Let the possibility to disable instance metadata collection

Fix #975

agent.py

@@ -182,7 +182,7 @@ def _set_agent_config_hostname(self, agentConfig):
         # in the config file.
         # DEPRECATED
         if agentConfig.get('hostname') is None and agentConfig.get('use_ec2_instance_id'):
-            instanceId = EC2.get_instance_id()
+            instanceId = EC2.get_instance_id(agentConfig)
             if instanceId is not None:
                 log.info("Running on EC2, instanceId: %s" % instanceId)
                 agentConfig['hostname'] = instanceId

checks/collector.py

@@ -422,12 +422,12 @@ def _build_payload(self, start_event=True):
                 host_tags.extend([unicode(tag.strip()) for tag in self.agentConfig['tags'].split(",")])
 
             if self.agentConfig['collect_ec2_tags']:
-                host_tags.extend(EC2.get_tags())
+                host_tags.extend(EC2.get_tags(self.agentConfig))
 
             if host_tags:
                 payload['host-tags']['system'] = host_tags
 
-            GCE_tags = GCE.get_tags()
+            GCE_tags = GCE.get_tags(self.agentConfig)
             if GCE_tags is not None:
                 payload['host-tags'][GCE.SOURCE_TYPE_NAME] = GCE_tags
 
@@ -438,7 +438,7 @@ def _build_payload(self, start_event=True):
         return payload
 
     def _get_metadata(self):
-        metadata = EC2.get_metadata()
+        metadata = EC2.get_metadata(self.agentConfig)
         if metadata.get('hostname'):
             metadata['ec2-hostname'] = metadata.get('hostname')
             del metadata['hostname']

config.py

@@ -400,6 +400,10 @@ def get_config(parse_args=True, cfg_path=None, options=None):
         if config.has_option("Main", "skip_ssl_validation"):
             agentConfig["skip_ssl_validation"] = _is_affirmative(config.get("Main", "skip_ssl_validation"))
 
+        agentConfig["collect_instance_metadata"] = True
+        if config.has_option("Main", "collect_instance_metadata"):
+            agentConfig["collect_instance_metadata"] = _is_affirmative(config.get("Main", "collect_instance_metadata"))
+
         agentConfig["collect_ec2_tags"] = False
         if config.has_option("Main", "collect_ec2_tags"):
             agentConfig["collect_ec2_tags"] = _is_affirmative(config.get("Main", "collect_ec2_tags"))

datadog.conf.example

@@ -26,6 +26,14 @@ api_key:
 # Collect AWS EC2 custom tags as agent tags
 # collect_ec2_tags: no
 
+# Collect instance metadata
+# The Agent will try to collect instance metadata for EC2 and GCE instances by 
+# trying to connect to the local endpoint: http://169.254.169.254
+# See http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/AESDG-chapter-instancedata.html
+# and https://developers.google.com/compute/docs/metadata
+# for more information
+# collect_instance_metadata: yes
+
 # Set the threshold for accepting points to allow anything
 # with recent_point_threshold seconds
 # Defaults to 30 seconds if no value is provided

packaging/datadog-agent/win32/install_files/datadog_win32.conf

@@ -32,6 +32,14 @@ api_key: APIKEYHERE
 # Certificate file.
 # ca_certs = datadog-cert.pem
 
+# Collect instance metadata
+# The Agent will try to collect instance metadata for EC2 and GCE instances by 
+# trying to connect to the local endpoint: http://169.254.169.254
+# See http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/AESDG-chapter-instancedata.html
+# and https://developers.google.com/compute/docs/metadata
+# for more information
+# collect_instance_metadata: yes
+
 # ========================================================================== #
 # Information page configuration
 # ========================================================================== #

tests/test_ec2.py

@@ -12,7 +12,7 @@ def test_metadata(self):
         if os.environ.get('TRAVIS', False): return
         # Test gathering metadata from ec2
         start = time.time()
-        d = EC2.get_metadata()
+        d = EC2.get_metadata({'collect_instance_metadata': True})
         end = time.time()
         assert type(d) == types.DictType
         # Either we're on ec2 or we're not (at least 7 attributes expected)

util.py

@@ -198,7 +198,7 @@ def get_hostname(config=None):
 
     #Try to get GCE instance name
     if hostname is None:
-        gce_hostname = GCE.get_hostname()
+        gce_hostname = GCE.get_hostname(config)
         if gce_hostname is not None:
             if is_valid_hostname(gce_hostname):
                 return gce_hostname
@@ -222,7 +222,7 @@ def _get_hostname_unix():
 
     # if we have an ec2 default hostname, see if there's an instance-id available
     if hostname is not None and True in [hostname.lower().startswith(p) for p in [u'ip-', u'domu']]:
-        instanceid = EC2.get_instance_id()
+        instanceid = EC2.get_instance_id(config)
         if instanceid:
             hostname = instanceid
 
@@ -249,10 +249,14 @@ class GCE(object):
 
 
     @staticmethod
-    def _get_metadata():
+    def _get_metadata(agentConfig):
         if GCE.metadata is not None:
             return GCE.metadata
 
+        if not agentConfig['collect_instance_metadata']:
+            GCE.metadata = {}
+            return GCE.metadata
+
         socket_to = None
         try:
             socket_to = socket.getdefaulttimeout()
@@ -279,10 +283,12 @@ def _get_metadata():
 
 
     @staticmethod
-    def get_tags():
+    def get_tags(agentConfig):
+        if not agentConfig['collect_instance_metadata']:
+            return []
 
         try:
-            host_metadata = GCE._get_metadata()
+            host_metadata = GCE._get_metadata(agentConfig)
             tags = []
 
             for key, value in host_metadata['instance'].get('attributes', {}).iteritems():
@@ -303,9 +309,9 @@ def get_tags():
             return None
 
     @staticmethod
-    def get_hostname():
+    def get_hostname(agentConfig):
         try:
-            host_metadata = GCE._get_metadata()
+            host_metadata = GCE._get_metadata(agentConfig)
             return host_metadata['instance']['hostname'].split('.')[0]
         except Exception:
             return None
@@ -320,7 +326,10 @@ class EC2(object):
     metadata = {}
 
     @staticmethod
-    def get_tags():
+    def get_tags(agentConfig):
+        if not agentConfig['collect_instance_metadata']:
+            return []
+
         socket_to = None
         try:
             socket_to = socket.getdefaulttimeout()
@@ -352,7 +361,7 @@ def get_tags():
 
 
     @staticmethod
-    def get_metadata():
+    def get_metadata(agentConfig):
         """Use the ec2 http service to introspect the instance. This adds latency if not running on EC2
         """
         # >>> import urllib2
@@ -366,6 +375,10 @@ def get_metadata():
         # Every call may add TIMEOUT seconds in latency so don't abuse this call
         # python 2.4 does not support an explicit timeout argument so force it here
         # Rather than monkey-patching urllib2, just lower the timeout globally for these calls
+
+        if not agentConfig['collect_instance_metadata']:
+            return {}
+
         socket_to = None
         try:
             socket_to = socket.getdefaulttimeout()
@@ -391,9 +404,9 @@ def get_metadata():
         return EC2.metadata
 
     @staticmethod
-    def get_instance_id():
+    def get_instance_id(agentConfig):
         try:
-            return EC2.get_metadata().get("instance-id", None)
+            return EC2.get_metadata(agentConfig).get("instance-id", None)
         except Exception:
             return None