[NDM] Add NDM metadata support for Cisco ACI (DataDog#17735)

* Add support for sending device metadata

* Add unit test for device metadata, update fixture

* Add license header, changelogs

* Lint

* First pass at submitting interface metadata, cleanup for test fixtures

* Fix for py2.7 support

* Try to fix imports

* Deal with pydantic stuff py2.7

* Allow namespace for Cisco ACI devices, static var for vendor

* Update device metadata to use the correct fieldname, add pydantic model for EvP intake

* Sync the conf.yaml example

* Add device type and integration to device metadata, fix ID field name

* Update interface statuses

* Deal with device status (use fabricSt)

* Update get_eth_list to get operStatus, update all tests and fixtures

* Amend docs for namespace

* Batch events sent to EvP

* Add interface status metric

* Only add to list for >py3.0

* Update default value for vendor, yield for batch events, use device type other

* Add source field to device metadata tags

* Add enums for interface status

* Use correct track type for NDM metadata

* Amend device id tag, collect timestamp ms -> s

* Add interface integration field

* More generic method to send EvP event

* Add docstring for the EvP method

* Update interface tagging, remove system_ip tag

* Fix linting for submit event platform event

* Use interface ID tags

Author: zoedt

cisco_aci/assets/configuration/spec.yaml

@@ -92,6 +92,13 @@ files:
       value:
         type: boolean
         example: False
+    - name: namespace
+      description: |
+        Namespace for differentiating between devices that share the same IP.
+        If not specified, the namespace will be 'default'.
+      value:
+        type: string
+        example: default
     - template: instances/http
       overrides:
         username.display_priority: 9

cisco_aci/changelog.d/17735.added

@@ -0,0 +1 @@
+[NDM] Add NDM metadata support for Cisco ACI

cisco_aci/datadog_checks/cisco_aci/__init__.py

@@ -3,6 +3,6 @@
 # Licensed under a 3-clause BSD style license (see LICENSE)
 
 from .__about__ import __version__
-from .cisco import CiscoACICheck
+from datadog_checks.cisco_aci.cisco import CiscoACICheck
 
 __all__ = ['__version__', 'CiscoACICheck']

cisco_aci/datadog_checks/cisco_aci/api.py

@@ -286,8 +286,8 @@ def get_spine_proc_metrics(self, pod, node):
         return self._parse_response(response)
 
     def get_eth_list(self, pod, node):
-        query = 'query-target=subtree&target-subtree-class=l1PhysIf'
-        path = '/api/mo/topology/pod-{}/node-{}/sys.json?{}'.format(pod, node, query)
+        query = 'rsp-subtree=children&rsp-subtree-class=ethpmPhysIf'
+        path = '/api/node/class/topology/pod-{}/node-{}/l1PhysIf.json?{}'.format(pod, node, query)
         response = self.make_request(path)
         return self._parse_response(response)
 

cisco_aci/datadog_checks/cisco_aci/cisco.py

@@ -6,13 +6,12 @@
 from datadog_checks.base import AgentCheck, ConfigurationError
 from datadog_checks.base.config import _is_affirmative
 from datadog_checks.base.utils.containers import hash_mutable
-
-from . import aci_metrics
-from .api import Api
-from .capacity import Capacity
-from .fabric import Fabric
-from .tags import CiscoTags
-from .tenant import Tenant
+from datadog_checks.cisco_aci.aci_metrics import make_tenant_metrics
+from datadog_checks.cisco_aci.api import Api
+from datadog_checks.cisco_aci.capacity import Capacity
+from datadog_checks.cisco_aci.fabric import Fabric
+from datadog_checks.cisco_aci.tags import CiscoTags
+from datadog_checks.cisco_aci.tenant import Tenant
 
 SOURCE_TYPE = 'cisco_aci'
 
@@ -25,7 +24,7 @@ class CiscoACICheck(AgentCheck):
 
     def __init__(self, name, init_config, instances):
         super(CiscoACICheck, self).__init__(name, init_config, instances)
-        self.tenant_metrics = aci_metrics.make_tenant_metrics()
+        self.tenant_metrics = make_tenant_metrics()
         self.last_events_ts = {}
         self.external_host_tags = {}
         self._api_cache = {}
@@ -109,7 +108,7 @@ def check(self, _):
             raise
 
         try:
-            fabric = Fabric(self, api, self.instance)
+            fabric = Fabric(self, api, self.instance, self.instance.get('namespace', 'default'))
             fabric.collect()
         except Exception as e:
             self.log.error('fabric collection failed: %s', e)

cisco_aci/datadog_checks/cisco_aci/config_models/defaults.py

@@ -60,6 +60,10 @@ def instance_min_collection_interval():
     return 15
 
 
+def instance_namespace():
+    return 'default'
+
+
 def instance_persist_connections():
     return False
 

cisco_aci/datadog_checks/cisco_aci/config_models/instance.py

@@ -82,6 +82,7 @@ class InstanceConfig(BaseModel):
     log_requests: Optional[bool] = None
     metric_patterns: Optional[MetricPatterns] = None
     min_collection_interval: Optional[float] = None
+    namespace: Optional[str] = None
     ntlm_domain: Optional[str] = None
     password: Optional[str] = None
     persist_connections: Optional[bool] = None

cisco_aci/datadog_checks/cisco_aci/data/conf.yaml.example

@@ -124,6 +124,12 @@ instances:
     #
     # appcenter: false
 
+    ## @param namespace - string - optional - default: default
+    ## Namespace for differentiating between devices that share the same IP.
+    ## If not specified, the namespace will be 'default'.
+    #
+    # namespace: default
+
     ## @param proxy - mapping - optional
     ## This overrides the `proxy` setting in `init_config`.
     ##

cisco_aci/datadog_checks/cisco_aci/fabric.py

@@ -2,21 +2,38 @@
 # All rights reserved
 # Licensed under a 3-clause BSD style license (see LICENSE)
 
-from six import iteritems
+from six import PY3, iteritems
+
+from datadog_checks.base.utils.serialization import json
+
+if PY3:
+    import time
+
+    from datadog_checks.cisco_aci.models import DeviceMetadata, InterfaceMetadata, NetworkDevicesMetadata, Node, PhysIf
+
+else:
+    DeviceMetadata = None
+    Eth = None
+    InterfaceMetadata = None
+    Node = None
 
 from . import aci_metrics, exceptions, helpers
 
+VENDOR_CISCO = 'cisco'
+PAYLOAD_METADATA_BATCH_SIZE = 100
+
 
 class Fabric:
     """
     Collect fabric metrics from the APIC
     """
 
-    def __init__(self, check, api, instance):
+    def __init__(self, check, api, instance, namespace):
         self.check = check
         self.api = api
         self.instance = instance
         self.check_tags = check.check_tags
+        self.namespace = namespace
 
         # grab some functions from the check
         self.gauge = check.gauge
@@ -25,13 +42,19 @@ def __init__(self, check, api, instance):
         self.submit_metrics = check.submit_metrics
         self.tagger = self.check.tagger
         self.external_host_tags = self.check.external_host_tags
+        self.event_platform_event = check.event_platform_event
 
     def collect(self):
         fabric_pods = self.api.get_fabric_pods()
         fabric_nodes = self.api.get_fabric_nodes()
         self.log.info("%s pods and %s nodes computed", len(fabric_nodes), len(fabric_pods))
         pods = self.submit_pod_health(fabric_pods)
-        self.submit_nodes_health(fabric_nodes, pods)
+        devices, interfaces = self.submit_nodes_health_and_metadata(fabric_nodes, pods)
+        if PY3:
+            collect_timestamp = int(time.time())
+            batches = self.batch_payloads(devices, interfaces, collect_timestamp)
+            for batch in batches:
+                self.event_platform_event(json.dumps(batch.model_dump(exclude_none=True)), "network-devices-metadata")
 
     def submit_pod_health(self, pods):
         pods_dict = {}
@@ -53,7 +76,9 @@ def submit_pod_health(self, pods):
 
         return pods_dict
 
-    def submit_nodes_health(self, nodes, pods):
+    def submit_nodes_health_and_metadata(self, nodes, pods):
+        device_metadata = []
+        interface_metadata = []
         for n in nodes:
             hostname = helpers.get_fabric_hostname(n)
 
@@ -70,17 +95,22 @@ def submit_nodes_health(self, nodes, pods):
                 continue
             self.log.info("processing node %s on pod %s", node_id, pod_id)
             try:
+                if PY3:
+                    device_metadata.append(self.submit_node_metadata(node_attrs, tags))
                 self.submit_process_metric(n, tags + self.check_tags + user_tags, hostname=hostname)
             except (exceptions.APIConnectionException, exceptions.APIParsingException):
                 pass
             if node_attrs.get('role') != "controller":
                 try:
                     stats = self.api.get_node_stats(pod_id, node_id)
                     self.submit_fabric_metric(stats, tags, 'fabricNode', hostname=hostname)
-                    self.process_eth(node_attrs)
+                    eth_metadata = self.process_eth(node_attrs)
+                    if PY3:
+                        interface_metadata.extend(eth_metadata)
                 except (exceptions.APIConnectionException, exceptions.APIParsingException):
                     pass
             self.log.info("finished processing node %s", node_id)
+        return device_metadata, interface_metadata
 
     def process_eth(self, node):
         self.log.info("processing ethernet ports for %s", node.get('id'))
@@ -90,16 +120,20 @@ def process_eth(self, node):
             eth_list = self.api.get_eth_list(pod_id, node['id'])
         except (exceptions.APIConnectionException, exceptions.APIParsingException):
             pass
+        interfaces = []
         for e in eth_list:
             eth_attrs = helpers.get_attributes(e)
             eth_id = eth_attrs['id']
             tags = self.tagger.get_fabric_tags(e, 'l1PhysIf')
+            if PY3:
+                interfaces.append(self.create_interface_metadata(e, node['address'], tags, hostname))
             try:
                 stats = self.api.get_eth_stats(pod_id, node['id'], eth_id)
                 self.submit_fabric_metric(stats, tags, 'l1PhysIf', hostname=hostname)
             except (exceptions.APIConnectionException, exceptions.APIParsingException):
                 pass
         self.log.info("finished processing ethernet ports for %s", node['id'])
+        return interfaces
 
     def submit_fabric_metric(self, stats, tags, obj_type, hostname=None):
         for s in stats:
@@ -209,3 +243,70 @@ def get_fabric_type(self, obj_type):
             return 'pod'
         if obj_type == 'l1PhysIf':
             return 'port'
+
+    def batch_payloads(self, devices, interfaces, collect_ts):
+        for device in devices:
+            yield NetworkDevicesMetadata(namespace=self.namespace, devices=[device], collect_timestamp=collect_ts)
+
+        payloads = []
+        for interface in interfaces:
+            if len(payloads) == PAYLOAD_METADATA_BATCH_SIZE:
+                yield NetworkDevicesMetadata(
+                    namespace=self.namespace, interfaces=payloads, collect_timestamp=collect_ts
+                )
+                payloads = []
+            payloads.append(interface)
+        if payloads:
+            yield NetworkDevicesMetadata(namespace=self.namespace, interfaces=payloads, collect_timestamp=collect_ts)
+
+    def submit_node_metadata(self, node_attrs, tags):
+        node = Node(attributes=node_attrs)
+        id_tags = ['namespace:{}'.format(self.namespace)]
+        device_tags = [
+            'device_vendor:{}'.format(VENDOR_CISCO),
+            'device_namespace:{}'.format(self.namespace),
+            'device_hostname:{}'.format(node.attributes.dn),
+            'hostname:{}'.format(node.attributes.dn),
+            'device_ip:{}'.format(node.attributes.address),
+            'device_id:{}:{}'.format(self.namespace, node.attributes.address),
+            "source:cisco-aci",
+        ]
+        device = DeviceMetadata(
+            id='{}:{}'.format(self.namespace, node.attributes.address),
+            id_tags=id_tags,
+            tags=device_tags + tags,
+            name=node.attributes.dn,
+            ip_address=node.attributes.address,
+            model=node.attributes.model,
+            fabric_st=node.attributes.fabric_st,
+            vendor=VENDOR_CISCO,
+            version=node.attributes.version,
+            serial_number=node.attributes.serial,
+            device_type=node.attributes.device_type,
+        )
+        return device.model_dump(exclude_none=True)
+
+    def create_interface_metadata(self, phys_if, address, tags, hostname):
+        eth = PhysIf(**phys_if.get('l1PhysIf', {}))
+        interface = InterfaceMetadata(
+            device_id='{}:{}'.format(self.namespace, address),
+            id_tags=['interface:{}'.format(eth.attributes.name)],
+            index=eth.attributes.id,
+            name=eth.attributes.name,
+            description=eth.attributes.desc,
+            mac_address=eth.attributes.router_mac,
+            admin_status=eth.attributes.admin_st,
+        )
+        if eth.ethpm_phys_if:
+            interface.oper_status = eth.ethpm_phys_if.attributes.oper_st
+        if interface.status:
+            new_tags = tags.copy()
+            new_tags.extend(
+                [
+                    "device_ip:{}".format(address),
+                    "device_namespace:{}".format(self.namespace),
+                    "interface.status:{}".format(interface.status),
+                ]
+            )
+            self.gauge('cisco_aci.fabric.node.interface.status', 1, tags=new_tags, hostname=hostname)
+        return interface.model_dump(exclude_none=True)