Add support for sending database_instance metadata

mysql/CHANGELOG.md

@@ -2,6 +2,10 @@
 
 ## Unreleased
 
+***Added***:
+
+* Add support for sending `database_instance` metadata ([#15524](https://github.com/DataDog/integrations-core/pull/15524))
+
 ***Fixed***:
 
 * Bump cryptography to 41.0.3 ([#15517](https://github.com/DataDog/integrations-core/pull/15517))

mysql/assets/configuration/spec.yaml

@@ -632,6 +632,16 @@ files:
               type: boolean
               example: false
               display_default: false
+          - name: database_instance_collection_interval
+            hidden: true
+            description: |
+              Set the database instance collection interval (in seconds). The database instance collection sends
+              basic information about the database instance along with a signal that it still exists. 
+              This collection does not involve any additional queries to the database.
+            value:
+              type: number
+              example: 1800
+              display_default: false
           - template: instances/default
             overrides:
               disable_generic_tags.hidden: false

mysql/datadog_checks/mysql/config.py

@@ -73,6 +73,7 @@ def __init__(self, instance):
         }
         self.log_unobfuscated_queries = is_affirmative(instance.get('log_unobfuscated_queries', False))
         self.log_unobfuscated_plans = is_affirmative(instance.get('log_unobfuscated_plans', False))
+        self.database_instance_collection_interval = instance.get('database_instance_collection_interval', 1800)
         self.configuration_checks()
 
     def _build_tags(self, custom_tags):

mysql/datadog_checks/mysql/config_models/defaults.py

@@ -12,6 +12,10 @@ def instance_connect_timeout():
     return 10
 
 
+def instance_database_instance_collection_interval():
+    return False
+
+
 def instance_dbm():
     return False
 

mysql/datadog_checks/mysql/config_models/instance.py

@@ -169,6 +169,7 @@ class InstanceConfig(BaseModel):
     collect_settings: Optional[CollectSettings] = None
     connect_timeout: Optional[float] = None
     custom_queries: Optional[tuple[CustomQuery, ...]] = None
+    database_instance_collection_interval: Optional[float] = None
     dbm: Optional[bool] = None
     defaults_file: Optional[str] = None
     disable_generic_tags: Optional[bool] = None

mysql/datadog_checks/mysql/mysql.py

@@ -5,18 +5,27 @@
 from __future__ import division
 
 import copy
+import time
 import traceback
 from collections import defaultdict
 from contextlib import closing, contextmanager
 from typing import Any, Dict, List, Optional  # noqa: F401
 
 import pymysql
+from cachetools import TTLCache
 from six import PY3, iteritems, itervalues
 
 from datadog_checks.base import AgentCheck, is_affirmative
 from datadog_checks.base.utils.db import QueryExecutor, QueryManager
-from datadog_checks.base.utils.db.utils import resolve_db_host as agent_host_resolver
+from datadog_checks.base.utils.db.utils import (
+    default_json_event_encoding,
+)
+from datadog_checks.base.utils.db.utils import (
+    resolve_db_host as agent_host_resolver,
+)
+from datadog_checks.base.utils.serialization import json
 
+from .__about__ import __version__
 from .activity import MySQLActivity
 from .collection_utils import collect_all_scalars, collect_scalar, collect_string, collect_type
 from .config import MySQLConfig
@@ -119,8 +128,16 @@ def __init__(self, name, init_config, instances):
         self._statement_samples = MySQLStatementSamples(self, self._config, self._get_connection_args())
         self._mysql_metadata = MySQLMetadata(self, self._config, self._get_connection_args())
         self._query_activity = MySQLActivity(self, self._config, self._get_connection_args())
+        # _database_instance_emitted: limit the collection and transmission of the database instance metadata
+        self._database_instance_emitted = TTLCache(
+            maxsize=1,
+            ttl=self._config.database_instance_collection_interval,
+        )  # type: TTLCache
 
         self._runtime_queries = None
+        # Keep a copy of the tags without the internal resource tags so they can be used for paths that don't
+        # go through the agent internal metrics submission processing those tags
+        self._non_internal_tags = copy.deepcopy(self.tags)
         self.set_resource_tags()
 
     def execute_query_raw(self, query):
@@ -254,6 +271,7 @@ def check(self, _):
                 # version collection
                 self.version = get_version(db)
                 self._send_metadata()
+                self._send_database_instance_metadata()
 
                 self.is_mariadb = self.version.flavor == "MariaDB"
                 if self._get_is_aurora(db):
@@ -1213,3 +1231,24 @@ def _report_warnings(self):
 
         for warning in messages:
             self.warning(warning)
+
+    def _send_database_instance_metadata(self):
+        if self.resolved_hostname not in self._database_instance_emitted:
+            event = {
+                "host": self.resolved_hostname,
+                "agent_version": datadog_agent.get_version(),
+                "dbms": "mysql",
+                "kind": "database_instance",
+                "collection_interval": self._config.database_instance_collection_interval,
+                'dbms_version': self.version.version + '+' + self.version.build,
+                'integration_version': __version__,
+                "tags": self._non_internal_tags,
+                "timestamp": time.time() * 1000,
+                "cloud_metadata": self._config.cloud_metadata,
+                "metadata": {
+                    "dbm": self._config.dbm_enabled,
+                    "connection_host": self._config.host,
+                },
+            }
+            self._database_instance_emitted[self.resolved_hostname] = event
+            self.database_monitoring_metadata(json.dumps(event, default=default_json_event_encoding))

mysql/tests/test_metadata.py

@@ -25,8 +25,8 @@ def test_collect_mysql_settings(aggregator, dbm_instance, dd_run_check):
     mysql_check = MySql(common.CHECK_NAME, {}, instances=[dbm_instance])
     dd_run_check(mysql_check)
     dbm_metadata = aggregator.get_event_platform_events("dbm-metadata")
-    event = dbm_metadata[0]
+    event = next((e for e in dbm_metadata if e['kind'] == 'mysql_variables'), None)
+    assert event is not None
     assert event['host'] == "stubbed.hostname"
     assert event['dbms'] == "mysql"
-    assert event['kind'] == "mysql_variables"
     assert len(event["metadata"]) > 0

mysql/tests/test_mysql.py

@@ -12,6 +12,7 @@
 from datadog_checks.base.utils.platform import Platform
 from datadog_checks.dev.utils import get_metadata_metrics
 from datadog_checks.mysql import MySql
+from datadog_checks.mysql.__about__ import __version__
 from datadog_checks.mysql.const import (
     BINLOG_VARS,
     GALERA_VARS,
@@ -640,3 +641,44 @@ def test_set_resources(aggregator, dd_run_check, instance_basic, cloud_metadata,
     aggregator.assert_metric_has_tag(
         "mysql.net.connections", tags.DATABASE_INSTANCE_RESOURCE_TAG.format(hostname=mysql_check.resolved_hostname)
     )
+
+
+@pytest.mark.parametrize(
+    'dbm_enabled, reported_hostname',
+    [
+        (True, None),
+        (False, None),
+        (True, 'forced_hostname'),
+        (True, 'forced_hostname'),
+    ],
+)
+@pytest.mark.integration
+@pytest.mark.usefixtures('dd_environment')
+def test_database_instance_metadata(aggregator, dd_run_check, instance_complex, dbm_enabled, reported_hostname):
+    instance_complex['dbm'] = dbm_enabled
+    if reported_hostname:
+        instance_complex['reported_hostname'] = reported_hostname
+    expected_host = reported_hostname if reported_hostname else 'stubbed.hostname'
+    mysql_check = MySql(common.CHECK_NAME, {}, [instance_complex])
+    dd_run_check(mysql_check)
+
+    dbm_metadata = aggregator.get_event_platform_events("dbm-metadata")
+    event = next((e for e in dbm_metadata if e['kind'] == 'database_instance'), None)
+    assert event is not None
+    assert event['host'] == expected_host
+    assert event['dbms'] == "mysql"
+    assert event['tags'].sort() == tags.METRIC_TAGS.sort()
+    assert event['integration_version'] == __version__
+    assert event['collection_interval'] == 1800
+    assert event['metadata'] == {
+        'dbm': dbm_enabled,
+        'connection_host': instance_complex['host'],
+    }
+
+    # Run a second time and expect the metadata to not be emitted again because of the cache TTL
+    aggregator.reset()
+    dd_run_check(mysql_check)
+
+    dbm_metadata = aggregator.get_event_platform_events("dbm-metadata")
+    event = next((e for e in dbm_metadata if e['kind'] == 'database_instance'), None)
+    assert event is None