DataDog · hithwen · Aug 4, 2021 · Jul 14, 2021 · Jul 14, 2021 · Jul 28, 2021
@@ -37,4 +37,12 @@ files:
       value:
         type: boolean
         example: false
+    - name: ip_cache_duration
+      value:
+        display_default: null
+        example: 120
+        type: number
+      description: |
+        Time in seconds to cache the IP address of the given host. If duration is not set,
+        the cache will expire only upon error.
     - template: instances/default
@@ -16,6 +16,10 @@ def instance_empty_default_hostname(field, value):
     return False
 
 
+def instance_ip_cache_duration(field, value):
+    return get_default_field_value(field, value)
+
+
 def instance_min_collection_interval(field, value):
     return 15
 

@@ -20,6 +20,7 @@ class Config:
     collect_response_time: Optional[bool]
     empty_default_hostname: Optional[bool]
     host: str
+    ip_cache_duration: Optional[float]
     min_collection_interval: Optional[float]
     name: str
     port: int

@@ -40,6 +40,12 @@ instances:
     #
     # collect_response_time: false
 
+    ## @param ip_cache_duration - number - optional
+    ## Time in seconds to cache the IP address of the given host. If duration is not set,
+    ## the cache will expire only upon error.
+    #
+    # ip_cache_duration: 120
+
     ## @param tags - list of strings - optional
     ## A list of tags to attach to every metric and service check emitted by this instance.
     ##

@@ -13,6 +13,8 @@ class TCPCheck(AgentCheck):
 
     SOURCE_TYPE_NAME = 'system'
     SERVICE_CHECK_NAME = 'tcp.can_connect'
+    CONFIGURATION_ERROR_MSG = "`{}` is an invalid `{}`; a {} must be specified."
+    DEFAULT_IP_CACHE_DURATION = None
 
     def __init__(self, name, init_config, instances):
         super(TCPCheck, self).__init__(name, init_config, instances)
@@ -24,16 +26,27 @@ def __init__(self, name, init_config, instances):
         self.host = instance.get('host', None)
         self.socket_type = None
         self._addr = None
+        self.ip_cache_last_ts = 0
+        self.ip_cache_duration = self.DEFAULT_IP_CACHE_DURATION
+
+        ip_cache_duration = instance.get('ip_cache_duration', None)
+        if ip_cache_duration is not None:
+            try:
+                self.ip_cache_duration = int(ip_cache_duration)
+            except Exception:
+                raise ConfigurationError(
+                    self.CONFIGURATION_ERROR_MSG.format(ip_cache_duration, 'ip_cache_duration', 'number')
+                )
 
         port = instance.get('port', None)
         try:
             self.port = int(port)
         except Exception:
-            raise ConfigurationError("{} is not a correct port.".format(str(port)))
+            raise ConfigurationError(self.CONFIGURATION_ERROR_MSG.format(port, 'port', 'number'))
         try:
             split_url = self.host.split(":")
         except Exception:  # Would be raised if url is not a string
-            raise ConfigurationError("A valid url must be specified")
+            raise ConfigurationError(self.CONFIGURATION_ERROR_MSG.format(self.host, 'url', 'string'))
 
         custom_tags = instance.get('tags', [])
         self.tags = [
@@ -51,7 +64,9 @@ def __init__(self, name, init_config, instances):
         if len(split_url) == 8:  # It may then be a IP V6 address, we check that
             for block in split_url:
                 if len(block) != 4:
-                    raise ConfigurationError("{} is not a correct IPv6 address.".format(self.host))
+                    raise ConfigurationError(
+                        self.CONFIGURATION_ERROR_MSG.format(self.host, 'IPv6 address', 'valid address')
+                    )
             # It's a correct IP V6 address
             self._addr = self.host
             self.socket_type = socket.AF_INET6
@@ -74,6 +89,11 @@ def resolve_ip(self):
         self._addr = socket.gethostbyname(self.host)
         self.log.debug("%s resolved to %s", self.host, self._addr)
 
+    def should_resolve_ip(self):
+        if self.ip_cache_duration is None:
+            return False
+        return get_precise_time() - self.ip_cache_last_ts > self.ip_cache_duration
+
     def connect(self):
         with closing(socket.socket(self.socket_type)) as sock:
             sock.settimeout(self.timeout)
@@ -84,7 +104,12 @@ def connect(self):
 
     def check(self, _):
         start = get_precise_time()  # Avoid initialisation warning
-        self.log.debug("Connecting to %s %d", self.host, self.port)
+
+        if self.should_resolve_ip():
+            self.resolve_ip()
+            self.ip_cache_last_ts = start
+
+        self.log.debug("Connecting to %s on port %d", self.host, self.port)
         try:
             response_time = self.connect()
             self.log.debug("%s:%d is UP", self.host, self.port)

@@ -27,11 +27,13 @@ def test_down(aggregator):
     assert len(aggregator.service_checks('tcp.can_connect')) == 1
 
 
-def test_reattempt_resolution():
+def test_reattempt_resolution_on_error():
     instance = deepcopy(common.INSTANCE)
     check = TCPCheck(common.CHECK_NAME, {}, [instance])
     check.check(instance)
 
+    assert check.ip_cache_duration is None
+
     # IP is not normally re-resolved after the first check run
     with mock.patch.object(check, 'resolve_ip', wraps=check.resolve_ip) as resolve_ip:
         check.check(instance)
@@ -49,6 +51,24 @@ def test_reattempt_resolution():
         assert resolve_ip.called
 
 
+def test_reattempt_resolution_on_duration():
+    instance = deepcopy(common.INSTANCE)
+    instance['ip_cache_duration'] = 0
+    check = TCPCheck(common.CHECK_NAME, {}, [instance])
+    check.check(instance)
+
+    assert check.ip_cache_duration is not None
+
+    # ip_cache_duration = 0 means IP is re-resolved every check run
+    with mock.patch.object(check, 'resolve_ip', wraps=check.resolve_ip) as resolve_ip:
+        check.check(instance)
+        assert resolve_ip.called
+        check.check(instance)
+        assert resolve_ip.called
+        check.check(instance)
+        assert resolve_ip.called
+
+
 def test_up(aggregator, check):
     """
     Service expected to be up