From 7856f7bc2eb9808537fd21b094cf15fc71abd233 Mon Sep 17 00:00:00 2001
From: Tanmoy Sarkar <57363826+tanmoysrt@users.noreply.github.com>
Date: Wed, 18 Dec 2024 10:52:37 +0000
Subject: [PATCH] feat(db-perf-report): added database performance report

Endpoint - /benches/<bench>/sites/<site>/database/performance-report
The API endpoint will send back some summarized reports
- Top 10 time consuming SQL Queries
- Top 10 queries with full table scans
- Unused Indexes
- Redundant Indexes
---
 agent/database.py | 58 +++++++++++++++++++++++++++++++++++++++++++++++
 agent/site.py     |  4 ++++
 agent/web.py      | 16 ++++++++++++-
 3 files changed, 77 insertions(+), 1 deletion(-)

diff --git a/agent/database.py b/agent/database.py
index f23c1849..91a37c93 100644
--- a/agent/database.py
+++ b/agent/database.py
@@ -321,6 +321,64 @@ def fetch_database_column_statistics(self, table):
 
         return result
 
+    def fetch_summarized_performance_report(self):
+        queries = f"""
+-- Top 10 time consuming queries;
+SELECT
+    (SUM_TIMER_WAIT / SUM(SUM_TIMER_WAIT) OVER() * 100) AS percent,
+    round(SUM_TIMER_WAIT/1000000000, 1) AS total_time_ms,
+    COUNT_STAR AS calls,
+    round(AVG_TIMER_WAIT/1000000000, 1) AS avg_time_ms,
+    DIGEST_TEXT AS query
+FROM performance_schema.events_statements_summary_by_digest
+    WHERE SCHEMA_NAME='{self.database_name}'
+    ORDER BY SUM_TIMER_WAIT DESC
+    LIMIT 10;
+
+-- Top 10 queries with full table scans;
+SELECT t1.exec_count AS calls,
+       t1.rows_examined AS rows_examined,
+       t1.rows_sent AS rows_sent,
+       t1.query AS query,
+       t2.DIGEST_TEXT AS example
+FROM sys.statements_with_full_table_scans AS t1
+LEFT JOIN (
+    SELECT DIGEST, FIRST_VALUE(DIGEST_TEXT) OVER (PARTITION BY DIGEST ORDER BY RAND()) AS DIGEST_TEXT
+    FROM performance_schema.events_statements_summary_by_digest
+) AS t2 ON t1.digest = t2.DIGEST
+ORDER BY rows_examined DESC
+LIMIT 10;
+
+-- Unused Indexes;
+SELECT
+    index_name,
+    object_name AS table_name
+FROM
+    sys.schema_unused_indexes
+WHERE
+    object_schema='{self.database_name}';
+
+-- Redundant Indexes;
+SELECT
+    table_name,
+    redundant_index_name,
+    redundant_index_columns,
+    dominant_index_name,
+    dominant_index_columns
+FROM
+    sys.schema_redundant_indexes
+WHERE
+    table_schema='{self.database_name}';
+"""
+
+        result = self._run_sql(queries, as_dict=True)
+        return {
+            "top_10_time_consuming_queries": result[0]["output"],
+            "top_10_queries_with_full_table_scan": result[1]["output"],
+            "unused_indexes": result[2]["output"],
+            "redundant_indexes": result[3]["output"],
+        }
+
     # Private helper methods
     def _run_sql(  # noqa C901
         self, query: str, commit: bool = False, as_dict: bool = False, allow_all_stmt_types: bool = False
diff --git a/agent/site.py b/agent/site.py
index 91f521ed..374ede8e 100644
--- a/agent/site.py
+++ b/agent/site.py
@@ -914,6 +914,10 @@ def analyze_slow_queries(self, queries: list[dict], database_root_password: str)
             result.append(query)
         return result
 
+    def fetch_summarized_database_performance_report(self, mariadb_root_password: str):
+        database = self.db_instance(username="root", password=mariadb_root_password)
+        return database.fetch_summarized_performance_report()
+
     def db_instance(self, username: str | None = None, password: str | None = None) -> Database:
         if not username:
             username = self.user
diff --git a/agent/web.py b/agent/web.py
index 9ecbeedf..219643d3 100644
--- a/agent/web.py
+++ b/agent/web.py
@@ -591,7 +591,7 @@ def run_sql(bench, site):
 
 
 @application.route(
-    "/benches/<string:bench>/sites/<string:site>/database/analyze_slow_queries", methods=["GET"]
+    "/benches/<string:bench>/sites/<string:site>/database/analyze-slow-queries", methods=["GET"]
 )
 @validate_bench_and_site
 def analyze_slow_queries(bench: str, site: str):
@@ -602,6 +602,20 @@ def analyze_slow_queries(bench: str, site: str):
     return {"job": job}
 
 
+@application.route(
+    "/benches/<string:bench>/sites/<string:site>/database/performance-report", methods=["POST"]
+)
+def database_performance_report(bench, site):
+    data = request.json
+    result = (
+        Server()
+        .benches[bench]
+        .sites[site]
+        .fetch_summarized_database_performance_report(data["mariadb_root_password"])
+    )
+    return jsonify(json.loads(json.dumps(result, cls=JSONEncoderForSQLQueryResult)))
+
+
 @application.route("/benches/<string:bench>/sites/<string:site>/database/users", methods=["POST"])
 @validate_bench_and_site
 def create_database_user(bench, site):