

disk/README.md

@@ -27,7 +27,9 @@ See [metadata.csv][4] for a list of metrics provided by this integration.
 The Disk check does not include any events at this time.
 
 ### Service Checks
-The Disk check does not include any service checks at this time.
+
+**`disk.service_check_rw`**:
+Returns `CRITICAL` if filesystem is in read-only mode.
 
 ## Troubleshooting
 Need help? Contact [Datadog Support][5].

disk/datadog_checks/disk/data/conf.yaml.default

@@ -55,3 +55,7 @@ instances:
     # tags:
     #   - optional_tag1
     #   - optional_tag2
+    #
+    # The (optional) service_check_rw parameter notifies when one of the partitions is in
+    # read-only mode
+    # service_check_rw: false

disk/datadog_checks/disk/disk.py

@@ -60,6 +60,8 @@ def _load_conf(self, instance):
             instance.get('all_partitions', False))
         self._device_tag_re = instance.get('device_tag_re', {})
         self._custom_tags = instance.get('tags', [])
+        self._service_check_rw = _is_affirmative(
+            instance.get('service_check_rw', False))
 
         # Force exclusion of CDROM (iso9660) from disk check
         self._excluded_filesystems.append('iso9660')
@@ -127,6 +129,22 @@ def collect_metrics_psutil(self):
                 self.gauge(metric_name, metric_value,
                            tags=tags, device_name=device_name)
 
+            # Add in a disk read write or read only check
+            if self._service_check_rw:
+                rwro = list(set(['rw', 'ro']) & set(part.opts.split(',')))
+                if len(rwro) == 1:
+                    self.service_check(
+                        'disk.read_write',
+                        AgentCheck.OK if rwro[0] == 'rw' else AgentCheck.CRITICAL,
+                        tags=tags+['device:%s' % (device_name)]
+                    )
+                else:
+                    self.service_check(
+                        'disk.read_write', AgentCheck.UNKNOWN,
+                        tags=tags+['device:%s' % (device_name)]
+                    )
+
+
         self.collect_latency_metrics()
 
     def _exclude_disk_psutil(self, part):

disk/service_checks.json

@@ -0,0 +1,11 @@
+[
+    {
+        "agent_version": "5.22.3",
+        "integration":"disk",
+        "check": "disk.service_check_rw",
+        "statuses": ["ok", "critical", "unknown"],
+        "groups": ["device"],
+        "name": "Disk read_write",
+        "description": "Returns `CRITICAL` if filesystem is in read-only mode."
+    }
+]

disk/tests/test_check.py

@@ -49,10 +49,11 @@
 
 
 class MockPart(object):
-    def __init__(self, device=DEFAULT_DEVICE_NAME, fstype='ext4', mountpoint='/'):
+    def __init__(self, device=DEFAULT_DEVICE_NAME, fstype='ext4', mountpoint='/', opts='ro'):
         self.device = device
         self.fstype = fstype
         self.mountpoint = mountpoint
+        self.opts = opts
 
 
 class MockDiskMetrics(object):
@@ -121,6 +122,7 @@ def test_default_options():
     assert check._all_partitions is False
     assert check._excluded_disk_re == re.compile('^$')
     assert check._device_tag_re == []
+    assert check._service_check_rw is False
 
 def test_disk_check(aggregator):
     """
@@ -224,6 +226,16 @@ def test_use_mount(aggregator, psutil_mocks):
 
     assert aggregator.metrics_asserted_pct == 100.0
 
+def test_psutil_rw(aggregator, psutil_mocks):
+    """
+    Check for 'ro' option in the mounts
+    """
+    instances = [{'service_check_rw': 'yes'}]
+    c = Disk('disk', None, {}, instances)
+    c.check(instances[0])
+
+    aggregator.assert_service_check('disk.read_write', status=Disk.CRITICAL)
+
 def mock_df_output(fname):
     """
     Load fixtures from tests/fixtures/ folder and return a tuple matching the