- 最后登录
- 2023-8-16
- 在线时间
- 1686 小时
- 威望
- 2135
- 金钱
- 50532
- 注册时间
- 2011-10-12
- 阅读权限
- 200
- 帖子
- 5207
- 精华
- 39
- 积分
- 2135
- UID
- 2
|
4#
发表于 2012-5-7 16:44:32
抱歉 周末比较忙 没有回复 , 就这个AWR 报告看:
10.2.0.4 on
Elapsed: 60.10 (mins)
DB Time: 209.87 (mins)
AAS=3 有一定的负载
Event Waits Time(s) Avg Wait(ms) % Total Call Time Wait Class
CPU time 6,079 48.3
db file sequential read 2,816,902 1,812 1 14.4 User I/O
gc cr disk read 2,206,028 999 0 7.9 Cluster
gc cr block 3-way 1,070,956 780 1 6.2 Cluster
gc cr block 2-way 1,307,860 689 1 5.5 Cluster
Top 5中 cluster + User I/O 占了大头
db file sequential read avg wait = 1ms 处于正常水平
Avg global cache cr block flush time (ms): 0.8
Avg global cache current block pin time (ms): 1,353.6
Avg global cache current block send time (ms): 0.1
Global cache log flushes for current blocks served %: 1.9
Avg global cache current block flush time (ms): 95,830.2
current block的 平均redo flush time 很高 =95,830.2 ms
wait class 中 cluster wait占了大头:
Wait Class Waits %Time -outs Total Wait Time (s) Avg wait (ms) Waits /txn
Cluster 6,312,252 0.07 3,768 1 361.75
User I/O 3,241,096 0.00 3,019 1 185.75
Other 14,225,006 85.09 304 0 815.23
log file sync 16,046 0.00 29 2 0.92
log file parallel write 33,034 0.00 27 1 1.89
log file parallel write 的avg wait =27ms 说明 日志文件写的性能存在瓶颈
gcs log flush sync 38,652 18.71 26 1 2.22
gcs log flush sync 的avg wait=26ms
SQL ordered by Cluster Wait Time
Cluster Wait Time (s) CWT % of Elapsd Time Elapsed Time(s) CPU Time(s) Executions SQL Id SQL Module SQL Text
239.99 92.98 258.11 16.27 21,728 9qgtwh66xg6nz update seg$ set type#=:4, bloc...
8.97 58.84 15.25 1.58 496 64wq3fa5a77sm insert into seg$ (file#, block...
2.07 75.56 2.74 1.74 1,570 2ym6hhaq30r73 select type#, blocks, extents,...
update seg$ set type#=:4, blocks=:5, extents=:6, minexts=:7, maxexts=:8, extsize=:9, extpct=:10, user#=:11, iniexts=:12, lists=decode(:13, 65535, NULL, :13), groups=decode(:14, 65535, NULL, :14), cachehint=:15, hwmincr=:16, spare1=DECODE(:17, 0, NULL, :17), scanhint=:18 where ts#=:1 and file#=:2 and block#=:3
update seg$ 语句 耗费大量时间在 cluster wait上
引起update seg$的 可能是 大的索引重建 ALTER INDEX L xxxxxx04_lakesii REBUILD PARTITION P06_12 parallel(degree 6)
配合statistics信息看 table scans (long tables) 9,005 2.50 0.52
存在 每秒 2.5次的 大表扫描
为什么要 非online的重建 index partition ? 是否因为index 失效 导致了 table scan?
尝试不要cross instance的 parallel rebuild index partition 减少 update seg$造成的cluster wait。
同时 建议你确认为什么 log file 日志文件写的性能变得很差。 |
|