From 354a7146d719146f8d9a2cccf530c20698b36b9e Mon Sep 17 00:00:00 2001
From: Eric Weaver <eweaver755@gmail.com>
Date: Tue, 2 Sep 2025 10:43:13 -0400
Subject: [PATCH 1/2] [DBMON-5587] Always clear out state between schema
 collections (#21198)

* Always clear out state between schema collections

* run format

* add changelog
---
 mysql/changelog.d/21198.fixed                |  1 +
 mysql/datadog_checks/mysql/databases_data.py | 36 +++++++++++---------
 mysql/datadog_checks/mysql/metadata.py       |  2 +-
 3 files changed, 22 insertions(+), 17 deletions(-)
 create mode 100644 mysql/changelog.d/21198.fixed

diff --git a/mysql/changelog.d/21198.fixed b/mysql/changelog.d/21198.fixed
new file mode 100644
index 0000000000000..e9f0d6342a66f
--- /dev/null
+++ b/mysql/changelog.d/21198.fixed
@@ -0,0 +1 @@
+Properly cleanup schema metadata memory references after all collection cycles
diff --git a/mysql/datadog_checks/mysql/databases_data.py b/mysql/datadog_checks/mysql/databases_data.py
index 2890725e93e9b..1adc026422154 100644
--- a/mysql/datadog_checks/mysql/databases_data.py
+++ b/mysql/datadog_checks/mysql/databases_data.py
@@ -54,6 +54,7 @@ def reset(self):
         self._total_columns_sent = 0
         self._columns_count = 0
         self.db_info.clear()
+        self.db_to_tables.clear()
         self.any_tables_found = False
 
     def store_db_infos(self, db_infos):
@@ -183,7 +184,7 @@ def _fetch_database_data(self, cursor, start_time, db_name):
         self._data_submitter.submit()
 
     @tracked_method(agent_check_getter=agent_check_getter)
-    def _collect_databases_data(self, tags):
+    def collect_databases_data(self, tags):
         """
         Collects database information and schemas and submits them to the agent's queue as dictionaries.
 
@@ -247,22 +248,25 @@ def _collect_databases_data(self, tags):
                             - data_length (int): The data length of the partition in bytes. If partition has
                                                  subpartitions, this is the sum of all subpartitions data_length.
         """
-        self._data_submitter.reset()
+        self._data_submitter.reset()  # Ensure we start fresh
         self._tags = tags
-        with closing(self._metadata.get_db_connection().cursor(CommenterDictCursor)) as cursor:
-            self._data_submitter.set_base_event_data(
-                self._check.reported_hostname,
-                self._check.database_identifier,
-                self._tags,
-                self._check._config.cloud_metadata,
-                self._check.version.version,
-                self._check.version.flavor,
-            )
-            db_infos = self._query_db_information(cursor)
-            self._data_submitter.store_db_infos(db_infos)
-            self._fetch_for_databases(db_infos, cursor)
-            self._data_submitter.submit()
-            self._log.debug("Finished collect_schemas_data")
+        self._data_submitter.set_base_event_data(
+            self._check.reported_hostname,
+            self._check.database_identifier,
+            self._tags,
+            self._check._config.cloud_metadata,
+            self._check.version.version,
+            self._check.version.flavor,
+        )
+        try:
+            with closing(self._metadata.get_db_connection().cursor(CommenterDictCursor)) as cursor:
+                db_infos = self._query_db_information(cursor)
+                self._data_submitter.store_db_infos(db_infos)
+                self._fetch_for_databases(db_infos, cursor)
+                self._data_submitter.submit()
+        finally:
+            self._data_submitter.reset()  # Ensure we reset in case of errors
+        self._log.debug("Finished collect_databases_data")
 
     def _fetch_for_databases(self, db_infos, cursor):
         start_time = time.time()
diff --git a/mysql/datadog_checks/mysql/metadata.py b/mysql/datadog_checks/mysql/metadata.py
index ac4f004c7694c..339a2e6336c6c 100644
--- a/mysql/datadog_checks/mysql/metadata.py
+++ b/mysql/datadog_checks/mysql/metadata.py
@@ -145,7 +145,7 @@ def run_job(self):
         if self._databases_data_enabled and elapsed_time_databases >= self._databases_data_collection_interval:
             self._last_databases_collection_time = time.time()
             try:
-                self._databases_data._collect_databases_data(self._tags)
+                self._databases_data.collect_databases_data(self._tags)
             except Exception as e:
                 self._log.error(
                     """An error occurred while collecting schema data.

From 6a1e3c8429b9658855265fbcfc550b78da3d8d8c Mon Sep 17 00:00:00 2001
From: Nathan Adams <nathan.adams@datadoghq.com>
Date: Tue, 2 Sep 2025 13:33:34 -0400
Subject: [PATCH 2/2] Define openai dataflows. (#21212)

---
 openai/assets/dataflows.yaml | 7 +++++++
 1 file changed, 7 insertions(+)
 create mode 100644 openai/assets/dataflows.yaml

diff --git a/openai/assets/dataflows.yaml b/openai/assets/dataflows.yaml
new file mode 100644
index 0000000000000..32de4a364b3b7
--- /dev/null
+++ b/openai/assets/dataflows.yaml
@@ -0,0 +1,7 @@
+provides:
+  - id: openai-usage-metrics
+    always_on: true
+  - id: openai-cloud-cost-data
+    always_on: false
+  - id: openai-llm-completions
+    always_on: false