Slow Query

On the Slow Query screen, you can monitor and view slow SQL that exceeds the threshold time set in DB parameters. MySQL records slow queries to a log file or table, but to check them, you need to connect to the DB server via terminal and read the text log directly or query the mysql.slow_log table. This screen collects slow query logs in real time and provides them as visual charts and lists.

• In the Slow Query chart, you can visually identify when slow queries occurred and their execution time.

• In Slow Query List, you can check detailed information for individual queries and view execution plans (Plan).

• Both real-time (Live) mode and historical time queries are supported.

Note

An additional agent is required to collect slow queries. In on-premises environments, install the XOS agent, and in cloud (AWS RDS, etc.) environments, install the DBXC agent.

The slow query chart is similar to the SQL Elapse Map chart on the dashboard, but with the following differences:

Table | Differences between Slow Query chart and SQL Elapse Map chart

Category	Slow Query	SQL Elapse Map
Data Source	Slow query log recorded by DB	Active sessions collected by the agent at 5-second intervals
Execution Time Accuracy	Accurate — actual execution time recorded by DB	Up to 5 seconds of error possible due to 5-second interval collection
Collection Target	Only queries exceeding the threshold time (long_query_time)	All active sessions running at the time of collection

Prerequisites

To display data on this screen, DB configuration, DBX Agent configuration, and either XOS Agent or DBXC Agent configuration depending on your environment are required.

DB configuration

Set the slow query-related parameters in my.cnf (or my.ini). MySQL restart is required after changes.

slow_query_log = 1
long_query_time = 3
log_output = FILE

Parameter	Description
slow_query_log	Whether to enable slow query logging. Enabled when set to 1.
long_query_time	Records SQL that exceeds the set value (seconds) in the log. For example, if set to 3, SQL taking 3 or more seconds is recorded.
log_output	Log storage method. FILE records to a file; TABLE records to the mysql.slow_log table.

Note

If log_output=TABLE is set, you must grant query permission to the monitoring account. Note that TABLE mode may impact DB performance if there are many slow queries.

GRANT SELECT ON mysql.slow_log TO whatap;

DBX Agent configuration

In whatap.conf, the slow_query_log option defaults to ON, so no additional configuration is needed.

slow_query_log=ON

Option	Type	Default	Description
slow_query_log	Boolean	ON	Enable slow query log collection

OS agent configuration

An agent matching the DB operating environment is required to collect slow query logs.

Environment	Required Agent	Key Configuration	Setup Guide
On-premises (Linux, Windows)	XOS agent	None (if auto-detection fails, specify log file path directly in slow_query=)	XOS Agent Configuration
Cloud (AWS RDS, NCP, etc.)	DBXC agent	slow_query: true (per-instance setting in config.yaml)	DBXC Agent Configuration

Slow Query

Basic options

In the top filter area of the slow query screen, you can select the time period and target DB instance to query.

• Time: Select real-time (Live) mode or a past time. Live mode automatically refreshes every 10 seconds.

• Instance: Select the target DB instance to query. You can select all or individual instances.

Slow Query chart

Displays the time when slow queries occurred as a scatter chart. The Y-axis is execution time (seconds) and the X-axis is time.

• Each dot represents one slow query; dots higher up indicate queries that took longer.

• Mouse drag on the chart area displays only the queries from that interval in the list below.

• The ∧ / ∨ buttons in the upper right adjust the Y-axis range.

Slow Query List

Displays the list of slow queries in the selected interval from the chart.

Column	Description
Instance	Name of the instance where the query was executed
DB	Database name
Start Time	Query start time
User	Executing user name
Host	Connection host
ID	Thread ID
Schema	Schema name
QC Hit	Whether the query cache was hit
Query Time (s)	Query execution time (seconds)
Lock Time (s)	Lock wait time (seconds)
Rows Sent	Number of rows returned to the client
Rows Examined	Number of rows examined for query processing
Rows Affected	Number of rows changed
Bytes Sent	Number of bytes sent to the client
SQL	Executed SQL statement

SQL details

Click the query in the SQL column to open the details window.

Data interpretation guide

Key checkpoints

• Queries with long Query Time: Queries with long individual execution time → SQL tuning, index check

• Queries with long Lock Time: Long lock wait causing performance degradation → Check for transaction contention and lock causes

• Queries with large Rows Examined: When the number of rows examined is large relative to rows returned → Missing index or inefficient scan

• When the same SQL appears repeatedly: A specific query is consistently slow → Fundamental tuning required

• Concentrated in a specific time period: Check for external factors such as DB load, lock contention, batch jobs during that time

Practical usage scenarios

Scenario 1: Real-time monitoring of slow queries

1. Observe the Slow Query chart in Live mode

2. When new dots appear, drag and select that area

3. Check queries in the Slow Query List → Click SQL to view details and execution plan

Scenario 2: Analyzing slow queries in a specific time period

1. Select the problematic time period in the time filter

2. Drag the interval where queries are concentrated in the chart

3. Check queries with the longest Query Time first to analyze the cause

Scenario 3: Adjusting the slow query threshold

1. If too many slow queries are being collected, raise the long_query_time value to loosen the threshold

2. If no slow queries are being collected, lower the long_query_time value or check the slow_query_log=ON setting

Analyzing with AI

Click the WhaTap AI button on the screen to open the WhaTap AI chat window, where the queried screen data is summarized and the analysis result is shown as a conversation. After reading the result, you can continue with follow-up questions in the same conversation.

• The data used for the analysis is the top 20 items with the longest elapsed time among the items selected on the scatter plot.

• The analysis target is the data queried on the screen. If you change the query conditions and analyze again, the analysis is based on the changed data.

• The conversation history is recorded in the AI Analysis - {screen name} format.