昨天整个系统慢,weblogic日志有大量stuck,数据库服务器IO负载高。取了数据库报告后发现了问题。
负载很高,平常DB Time最多1000.
Host Name | Platform | CPUs | Cores | Sockets | Memory (GB) |
---|---|---|---|---|---|
XXXXXX | AIX-Based Systems (64-bit) | 128 | 32 | 123.50 |
Snap Id | Snap Time | Sessions | Cursors/Session | Instances | |
---|---|---|---|---|---|
Begin Snap: | 8940 | 01-12月-15 13:59:29 | 1263 | 114.2 | 2 |
End Snap: | 8943 | 01-12月-15 16:19:14 | 1648 | 96.0 | 2 |
Elapsed: | 139.75 (mins) | ||||
DB Time: | 30,337.56 (mins) |
TOP等待事件都是IO
Top 10 Foreground Events by Total Wait Time
Event | Waits | Total Wait Time (sec) | Wait Avg(ms) | % DB time | Wait Class |
---|---|---|---|---|---|
db file sequential read | 19,395,915 | 512.6K | 26 | 28.2 | User I/O |
read by other session | 9,753,486 | 385.8K | 40 | 21.2 | User I/O |
direct path read temp | 4,205,461 | 310.7K | 74 | 17.1 | User I/O |
direct path read | 1,886,691 | 133.4K | 71 | 7.3 | User I/O |
DB CPU | 86.1K | 4.7 | |||
direct path write temp | 840,182 | 61.9K | 74 | 3.4 | User I/O |
gc buffer busy acquire | 3,103,568 | 58.2K | 19 | 3.2 | Cluster |
db file scattered read | 1,721,381 | 55.1K | 32 | 3.0 | User I/O |
db file parallel read | 859,955 | 33.5K | 39 | 1.8 | User I/O |
表XX_PURCHASE_CONTRACT_ITEM_MV物理读非常高,284623132*8/1024/1024=2171.5G的数据,在磁盘上读了这么多数据,当然,这个数字需要*2, 因为是RAC的数据库。
Segments by Physical Reads
- Total Physical Reads: 620,000,121
- Captured Segments account for 52.3% of Total
Owner | Tablespace Name | Object Name | Subobject Name | Obj. Type | Physical Reads | %Total |
---|---|---|---|---|---|---|
XXXXX | Www | XX_PURCHASE_CONTRACT_ITEM_MV | TABLE | 284,623,132 | 45.91 | |
XXXXX | Www | XX_PURCHASE_ITEM | P0308 | TABLE PARTITION | 7,707,543 | 1.24 |
XXXXX | Www | XX_PURCHASE_ITEM | P0314 | TABLE PARTITION | 6,211,667 | 1.00 |
XXXXX | Www | XX_PURCHASE_ITEM | P0313 | TABLE PARTITION | 5,502,512 | 0.89 |
XXXXX | Www | XX_PURCHASE_ITEM | P0318 | TABLE PARTITION | 3,820,804 | 0.62 |