Postgres cross-org telemetry (#18758)

datadog_checks_base/changelog.d/18758.added

@@ -0,0 +1 @@
+Added Postgres cross-org telemetry metrics.

datadog_checks_base/datadog_checks/base/stubs/datadog_agent.py

@@ -29,6 +29,7 @@ def __init__(self):
         self._process_start_time = 0
         self._external_tags = []
         self._host_tags = "{}"
+        self._sent_telemetry = defaultdict(list)
 
     def get_default_config(self):
         return {'enable_metadata_collection': True, 'disable_unsafe_yaml': True}
@@ -84,6 +85,12 @@ def assert_external_tags_count(self, count):
             count, tags_count, repr(self._external_tags)
         )
 
+    def assert_telemetry(self, check_name, metric_name, metric_type, metric_value):
+        values = self._sent_telemetry[(check_name, metric_name, metric_type)]
+        assert metric_value in values, 'Expected value {} for check {}, metric {}, type {}. Found {}.'.format(
+            metric_value, check_name, metric_name, metric_type, values
+        )
+
     def get_hostname(self):
         return self._hostname
 
@@ -152,6 +159,9 @@ def obfuscate_mongodb_string(self, command):
         # Passthrough stub: obfuscation implementation is in Go code.
         return command
 
+    def emit_agent_telemetry(self, check_name, metric_name, metric_value, metric_type):
+        self._sent_telemetry[(check_name, metric_name, metric_type)].append(metric_value)
+
 
 # Use the stub as a singleton
 datadog_agent = DatadogAgentStub()

postgres/changelog.d/18758.added

@@ -0,0 +1 @@
+Added Postgres cross-org telemetry metrics.

postgres/datadog_checks/postgres/metadata.py

@@ -297,6 +297,8 @@ def report_postgres_metadata(self):
 
             # Tuned from experiments on staging, we may want to make this dynamic based on schema size in the future
             chunk_size = 50
+            total_tables = 0
+            start_time = time.time()
 
             for database in schema_metadata:
                 dbname = database["name"]
@@ -332,11 +334,31 @@ def report_postgres_metadata(self):
 
                                 if buffer_column_count >= 100_000:
                                     self._flush_schema(base_event, database, schema, tables_buffer)
+                                    total_tables += len(tables_buffer)
                                     tables_buffer = []
                                     buffer_column_count = 0
 
                             if len(tables_buffer) > 0:
                                 self._flush_schema(base_event, database, schema, tables_buffer)
+                                total_tables += len(tables_buffer)
+            elapsed_ms = (time.time() - start_time) * 1000
+            self._check.histogram(
+                "dd.postgres.schema.time",
+                elapsed_ms,
+                tags=self._check.tags,
+                hostname=self._check.resolved_hostname,
+                raw=True,
+            )
+            self._check.gauge(
+                "dd.postgres.schema.tables_count",
+                total_tables,
+                tags=self._check.tags,
+                hostname=self._check.resolved_hostname,
+                raw=True,
+            )
+            datadog_agent.emit_agent_telemetry("postgres", "schema_tables_elapsed_ms", elapsed_ms, "gauge")
+            datadog_agent.emit_agent_telemetry("postgres", "schema_tables_count", total_tables, "gauge")
+
             self._is_schemas_collection_in_progress = False
 
     def _should_collect_metadata(self, name, metadata_type):

postgres/datadog_checks/postgres/postgres.py

@@ -646,6 +646,8 @@ def _collect_metric_autodiscovery(self, instance_tags, scopes, scope_type):
             hostname=self.resolved_hostname,
             raw=True,
         )
+        telemetry_metric = scope_type.replace("_", "", 1)  # remove the first underscore to match telemetry convention
+        datadog_agent.emit_agent_telemetry("postgres", f"{telemetry_metric}_ms", elapsed_ms, "histogram")
         if elapsed_ms > self._config.min_collection_interval * 1000:
             self.record_warning(
                 DatabaseConfigurationError.autodiscovered_metrics_exceeds_collection_interval,

postgres/datadog_checks/postgres/statement_samples.py

@@ -434,6 +434,18 @@ def _report_check_hist_metrics(self, start_time, row_len, method_name):
             hostname=self._check.resolved_hostname,
             raw=True,
         )
+        datadog_agent.emit_agent_telemetry(
+            "postgres",
+            f"{method_name}_ms",
+            (time.time() - start_time) * 1000,
+            "histogram",
+        )
+        datadog_agent.emit_agent_telemetry(
+            "postgres",
+            f"{method_name}_count",
+            row_len,
+            "histogram",
+        )
 
     def run_job(self):
         # do not emit any dd.internal metrics for DBM specific check code
@@ -478,6 +490,13 @@ def _collect_statement_samples(self):
                 tags=self.tags,
                 raw=True,
             )
+            datadog_agent.emit_agent_telemetry(
+                "postgres",
+                "collect_activity_snapshot_ms",
+                (time.time() - start_time) * 1000,
+                "histogram",
+            )
+
         elapsed_ms = (time.time() - start_time) * 1000
         self._check.histogram(
             "dd.postgres.collect_statement_samples.time",
@@ -514,6 +533,18 @@ def _collect_statement_samples(self):
             hostname=self._check.resolved_hostname,
             raw=True,
         )
+        datadog_agent.emit_agent_telemetry(
+            "postgres",
+            "collect_statement_samples_ms",
+            elapsed_ms,
+            "histogram",
+        )
+        datadog_agent.emit_agent_telemetry(
+            "postgres",
+            "collect_statement_samples_count",
+            submitted_count,
+            "gauge",
+        )
 
     @staticmethod
     def _to_active_session(row, track_activity_query_size):