Performing Full-Link Analysis for a SQL Record

Scenarios

You can query details about SQL execution links using the unique SQL ID or transaction ID of an application SQL statement.

Prerequisites

• You have logged in using an account assigned the gaussdb:instance:listSqlLink action.

• SQL Explorer has been enabled, and valid data has been collected.

Constraints

• If the transaction ID is 0, it cannot be used to query link details.
• If the unique SQL ID is 0, no link information is available.
• The DB engine version of GaussDB instances must be V2.0-8.0.1 or later, and the kernel version must be 505.0.0 or later. For details about how to query the version, see Checking the DB Engine and Kernel Engine Version of a GaussDB Instance.

Procedure

1. Log in to the management console.
2. Click in the upper left corner and select a region and project.
3. Click in the upper left corner of the page and choose Databases > GaussDB.
4. On the Instances page, click the name of the target instance to go to the Basic Information page.
5. Choose Diagnostics and Optimization > SQL Diagnosis and click the SQL Explorer tab.
6. On the Full SQL Statistics tab, click to expand the target row. Alternatively, click the Full Data by SQL Statement tab.
7. Click the transaction ID or click More in the Operation column and choose End-to-End Trace to go to the SQL full-link analysis page.
   • If you click More in the Operation column and choose End-to-End Trace, the execution link details of the SQL statement are displayed.
   • If you click a transaction ID (which is not 0), the execution link details of the transaction are displayed.
   Figure 1 SQL link information
   

   Table 1 Parameters for SQL link analysis

   Module	Parameter	Description
   Elapsed Time for Resources	CPU	Time spent on CPU processing
   	I/O	Time spent on I/O operations
   	Others	Valid DB time minus the CPU and I/O time
   Elapsed Time for Kernel Execution	SQL Parsing	Time spent on parsing the SQL statement
   	Query Rewriting	Time spent on rewriting the SQL statement
   	Execution Plan Generation	Time spent on generating a SQL execution plan
   	Execution in Executor	Time spent on executing the SQL statement
   	Others	Total SQL execution time minus the time spent on SQL parsing, SQL rewriting, SQL plan generation, and execution in the executor
   Elapsed Time for Code-related Wait Events	Top 5 Events by Elapsed Time	Top events with the longest elapsed time
   	Elapsed Time for Other Events	Time consumed by all events minus the time consumed by top 5 events
   	Others	Valid DB time minus the time consumed by all wait event code
   Elapsed Time for Resource-related Wait Events - Elapsed Time for I/O	Top 5 Events by Elapsed Time	The most time-consuming I/O events
   	Elapsed Time for Other Events	Time consumed by all I/O events minus the time consumed by top 5 I/O events
   	Others	Valid DB time minus the time consumed by all I/O wait events
   Elapsed Time for Resource-related Wait Events - Elapsed Time for Locks	Top 5 Events by Elapsed Time	The most time-consuming lock events
   	Elapsed Time for Other Events	Time consumed by all lock events minus the time consumed by top 5 lock events
   	Others	Valid DB time minus the time consumed by all lock events
   Elapsed Time for Resource-related Wait Events - Elapsed Time for Lightweight Locks	Top 5 Events by Elapsed Time	The most time-consuming lightweight lock events
   	Elapsed Time for Other Events	Time consumed by all lightweight lock events minus the time consumed by top 5 lightweight lock events
   	Others	Valid DB time minus the time consumed by all lightweight lock events

   For details about wait event parameters, see the description of wait_status and wait_event in "Developer Guide" > "System Catalogs and System Views" > "System Views" > "Other System Views" > "PG_THREAD_WAIT_STATUS."

8. Hover the pointer over a node to view the details about SQL execution on that node.
   Figure 2 SQL execution information
   

   DB Time indicates the total DB time, in microseconds, which is accumulated if multiple threads are involved. The way for collecting DB time statistics is slightly different from that for collecting start/end time statistics, but the difference does not affect performance issue analysis.