Revert "[SDBM-842] Postgres integration performance optimization: Limit how many records are pulled from pg_stat_statements. (#17187)" (#17397)

This reverts commit 798f2f4.

Author: amw-zero

postgres/changelog.d/17397.fixed

@@ -0,0 +1,3 @@
+Revert Postgres Optimization (#17187).
+
+This appears to lead to inflated metrics in certain cases. Removing this optimization while we fix the inflated metrics.

postgres/datadog_checks/postgres/query_calls_cache.py

@@ -19,7 +19,6 @@
 from datadog_checks.base.utils.tracking import tracked_method
 from datadog_checks.postgres.cursor import CommenterCursor, CommenterDictCursor
 
-from .query_calls_cache import QueryCallsCache
 from .util import DatabaseConfigurationError, payload_pg_version, warning_with_tags
 from .version_utils import V9_4, V14
 
@@ -28,21 +27,15 @@
 except ImportError:
     from ..stubs import datadog_agent
 
-QUERYID_TO_CALLS_QUERY = """
-SELECT queryid, calls
-  FROM {pg_stat_statements_view}
-  WHERE queryid IS NOT NULL
-"""
-
 STATEMENTS_QUERY = """
 SELECT {cols}
   FROM {pg_stat_statements_view} as pg_stat_statements
   LEFT JOIN pg_roles
          ON pg_stat_statements.userid = pg_roles.oid
   LEFT JOIN pg_database
          ON pg_stat_statements.dbid = pg_database.oid
-  WHERE query NOT LIKE 'EXPLAIN %%'
-  AND queryid = ANY('{{ {called_queryids} }}'::bigint[])
+  WHERE query != '<insufficient privilege>'
+  AND query NOT LIKE 'EXPLAIN %%'
   {filters}
   {extra_clauses}
 """
@@ -153,7 +146,6 @@ def __init__(self, check, config, shutdown_callback):
         self.tags = None
         self._state = StatementMetrics()
         self._stat_column_cache = []
-        self._query_calls_cache = QueryCallsCache()
         self._track_io_timing_cache = None
         self._obfuscate_options = to_native_string(json.dumps(self._config.obfuscator_options))
         # full_statement_text_cache: limit the ingestion rate of full statement text events per query_signature
@@ -185,11 +177,7 @@ def _get_pg_stat_statements_columns(self):
 
         # Querying over '*' with limit 0 allows fetching only the column names from the cursor without data
         query = STATEMENTS_QUERY.format(
-            cols='*',
-            pg_stat_statements_view=self._config.pg_stat_statements_view,
-            extra_clauses="LIMIT 0",
-            filters="",
-            called_queryids="",
+            cols='*', pg_stat_statements_view=self._config.pg_stat_statements_view, extra_clauses="LIMIT 0", filters=""
         )
         with self._check._get_main_db() as conn:
             with conn.cursor(cursor_factory=CommenterCursor) as cursor:
@@ -198,39 +186,6 @@ def _get_pg_stat_statements_columns(self):
                 self._stat_column_cache = col_names
                 return col_names
 
-    def _check_called_queries(self):
-        pgss_view_without_query_text = self._config.pg_stat_statements_view
-        if pgss_view_without_query_text == "pg_stat_statements":
-            # Passing false for the showtext argument leads to a huge performance increase. This
-            # allows the engine to avoid retrieving the potentially large amount of text data.
-            # The query count query does not depend on the statement text, so it's safe for this use case.
-            # For more info: https://www.postgresql.org/docs/current/pgstatstatements.html#PGSTATSTATEMENTS-FUNCS
-            pgss_view_without_query_text = "pg_stat_statements(false)"
-
-        with self._check._get_main_db() as conn:
-            with conn.cursor(cursor_factory=CommenterCursor) as cursor:
-                called_queryids = []
-                query = QUERYID_TO_CALLS_QUERY.format(pg_stat_statements_view=pgss_view_without_query_text)
-                rows = self._execute_query(cursor, query, params=(self._config.dbname,))
-                for row in rows:
-                    queryid = row[0]
-                    if queryid is None:
-                        continue
-
-                    calls = row[1]
-                    calls_changed = self._query_calls_cache.set_calls(queryid, calls)
-                    if calls_changed:
-                        called_queryids.append(queryid)
-                self._query_calls_cache.end_query_call_snapshot()
-                self._check.gauge(
-                    "dd.postgresql.pg_stat_statements.calls_changed",
-                    len(called_queryids),
-                    tags=self.tags,
-                    hostname=self._check.resolved_hostname,
-                )
-
-                return called_queryids
-
     def run_job(self):
         # do not emit any dd.internal metrics for DBM specific check code
         self.tags = [t for t in self._tags if not t.startswith('dd.internal')]
@@ -267,7 +222,6 @@ def collect_per_statement_metrics(self):
     @tracked_method(agent_check_getter=agent_check_getter, track_result_length=True)
     def _load_pg_stat_statements(self):
         try:
-            called_queryids = self._check_called_queries()
             available_columns = set(self._get_pg_stat_statements_columns())
             missing_columns = PG_STAT_STATEMENTS_REQUIRED_COLUMNS - available_columns
             if len(missing_columns) > 0:
@@ -342,7 +296,6 @@ def _load_pg_stat_statements(self):
                             pg_stat_statements_view=self._config.pg_stat_statements_view,
                             filters=filters,
                             extra_clauses="",
-                            called_queryids=', '.join([str(i) for i in called_queryids]),
                         ),
                         params=params,
                     )