Recently, a very strange thing encountered in the maintenance of oracle rac is that the business occasionally appears extremely slow. The server load and database load are very low. There are no obvious errors in the database and host logs, and there are no congested sessions inside the database. Case. This article is hereby recorded for future reference!
Background: Since December, the application has started to give feedback, and there are no new product releases, and the issuance of documents will be particularly slow. Later, in order to temporarily solve the problem, the business was directed to a single node, and it was found that the business delivery speed returned to normal. Later, after a week or so, the document issuance speed was slow again. However, the server load and database load are very low. There are no obvious errors in the database and host logs, and there is no congested session inside the database. In order to resume business as soon as possible, the database service was restarted,
Business temporarily returned to normal. However, three or four days later, that is today, the issue of receipts was slow again.
problem analysis:
1. Based on the above background, there is no high load on the database and the host, especially if there is no session congestion inside the database. What I can think of is that the SQL execution plan may have changed.
2. View the SQL statements that are being executed by the business bearer node
3. View the execution plan of sql-5216pwt38ckhp
--Excellent execution plan, can be executed nearly 30 times at 3-8 points in a single document issuance
--Inefficient execution plan after the change, 3-10 points are issued for a single receipt, and only 1 execution is completed multiple times
4. Through the comparison of the execution plan before and after the change in step 3, we find
--The primary key unique index scan of the execution plan before the change, the range scan of the ordinary index of the execution plan after the change
--The execution plan buffer get before the change is very low as 11, and there is no disk read, the execution plan buffer after the change is as high as 280,000, and there is a physical disk read
--The execution plan before the change, after the unique index scan, scans a row of data back to the table, and the execution plan after the change scans a row of data back to the table after scanning the full partition of the index
--The execution plan before the change can be executed at 3-8 points for a single document issued for nearly 30 points, and the changed execution plan can only be executed more than 10,000 times after a single document issued at 3-10 points.
Problem handling:
Based on the analysis of the above problems, the SQL-5216pwt38ckhp statement is an update statement and uses bind variables. Therefore, without changing the sql, we
Choose a fixed sql execution plan. The method is to use the script coe_load_sql_profile.sql to fix the execution plan. Pay attention when the execution plan is fixed.
It is not possible to log in to the database with sys to execute the fixed-line execution plan. It needs to be executed by the business user. During the fixed-line process, you need to enter the sql_id twice, and then select the plan hash value of the excellent execution plan.
Attach the execution plan fixed script coe_load_sql_profile.sql:
SPO coe_load_sql_profile.log; SET DEF ON TERM OFF ECHO ON FEED OFF VER OFF HEA ON LIN 2000 PAGES 100 LONG 8000000 LONGC 800000 TRIMS ON TI OFF TIMI OFF SERVEROUT ON SIZE 1000000 NUM 20 SQLP SQL>; SET SERVEROUT ON SIZE UNL; REM REM $Header: 215187.1 coe_load_sql_profile.sql 11.4.5.5 2013/03/01 carlos.sierra $ REM REM Copyright (c) 2000-2013, Oracle Corporation. All rights reserved. REM REM AUTHOR REM [email protected] REM REM SCRIPT REM coe_load_sql_profile.sql REM REM DESCRIPTION REM This script loads a plan from a modified SQL into a Custom SQL REM Profile for the original SQL. REM If a good performing plan only reproduces with CBO Hints REM then you can load the plan of the modified version of the REM SQL into a Custom SQL Profile for the orignal SQL. REM In other words, the original SQL can use the plan that was REM generated out of the SQL with hints. REM REM PRE-REQUISITES REM 1. Have in cache or AWR the text for the original SQL. REM 2. Have in cache or AWR the plan for the modified SQL REM (usually with hints). REM REM PARAMETERS REM 1. ORIGINAL_SQL_ID (required) REM 2. MODIFIED_SQL_ID (required) REM 3. PLAN_HASH_VALUE (required) REM REM EXECUTION REM 1. Connect into SQL*Plus as user with access to data dictionary REM and privileges to create SQL Profiles. Do not use SYS. REM 2. Execute script coe_load_sql_profile.sql passing first two REM parameters inline or until requested by script. REM 3. Provide plan hash value of the modified SQL when asked. REM 4. Use a DBA user but not SYS. Do not connect as SYS as the staging REM table cannot be created in SYS schema and you will receive an error: REM ORA-19381: cannot create staging table in SYS schema REM REM EXAMPLE REM # sqlplus system REM SQL> START coe_load_sql_profile.sql gnjy0mn4y9pbm b8f3mbkd8bkgh REM SQL> START coe_load_sql_profile.sql; REM REM NOTES REM 1. This script works on 10g or higher. REM 2. For a similar script for 11g use coe_load_sql_baseline.sql, REM which uses SQL Plan Baselines instead of Custom SQL Profiles. REM 3. For possible errors see coe_load_sql_profile.log REM 4. If you get "ORA-06532: Subscript outside of limit, ORA-06512: at line 1" REM Then you may consider this change (only in a test and disposable system): REM create or replace TYPE sys.sqlprof_attr AS VARRAY(5000) of VARCHAR2(500); REM SET TERM ON ECHO OFF; PRO PRO Parameter 1: PRO ORIGINAL_SQL_ID (required) PRO DEF original_sql_id = '&1'; PRO PRO Parameter 2: PRO MODIFIED_SQL_ID (required) PRO DEF modified_sql_id = '&2'; PRO WITH p AS ( SELECT plan_hash_value FROM gv$sql_plan WHERE sql_id = TRIM('&&modified_sql_id.') AND other_xml IS NOT NULL UNION SELECT plan_hash_value FROM dba_hist_sql_plan WHERE sql_id = TRIM('&&modified_sql_id.') AND other_xml IS NOT NULL ), m AS ( SELECT plan_hash_value, SUM(elapsed_time)/SUM(executions) avg_et_secs FROM gv$sql WHERE sql_id = TRIM('&&modified_sql_id.') AND executions > 0 GROUP BY plan_hash_value ), a AS ( SELECT plan_hash_value, SUM(elapsed_time_total)/SUM(executions_total) avg_et_secs FROM dba_hist_sqlstat WHERE sql_id = TRIM('&&modified_sql_id.') AND executions_total > 0 GROUP BY plan_hash_value ) SELECT p.plan_hash_value, ROUND(NVL(m.avg_et_secs, a.avg_et_secs)/1e6, 3) avg_et_secs FROM p, m, a WHERE p.plan_hash_value = m.plan_hash_value(+) AND p.plan_hash_value = a.plan_hash_value(+) ORDER BY avg_et_secs NULLS LAST; PRO PRO Parameter 3: PRO PLAN_HASH_VALUE (required) PRO DEF plan_hash_value = '&3'; PRO PRO Values passed to coe_load_sql_profile: PRO ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PRO ORIGINAL_SQL_ID: "&&original_sql_id." PRO MODIFIED_SQL_ID: "&&modified_sql_id." PRO PLAN_HASH_VALUE: "&&plan_hash_value." PRO WHENEVER SQLERROR EXIT SQL.SQLCODE; SET TERM OFF ECHO ON; -- trim parameters COL original_sql_id NEW_V original_sql_id FOR A30; COL modified_sql_id NEW_V modified_sql_id FOR A30; COL plan_hash_value NEW_V plan_hash_value FOR A30; SELECT TRIM('&&original_sql_id.') original_sql_id, TRIM('&&modified_sql_id.') modified_sql_id, TRIM('&&plan_hash_value.') plan_hash_value FROM DUAL; -- open log file SPO coe_load_sql_profile_&&original_sql_id..log; GET coe_load_sql_profile.log; . -- get user COL connected_user NEW_V connected_user FOR A30; SELECT USER connected_user FROM DUAL; VAR sql_text CLOB; VAR other_xml CLOB; VAR signature NUMBER; VAR name VARCHAR2 (30); EXEC :sql_text := NULL; EXEC :other_xml := NULL; EXEC :signature := NULL; EXEC :name := NULL; -- get sql_text from memory DECLARE l_sql_text VARCHAR2(32767); BEGIN -- 10g see bug 5017909 FOR i IN (SELECT DISTINCT piece, sql_text FROM gv$sqltext_with_newlines WHERE sql_id = TRIM('&&original_sql_id.') ORDER BY 1, 2) LOOP IF :sql_text IS NULL THEN DBMS_LOB.CREATETEMPORARY(:sql_text, TRUE); DBMS_LOB.OPEN(:sql_text, DBMS_LOB.LOB_READWRITE); END IF; l_sql_text := REPLACE(i.sql_text, CHR(00), ' '); DBMS_LOB.WRITEAPPEND(:sql_text, LENGTH(l_sql_text), l_sql_text); END LOOP; IF :sql_text IS NOT NULL THEN DBMS_LOB.CLOSE(:sql_text); END IF; EXCEPTION WHEN OTHERS THEN DBMS_OUTPUT.PUT_LINE('getting original sql_text from memory: '||SQLERRM); :sql_text := NULL; END; / -- get sql_text from awr BEGIN IF :sql_text IS NULL OR NVL(DBMS_LOB.GETLENGTH(:sql_text), 0) = 0 THEN SELECT REPLACE(sql_text, CHR(00), ' ') INTO :sql_text FROM dba_hist_sqltext WHERE sql_id = TRIM('&&original_sql_id.') AND sql_text IS NOT NULL AND ROWNUM = 1; END IF; EXCEPTION WHEN OTHERS THEN DBMS_OUTPUT.PUT_LINE('getting original sql_text from awr: '||SQLERRM); :sql_text := NULL; END; / -- sql_text as found SELECT :sql_text FROM DUAL; -- check is sql_text for original sql is available SET TERM ON; BEGIN IF :sql_text IS NULL THEN RAISE_APPLICATION_ERROR(-20100, 'SQL_TEXT for original SQL_ID &&original_sql_id. was not found in memory (gv$sqltext_with_newlines) or AWR (dba_hist_sqltext).'); END IF; END; / SET TERM OFF; -- get other_xml from memory BEGIN FOR i IN (SELECT other_xml FROM gv$sql_plan WHERE sql_id = TRIM('&&modified_sql_id.') AND plan_hash_value = TO_NUMBER(TRIM('&&plan_hash_value.')) AND other_xml IS NOT NULL ORDER BY child_number, id) LOOP :other_xml := i.other_xml; EXIT; -- 1st END LOOP; EXCEPTION WHEN OTHERS THEN DBMS_OUTPUT.PUT_LINE('getting modified other_xml from memory: '||SQLERRM); :other_xml := NULL; END; / -- get other_xml from awr BEGIN IF :other_xml IS NULL OR NVL(DBMS_LOB.GETLENGTH(:other_xml), 0) = 0 THEN FOR i IN (SELECT other_xml FROM dba_hist_sql_plan WHERE sql_id = TRIM('&&modified_sql_id.') AND plan_hash_value = TO_NUMBER(TRIM('&&plan_hash_value.')) AND other_xml IS NOT NULL ORDER BY id) LOOP :other_xml := i.other_xml; EXIT; -- 1st END LOOP; END IF; EXCEPTION WHEN OTHERS THEN DBMS_OUTPUT.PUT_LINE('getting modified other_xml from awr: '||SQLERRM); :other_xml := NULL; END; / -- other_xml as found SELECT :other_xml FROM DUAL; -- validate other_xml SET TERM ON; BEGIN IF :other_xml IS NULL THEN RAISE_APPLICATION_ERROR(-20101, 'PLAN for modified SQL_ID &&modified_sql_id. and PHV &&plan_hash_value. was not found in memory (gv$sql_plan) or AWR (dba_hist_sql_plan).'); END IF; END; / SET ECHO OFF; DECLARE h SYS.SQLPROF_ATTR := SYS.SQLPROF_ATTR (); idx INTEGER := 0; l_pos NUMBER; l_hint VARCHAR2(32767); description VARCHAR2(500); PROCEDURE add_hint (p_hint IN VARCHAR2) IS BEGIN idx := idx + 1; DBMS_OUTPUT.PUT_LINE(LPAD(idx, 4, '0')||' '||p_hint); h.EXTEND; h(idx) := p_hint; END add_hint; BEGIN add_hint('BEGIN_OUTLINE_DATA'); FOR i IN (SELECT /*+ opt_param('parallel_execution_enabled', 'false') */ SUBSTR(EXTRACTVALUE(VALUE(d), '/hint'), 1, 4000) hint FROM TABLE(XMLSEQUENCE(EXTRACT(XMLTYPE(:other_xml), '/*/outline_data/hint'))) d) LOOP l_hint := i.hint; WHILE NVL(LENGTH(l_hint), 0) > 0 LOOP IF LENGTH(l_hint) <= 500 THEN add_hint(l_hint); l_hint := NULL; ELSE l_pos := INSTR(SUBSTR(l_hint, 1, 500), ' ', -1); add_hint(SUBSTR(l_hint, 1, l_pos)); l_hint := ' '||SUBSTR(l_hint, l_pos); END IF; END LOOP; END LOOP; add_hint('END_OUTLINE_DATA'); :signature := DBMS_SQLTUNE.SQLTEXT_TO_SIGNATURE(:sql_text); :name := UPPER(TRIM('&&original_sql_id.'))||'_'||TRIM('&&plan_hash_value.'); description := UPPER('original:'||TRIM('&&original_sql_id.')||' modified:'||TRIM('&&modified_sql_id.')||' phv:'||TRIM('&&plan_hash_value.')||' signature:'||:signature||' created by coe_load_sql_profile.sql'); -- create custom sql profile for original sql using plan from modified sql DBMS_SQLTUNE.IMPORT_SQL_PROFILE ( sql_text => :sql_text, -- original sql profile => h, -- plan from modified sql name => :name, description => description, category => 'DEFAULT', validate => TRUE, replace => TRUE, force_match => FALSE /* TRUE:FORCE (match even when different literals in SQL). FALSE:EXACT (similar to CURSOR_SHARING) */ ); -- drop sql profile staging table for original sql (if one exists) BEGIN DBMS_OUTPUT.PUT_LINE('dropping staging table "STGTAB_SQLPROF_'||UPPER(TRIM('&&original_sql_id.'))||'"'); EXECUTE IMMEDIATE 'DROP TABLE STGTAB_SQLPROF_'||UPPER(TRIM('&&original_sql_id.')); EXCEPTION WHEN OTHERS THEN DBMS_OUTPUT.PUT_LINE('staging table "STGTAB_SQLPROF_'||UPPER(TRIM('&&original_sql_id.'))||'" did not exist'); END; -- create sql profile staging table for original sql DBMS_OUTPUT.PUT_LINE('creating staging table "STGTAB_SQLPROF_'||UPPER(TRIM('&&original_sql_id.'))||'"'); DBMS_SQLTUNE.CREATE_STGTAB_SQLPROF ( table_name => 'STGTAB_SQLPROF_'||UPPER(TRIM('&&original_sql_id.')), schema_name => '&&connected_user.' ); -- packs new sql profile for original sql DBMS_OUTPUT.PUT_LINE('packaging new sql profile into staging table "STGTAB_SQLPROF_'||UPPER(TRIM('&&original_sql_id.'))||'"'); DBMS_SQLTUNE.PACK_STGTAB_SQLPROF ( profile_name => :name, staging_table_name => 'STGTAB_SQLPROF_'||UPPER(TRIM('&&original_sql_id.')), staging_schema_owner => '&&connected_user.' ); END; / -- profile_name COL profile_name NEW_V profile_name FOR A30; SELECT :name profile_name FROM DUAL; -- display details of new sql_profile SET ECHO ON; REM REM SQL Profile REM ~~~~~~~~~~~ REM SELECT signature, name, category, type, status FROM dba_sql_profiles WHERE name = :name; SELECT description FROM dba_sql_profiles WHERE name = :name; SET ECHO OFF; PRO PRO **************************************************************************** PRO * Enter &&connected_user. password to export staging table STGTAB_SQLPROF_&&original_sql_id. PRO **************************************************************************** HOS exp &&connected_user. tables=&&connected_user..STGTAB_SQLPROF_&&original_sql_id. file=STGTAB_SQLPROF_&&original_sql_id..dmp statistics=NONE indexes=N constraints=N grants=N triggers=N PRO PRO If you need to implement this Custom SQL Profile on a similar system, PRO import and unpack using these commands: PRO PRO imp &&connected_user. file=STGTAB_SQLPROF_&&original_sql_id..dmp tables=STGTAB_SQLPROF_&&original_sql_id. ignore=Y PRO PRO BEGIN PRO DBMS_SQLTUNE.UNPACK_STGTAB_SQLPROF ( PRO profile_name => '&&profile_name.', PRO replace => TRUE, PRO staging_table_name => 'STGTAB_SQLPROF_&&original_sql_id.', PRO staging_schema_owner => '&&connected_user.' );; PRO END;; PRO / PRO SPO OFF; HOS zip -m coe_load_sql_profile_&&original_sql_id. coe_load_sql_profile_&&original_sql_id..log STGTAB_SQLPROF_&&original_sql_id..dmp coe_load_sql_profile.log HOS zip -d coe_load_sql_profile_&&original_sql_id. coe_load_sql_profile.log WHENEVER SQLERROR CONTINUE; SET DEF ON TERM ON ECHO OFF FEED 6 VER ON HEA ON LIN 80 PAGES 14 LONG 80 LONGC 80 TRIMS OFF TI OFF TIMI OFF SERVEROUT OFF NUM 10 SQLP SQL>; SET SERVEROUT OFF; UNDEFINE 1 2 3 original_sql_id modified_sql_id plan_hash_value profile_name CL COL PRO PRO coe_load_sql_profile completed.
Results of problem handling:
Summary: I have dealt with similar oracle sql execution plan changes many times. They have the same business characteristics, that is, dml frequently changes the data in the business table. Like this case, it needs to be changed nearly 300,000 times at a time. .