Add configurable limit to index metric (#20012)

* testing a branch

* limit cardinality of index metrics

* add changelog

* delete zombie file and clarify config description

Author: azhou-datadog

mysql/assets/configuration/spec.yaml

@@ -567,6 +567,13 @@ files:
                 value:
                   type: number
                   example: 300
+              - name: limit
+                description: |
+                  Set the maximum number of unique (by combination of database, table, and index name)
+                  indexes to collect. Defaults to 1000.
+                value:
+                  type: integer
+                  example: 1000
           - name: aws
             description: |
               This block defines the configuration for AWS RDS and Aurora instances.

mysql/changelog.d/20012.added

@@ -0,0 +1 @@
+added configurable limit to index metric collection

mysql/datadog_checks/mysql/config_models/instance.py

@@ -74,6 +74,7 @@ class IndexMetrics(BaseModel):
     )
     collection_interval: Optional[float] = None
     enabled: Optional[bool] = None
+    limit: Optional[int] = None
 
 
 class MetricPatterns(BaseModel):

mysql/datadog_checks/mysql/data/conf.yaml.example

@@ -537,6 +537,12 @@ instances:
         #
         # enabled: true
 
+        ## @param limit - integer - optional - default: 1000
+        ## Set the maximum number of unique (by combination of database, table, and index name)
+        ## indexes to collect. Defaults to 1000.
+        #
+        # limit: 1000
+
     ## This block defines the configuration for AWS RDS and Aurora instances.
     ##
     ## Complete this section if you have installed the Datadog AWS Integration

mysql/datadog_checks/mysql/index_metrics.py

@@ -9,13 +9,21 @@
     'query': """
         SELECT
             database_name,
-            table_name,
+            CASE
+                WHEN table_name LIKE '%#p#%' THEN SUBSTRING_INDEX(table_name, '#p#', 1)
+                ELSE table_name
+            END AS base_table_name,
             index_name,
-            stat_value * @@innodb_page_size AS index_size_bytes
+            SUM(stat_value * @@innodb_page_size) AS index_size_bytes
         FROM
             mysql.innodb_index_stats
         WHERE
             stat_name = 'size'
+        GROUP BY
+            database_name, base_table_name, index_name
+        ORDER BY
+            index_size_bytes DESC
+        LIMIT {INDEX_LIMIT}
     """.strip(),
     'columns': [
         {'name': 'db', 'type': 'tag'},
@@ -28,16 +36,35 @@
     'name': 'performance_schema.table_io_waits_summary_by_index_usage',
     'query': """
         SELECT
-            object_schema,
-            object_name,
-            index_name,
-            count_read,
-            count_update,
-            count_delete
-        FROM
-            performance_schema.table_io_waits_summary_by_index_usage
-        WHERE index_name IS NOT NULL
-        AND object_schema NOT IN ('mysql', 'performance_schema')
+            t.object_schema AS schema_name,
+            t.object_name AS table_name,
+            t.index_name,
+            t.count_read,
+            t.count_update,
+            t.count_delete
+        FROM performance_schema.table_io_waits_summary_by_index_usage t
+        JOIN (
+            SELECT
+                database_name,
+                CASE
+                    WHEN table_name LIKE '%#p#%' THEN SUBSTRING_INDEX(table_name, '#p#', 1)
+                    ELSE table_name
+                END AS base_table_name,
+                index_name,
+                SUM(stat_value * @@innodb_page_size) AS index_size_bytes
+            FROM
+                mysql.innodb_index_stats
+            WHERE
+                stat_name = 'size'
+            GROUP BY
+                database_name, base_table_name, index_name
+            ORDER BY
+                index_size_bytes DESC
+            LIMIT {INDEX_LIMIT}
+        ) i ON t.object_schema = i.database_name
+            AND t.object_name = i.base_table_name
+            AND t.index_name = i.index_name
+            AND t.index_name IS NOT NULL
     """.strip(),
     'columns': [
         {'name': 'db', 'type': 'tag'},
@@ -50,6 +77,7 @@
 }
 
 DEFAULT_INDEX_METRIC_COLLECTION_INTERVAL = 300  # 5 minutes
+DEFAULT_INDEX_LIMIT = 1000  # Default number of top indexes to collect
 
 
 class MySqlIndexMetrics:
@@ -64,12 +92,20 @@ def include_index_metrics(self) -> bool:
     def collection_interval(self) -> int:
         return int(self._config.index_config.get('collection_interval', DEFAULT_INDEX_METRIC_COLLECTION_INTERVAL))
 
+    @property
+    def index_limit(self) -> int:
+        return int(self._config.index_config.get('limit', DEFAULT_INDEX_LIMIT))
+
     @property
     def queries(self):
         # make a copy of the query to avoid modifying the original
         # in case different instances have different collection intervals
         usage_query = QUERY_INDEX_USAGE.copy()
         size_query = QUERY_INDEX_SIZE.copy()
+
+        # Update the index limit in the queries
+        size_query['query'] = size_query['query'].format(INDEX_LIMIT=self.index_limit)
+        usage_query['query'] = usage_query['query'].format(INDEX_LIMIT=self.index_limit)
         usage_query['collection_interval'] = self.collection_interval
         size_query['collection_interval'] = self.collection_interval
         return [size_query, usage_query]