[DBMON-6076] fix collect_wal_metrics default on 9.6 and below (#22377)

* fix collect_wal_metrics default on 9.6 and below

* Fix collect_wal_metrics default in tests

* Add changelog

Author: eric-weaver

postgres/assets/configuration/spec.yaml

@@ -364,16 +364,20 @@ files:
     - name: collect_wal_metrics
       description: |
         Collect metrics about WAL file age.
-        NOTE:
-        For Postgres 9.6 and below, you must run the check local to your database if you want to enable this option.
-        Starting Postgres 10, WAL file metrics are enabled by default using `pg_ls_waldir` and don't need local access.
+
+        When unset, this defaults to `true` for Postgres 10+ (uses `pg_ls_waldir` via SQL) and `false` for
+        Postgres 9.6 and below (requires local filesystem access to the WAL directory).
+
+        Set to `true` explicitly for Postgres 9.6 and below only if the Agent runs on the same host as the
+        database and has read access to the WAL directory. You must also configure `data_directory`.
       value:
         type: boolean
         example: true
+        display_default: null
     - name: data_directory
       description: |
-        The data directory of your postgres installation
-        Required when collecting WAL metrics on Postgres 9.6.
+        The data directory of your postgres installation.
+        Only used when `collect_wal_metrics` is explicitly set to `true` on Postgres 9.6 and below.
       value:
         type: string
         example: /usr/local/pgsql/data

postgres/changelog.d/22377.fixed

@@ -0,0 +1 @@
+Fixes collect_wal_metrics config default on 9.6 and below to correctly default to disabled

postgres/datadog_checks/postgres/config_models/defaults.py

@@ -52,10 +52,6 @@ def instance_collect_function_metrics():
     return False
 
 
-def instance_collect_wal_metrics():
-    return True
-
-
 def instance_data_directory():
     return '/usr/local/pgsql/data'
 

postgres/datadog_checks/postgres/data/conf.yaml.example

@@ -271,17 +271,20 @@ instances:
     #
     collect_bloat_metrics: false
 
-    ## @param collect_wal_metrics - boolean - optional - default: true
+    ## @param collect_wal_metrics - boolean - optional
     ## Collect metrics about WAL file age.
-    ## NOTE:
-    ## For Postgres 9.6 and below, you must run the check local to your database if you want to enable this option.
-    ## Starting Postgres 10, WAL file metrics are enabled by default using `pg_ls_waldir` and don't need local access.
+    ##
+    ## When unset, this defaults to `true` for Postgres 10+ (uses `pg_ls_waldir` via SQL) and `false` for
+    ## Postgres 9.6 and below (requires local filesystem access to the WAL directory).
+    ##
+    ## Set to `true` explicitly for Postgres 9.6 and below only if the Agent runs on the same host as the
+    ## database and has read access to the WAL directory. You must also configure `data_directory`.
     #
     # collect_wal_metrics: true
 
     ## @param data_directory - string - optional - default: /usr/local/pgsql/data
-    ## The data directory of your postgres installation
-    ## Required when collecting WAL metrics on Postgres 9.6.
+    ## The data directory of your postgres installation.
+    ## Only used when `collect_wal_metrics` is explicitly set to `true` on Postgres 9.6 and below.
     #
     # data_directory: /usr/local/pgsql/data
 

postgres/datadog_checks/postgres/postgres.py

@@ -397,7 +397,8 @@ def dynamic_queries(self):
             if self.is_aurora is False:
                 queries.append(QUERY_PG_STAT_WAL_RECEIVER)
                 if self._config.collect_wal_metrics is not False:
-                    # collect wal metrics for pg >= 10 only if the user has not explicitly disabled it
+                    # collect wal metrics for pg >= 10 by default (uses pg_ls_waldir via SQL)
+                    # unless the user has explicitly disabled it
                     queries.append(WAL_FILE_METRICS)
             if self._config.collect_buffercache_metrics:
                 queries.append(BUFFERCACHE_METRICS)
@@ -1137,8 +1138,9 @@ def check(self, _):
                     self.statement_metrics.run_job_loop(tags)
                     self.statement_samples.run_job_loop(tags)
                     self.metadata_samples.run_job_loop(tags)
-                if self._config.collect_wal_metrics:
-                    # collect wal metrics for pg < 10, disabled by enabled
+                if self._config.collect_wal_metrics is True:
+                    # collect wal metrics for pg < 10 only when explicitly enabled
+                    # (requires local filesystem access to the WAL directory)
                     self._collect_wal_metrics()
 
             if self._query_manager.queries:

postgres/tests/test_config_defaults.py

@@ -50,7 +50,7 @@
     'collect_database_size_metrics': True,
     'collect_default_database': True,
     'collect_bloat_metrics': False,
-    'collect_wal_metrics': True,
+    'collect_wal_metrics': None,  # Version-aware: defaults to True for PG 10+, False for PG 9.6 and below
     'tag_replication_role': True,
     'table_count_limit': 200,
     'max_relations': 300,