[20200808]優化的困惑10.txt
[20200808]優化的困惑10.txt
--//前一段時間遇到的問題,連結http://www.itpub.net/thread-2137868-1-1.html,CPU使用率100%.
--//我下載對方提供的awr報表,順便說一下.對方提供的awr html檔案無法正常跳轉.必須使用vim執行如下:
--// :%s+http://172.19.11.136/awr/dongguan/DGMESDB-mesdb/202007/mesdb_73613_73614_202007220700_202007220800.html++
--//很容易定位裡面的問題,邏輯讀太高,消耗大量CPU資源,而裡面sql語句使用rule提示.
--//但是我開始看時還是遇到1點點疑惑.
--//注:資料庫版本是10.2.0.5.0,作業系統:windows 64位版本的,實際上我個人建議不要在windows下跑資料庫,特別在生產環境,無形增
--//加問題定位的難度.
1.DB Time問題:
Snap Id Snap Time Sessions Cursors/Session
Begin Snap: 73613 22-Jul-20 07:00:36 206 24.9
End Snap: 73614 22-Jul-20 08:00:16 212 23.2
Elapsed: 59.66 (mins)
DB Time: 1,070.16 (mins)
--//取樣1個小時,而DB Time達到1070/60 = 17.83 小時.而實際的CPU數量是8.:
Operating System Statistics
Statistic Total
AVG_BUSY_TIME 354,972
AVG_IDLE_TIME 2,432
AVG_SYS_TIME 4,388
AVG_USER_TIME 350,487
BUSY_TIME 2,840,695
IDLE_TIME 19,653
SYS_TIME 36,011
USER_TIME 2,804,684
RSRC_MGR_CPU_WAIT_TIME 0
VM_IN_BYTES -1.5E+19
VM_OUT_BYTES 2.1E+16
PHYSICAL_MEMORY_BYTES 85,862,932,480
NUM_CPUS 8
NUM_CPU_CORES 8
--//實際的CPU數量僅僅8個,感覺不大可能DB time大於8小時.
--//再回過頭看.
Top 5 Timed Events
Event Waits Time(s) Avg Wait(ms) % Total Call Time Wait Class
CPU time 29,020 45.2
latch: cache buffers chains 101,977 4,455 44 6.9 Concurrency
db file sequential read 261,308 3,864 15 6.0 User I/O
log file sync 124,032 3,540 29 5.5 Commit
direct path read temp 38,162 1,321 35 2.1 User I/O
--//注意觀察列 Total Call Time,CPU time僅僅佔45.2. 這些前5名相加僅僅45.2+6.9+6.0+5.5+2.1 = 65.7.
--//還有什麼其它事件佔3X%呢?
--//另外latch: cache buffers chains 的Avg Wait(ms) 達到 44ms,真心覺得有一些誇張...CPU已經消耗殆盡了.
--//順便說一下,我以前學習oracle一直認為CPU資源很難消耗100%.除非人為"破壞".直到隨著學習的深入...
2.繼續:
--//再看SQL ordered by Elapsed Time部分.
SQL ordered by Elapsed Time
Resources reported for PL/SQL code includes the resources used by all SQL statements called by the code.
% Total DB Time is the Elapsed Time of the SQL statement divided into the Total Database Time multiplied by 100
Total DB Time (s): 64,209
Captured SQL account for 55.0% of Total
--//注意Resources reported for PL/SQL code includes the resources used by all SQL statements called by the code.
Elapsed Time (s) CPU Time (s) Executions Elap per Exec (s) % Total DB Time SQL Id SQL Module SQL Text
6,346 2,655 164,771 0.04 9.88 gyc05t01rz49h TGS2.exe declare p_owner varchar(30) :=...
6,110 2,551 164,791 0.04 9.52 0tpg5r73s1146 TGS2.exe SELECT /*+ RULE */ COUNT(*) FR...
5,389 2,601 164,835 0.03 8.39 8cd1y3p3rnsys TGS2.exe declare p_owner varchar(30) :=...
5,316 2,563 164,829 0.03 8.28 6x6aknt5p0uqd TGS2.exe SELECT /*+ RULE */ ARGUMENT_NA...
--//一定是sql_id=gyc05t01rz49h裡面呼叫sql_id=0tpg5r73s1146語句.
--//sql_id=gyc05t01rz49h格式化如下:
DECLARE
p_owner VARCHAR (30) := :owner;
p_object_name VARCHAR (30) := :object_name;
p_proc_name VARCHAR (30) := :procedure_name;
p_overload NUMBER := :overload;
p_object_type VARCHAR (19);
p_count NUMBER;
BEGIN
IF p_owner IS NULL
THEN
BEGIN
SELECT /*+ RULE */
object_type, owner
INTO p_object_type, p_owner
FROM sys.all_objects
WHERE owner = SYS_CONTEXT ('USERENV', 'CURRENT_SCHEMA')
AND object_type IN
('PROCEDURE'
,'FUNCTION'
,'PACKAGE'
,'PACKAGE BODY'
,'SYNONYM')
AND object_name = p_object_name
AND ROWNUM <= 1;
EXCEPTION
WHEN NO_DATA_FOUND
THEN
BEGIN
SELECT /*+ RULE */
object_type, owner
INTO p_object_type, p_owner
FROM sys.all_objects
WHERE owner = 'PUBLIC'
AND object_type IN
('PROCEDURE'
,'FUNCTION'
,'PACKAGE'
,'PACKAGE BODY'
,'SYNONYM')
AND object_name = p_object_name
AND ROWNUM <= 1;
EXCEPTION
WHEN NO_DATA_FOUND
THEN
IF p_proc_name IS NOT NULL
THEN
p_owner := p_object_name;
p_object_name := p_proc_name;
p_proc_name := NULL;
ELSE
RAISE;
END IF;
END;
END;
END IF;
IF p_object_type IS NULL
THEN
SELECT /*+ RULE */
object_type
INTO p_object_type
FROM sys.all_objects
WHERE object_type IN
('PROCEDURE'
,'FUNCTION'
,'PACKAGE'
,'PACKAGE BODY'
,'SYNONYM')
AND owner = p_owner
AND object_name = p_object_name
AND ROWNUM <= 1;
END IF;
WHILE p_object_type = 'SYNONYM'
LOOP
SELECT /*+ RULE */
table_owner, table_name
INTO p_owner, p_object_name
FROM sys.all_synonyms
WHERE owner = p_owner AND synonym_name = p_object_name;
SELECT /*+ RULE */
object_type
INTO p_object_type
FROM sys.all_objects
WHERE owner = p_owner AND object_name = p_object_name AND ROWNUM <= 1;
END LOOP;
SELECT /*+ RULE */
COUNT (*)
INTO p_count
FROM sys.all_procedures
WHERE owner = p_owner
AND object_name = p_object_name
AND ( procedure_name = p_proc_name
OR p_proc_name IS NULL AND procedure_name IS NULL);
IF p_count < p_overload
THEN
RAISE NO_DATA_FOUND;
END IF;
:owner := p_owner;
:object_name := p_object_name;
:procedure_name := p_proc_name;
SELECT /*+ RULE */
COUNT (*)
INTO :param_count
FROM sys.all_arguments
WHERE owner = p_owner
AND ( p_proc_name IS NULL
AND package_name IS NULL
AND object_name = p_object_name
OR p_proc_name IS NOT NULL
AND package_name = p_object_name
AND object_name = p_proc_name
AND NVL (overload, 1) = p_overload)
AND data_type IS NOT NULL;
END;
--//sql_id = 0tpg5r73s1146.
SELECT /*+ RULE */
COUNT (*)
FROM SYS.ALL_ARGUMENTS
WHERE OWNER = :B4
AND ( :B2 IS NULL AND PACKAGE_NAME IS NULL AND OBJECT_NAME = :B3
OR :B2 IS NOT NULL
AND PACKAGE_NAME = :B3
AND OBJECT_NAME = :B2
AND NVL (OVERLOAD, 1) = :B1)
AND DATA_TYPE IS NOT NULL;
--//仔細比較可以發現sql_id = 0tpg5r73s1146正是前面儲存過程最後要執行的語句.
--//我有點奇怪的是使用rule提示.這些儲存過程是開發寫的嗎?還是程式後臺自動生成執行的,這個也是我的疑問.
--//這樣就很好理解我為什麼看到db time很高了,db time的執行時間被重複計算了.
--//再看SQL ordered by Gets部分.
SQL ordered by Gets
Resources reported for PL/SQL code includes the resources used by all SQL statements called by the code.
Total Buffer Gets: 8,982,272,506
Captured SQL account for 29.5% of Total
Buffer Gets Executions Gets per Exec %Total CPU Time (s) Elapsed Time (s) SQL Id SQL Module SQL Text
945,581,260 164,771 5,738.76 10.53 2655.28 6345.51 gyc05t01rz49h TGS2.exe declare p_owner varchar(30) :=...
940,649,793 164,835 5,706.61 10.47 2600.58 5389.32 8cd1y3p3rnsys TGS2.exe declare p_owner varchar(30) :=...
940,453,081 164,791 5,706.94 10.47 2550.73 6109.69 0tpg5r73s1146 TGS2.exe SELECT /*+ RULE */ COUNT(*) FR...
940,296,517 164,829 5,704.68 10.47 2562.72 5316.13 6x6aknt5p0uqd TGS2.exe SELECT /*+ RULE */ ARGUMENT_NA...
--//可以相互驗證.其實每次並不是很多,僅僅57XX 邏輯讀,但是執行頻率非常高.
--//164771/3600 = 45.76次/秒.
3.解決方案:
--//設定 alter system set "_optimizer_ignore_hints"=true看看.
SYS@test> alter system set "_optimizer_ignore_hints"=true scope=memory;
System altered.
--//也可以在session 設定"_optimizer_ignore_hints"=true;引數. 通過登陸觸發器設定.
--//分析資料字典以及固定物件.
execute sys.dbms_stats.GATHER_FIXED_OBJECTS_STATS()
execute sys.dbms_stats.GATHER_DICTIONARY_STATS()
--//可惜提出解決方案後,樓主一直沒給出反饋建議.也不知道我分析是否正確.
--//還有1個非常無賴的方案就是玩一點點欺騙,改名檢視sys.all_arguments,然後建立一張真實的表代替.
--//sys.all_arguments裡面實際上記錄呼叫函式,儲存過程的引數.
--//還有我記憶遇到程式大量呼叫SYS.ALL_ARGUMENTS是使用deliph程式開發的.
--//不過我以前遇到的情況不同,執行語句裡面沒owner,而是直接寫ALL_ARGUMENTS,這樣更加簡單,我直接在使用者的schema下建議一張
--//ALL_ARGUMENTS表,來欺騙應用程式,風險就是如果改儲存過程帶入引數發生變化,ALL_ARGUMENTS也要同步更新.不知道不更新情況會怎
--//樣??
--//當然解決完上述問題,還可能需要優化剩下的sql語句,不過lz沒反饋,只能分析到這裡了.比如:
a4g0jrkkr3ggn
--//格式化:
SELECT COUNT (*) AS fail_QTY, T1.WORK_ORDER, T2.DEFECT_DESC
FROM (SELECT a.recid
,a.DEFECT_id
,A.WORK_ORDER
,a.SERIAL_NUMBER
,a.rec_time
,B.Defect_Code
,ROW_NUMBER () OVER ( PARTITION BY A.WORK_ORDER, A.SERIAL_NUMBER ORDER BY A.rec_time ASC, B.Defect_Code DESC)
AS rowindex
FROM sajet.G_SN_DEFECT a, sajet.sys_defect B
WHERE A.PROCESS_ID = :ProcessId
AND A.WORK_ORDER = :WO
AND a.defect_id = B.defect_id
AND a.SERIAL_NUMBER IN
(SELECT SERIAL_NUMBER
FROM (SELECT A.WORK_ORDER
,a.SERIAL_NUMBER
,a.CURRENT_STATUS
,a.OUT_PROCESS_TIME
,RANK () OVER ( PARTITION BY A.WORK_ORDER ,A.SERIAL_NUMBER ORDER BY a.OUT_PROCESS_TIME)
AS rowindex
FROM sajet.G_SN_TRAVEL a
WHERE A.WORK_ORDER = :WO
AND A.PROCESS_ID = :ProcessId) b
WHERE b.CURRENT_STATUS = 1
AND b.rowindex = 1
AND TO_CHAR ( b.OUT_PROCESS_TIME ,'yyyymmddHH24') BETWEEN :STime AND :ETime)) T1
,sajet.SYS_DEFECT T2
WHERE T1.rowindex = 1 AND T1.DEFECT_id = T2.DEFECT_id
GROUP BY T1.WORK_ORDER, T2.DEFECT_DESC
ORDER BY COUNT (*) DESC;
--//使用TO_CHAR函式,以及內層IN使用分析函式是否多餘,理論講只要exists應該就可以了.看看是否有優化餘地.
4.繼續分析:
--//看看Load Profile部分:
Load Profile
Per Second Per Transaction
Redo size: 338,989.88 2,438.14
Logical reads: 2,509,093.86 18,046.33
Block changes: 3,404.38 24.49
Physical reads: 284.63 2.05
Physical writes: 257.32 1.85
User calls: 975.01 7.01
Parses: 925.49 6.66
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Hard parses: 33.63 0.24
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Sorts: 940.09 6.76
Logons: 0.64 0.00
Executes: 10,063.58 72.38
Transactions: 139.04
--//硬分析不是很大.但是在SQL ordered by CPU Time部分出現:
CPU Time (s) Elapsed Time (s) Executions CPU per Exec (s) % Total % Total DB Time SQL Id SQL Module SQL Text
2,655 6,346 164,771 0.02 9.15 9.88 gyc05t01rz49h TGS2.exe declare p_owner varchar(30) :=...
2,601 5,389 164,835 0.02 8.96 8.39 8cd1y3p3rnsys TGS2.exe declare p_owner varchar(30) :=...
2,563 5,316 164,829 0.02 8.83 8.28 6x6aknt5p0uqd TGS2.exe SELECT /*+ RULE */ ARGUMENT_NA...
2,551 6,110 164,791 0.02 8.79 9.52 0tpg5r73s1146 TGS2.exe SELECT /*+ RULE */ COUNT(*) FR...
848 984 30 28.26 2.92 1.53 67086ft5pmkbn DECLARE job BINARY_INTEGER := ...
582 1,174 6,607 0.09 2.01 1.83 04t2q2nf3pd4b APService.exe begin SAJET.SJ_CKRT_INPUTMATER...
579 1,165 6,607 0.09 1.99 1.81 cca7qc96c2k2h APService.exe SELECT COUNT (DISTINCT STATION...
293 736 128,775 0.00 1.01 1.15 7dupqruy76j0u TGS2.exe BEGIN COMMAND_CODE(:TSAJET1, :...
213 380 690 0.31 0.73 0.59 gdb3vjjk0m4xs APService.exe select max(END_TIME) from SAJE...
211 392 690 0.31 0.73 0.61 88jc1wztmqbky APService.exe select min(END_TIME) from SAJE...
--//sql_id=gdb3vjjk0m4xs
gdb3vjjk0m4xs select max(END_TIME) from SAJET.G_WH_HR A where A.WORK_ORDER= '20405946' AND PROCESS_ID='100085' AND SHIFT_ID IN ( 10000017) AND SHIFT_DATE ='2020/07/21' AND PDLINE_ID = 10075
SELECT max(END_TIME)
FROM SAJET.G_WH_HR A
WHERE A.WORK_ORDER = '20405946'
AND PROCESS_ID = '100085'
AND SHIFT_ID IN ( 10000017)
AND SHIFT_DATE = '2020/07/21'
AND PDLINE_ID = 10075
--//沒有使用繫結變數,應該可以有優化餘地.
88jc1wztmqbky select min(END_TIME) from SAJET.G_WH_HR A where A.WORK_ORDER= '20405946' AND PROCESS_ID='100085' AND SHIFT_ID IN ( 10000017) AND SHIFT_DATE ='2020/07/21' AND PDLINE_ID = 10075
SELECT min(END_TIME)
FROM SAJET.G_WH_HR A
WHERE A.WORK_ORDER = '20405946'
AND PROCESS_ID = '100085'
AND SHIFT_ID IN ( 10000017)
AND SHIFT_DATE = '2020/07/21'
AND PDLINE_ID = 10075
--//沒有使用繫結變數,應該可以有優化餘地.我估計程式應該有很多類似的語句.awr報表僅僅出現2次.
來自 “ ITPUB部落格 ” ,連結:http://blog.itpub.net/267265/viewspace-2710302/,如需轉載,請註明出處,否則將追究法律責任。
相關文章
- [20210203]max優化的困惑.txt優化
- [20220507]優化的困惑13.txt優化
- [20220428]優化的困惑12.txt優化
- [20200408]優化的困惑6.txt優化
- [20200401]優化的困惑5.txt優化
- 優必選的商業化困惑
- [20201224]sql優化困惑.txtSQL優化
- [20200320]SQL語句優化的困惑.txtSQL優化
- [20200324]SQL語句優化的困惑2.txtSQL優化
- [20230511]最佳化的困惑17.txt
- [20230512]最佳化的困惑19.txt
- [20200422]最佳化的困惑9.txt
- strtotime 的困惑
- 困惑度的計算
- 學習java的困惑Java
- [20211207]變態的windows批處理10.txtWindows
- [20221030]繫結變數的分配長度10.txt變數
- 令人困惑的 TensorFlow!(II)
- 個人部落格的困惑
- [20180612]函式與標量子查詢10.txt函式
- 【原創】多專案控制的困惑
- MySQL優化(1)——–常用的優化步驟MySql優化
- hive的優化Hive優化
- 關於非同步爬蟲排序的困惑非同步爬蟲排序
- 前端效能優化(JS/CSS優化,SEO優化)前端優化JSCSS
- 數值最優化—優化問題的解(二)優化
- Android效能優化——效能優化的難題總結Android優化
- 如何消除ERP選型困惑
- Flutter的效能優化Flutter優化
- 優化你的 PhpStorm優化PHPORM
- Hive --------- hive 的優化Hive優化
- nginx的location優化Nginx優化
- java Synchronized的優化Javasynchronized優化
- Oracle優化的方法Oracle優化
- 設計模式系列 · 無從下手的困惑 (一)設計模式
- 資料庫優化 - SQL優化資料庫優化SQL
- Android 效能優化 ---- 啟動優化Android優化
- Android效能優化----卡頓優化Android優化