I insisted on reading three sets of RAC AWR every day and summarized it.
Query the SQL statement of the log wait event:
9i: Select * from V $ event_name where name like '% log %'; (about 14)
10g: Select name, wait_class from V $ event_name where name like '% log %'; (about 35)
11G: Select name, wait_class from V $ event_name where name like '% log %'; (about 30)
The following are the most common examples:
I,Log File switch (archiving needed)
That is, when the log is switched to the target log group, archiving is not yet completed. You must wait.
Possible causes:Redo,ArchivePartitionPoor I/O performanceSlow archiving and writing, and unreasonable setting of the number of log groups
Solution: 1, Increase the size of the log group or log group members
2, Adjust archive log to a disk with high Io performance, such as storage
3, AdjustmentLog_archive_max_processesParameters
II,Log File switch (checkpoint incomplete)
Indicates that the dirty data protected by the log group has not been written to the target log group during log switching.
Possible causes:DbwnWriting is too slow,I/OProblems
Solution: 1, Add additionalDbwn
2, Increase the size of the log group or log group member
III,Log File sync
Possible causes: LgwrLow write efficiency,CommitToo frequent
Solution: 1, ImproveLgwrWrite efficiency, using disks with good Io Performance
2Batch submission,(Real-time online business operations with caution)
3, UseNologging/unreoverableOption ()
IV,Log File single write
Cause:Wait for the log file header to be updated
V,Log File parallel write
Cause:Waiting for concurrent writing to multiple log group members
VI,Log buffer space
Cause:Database-generated log ratioLgwrThe log writing speed is fast, or the log switching speed is too slow.
Solution: 1, IncreaseLog Buffer
2, DiskI/OBottlenecks
Note: The DB plan is a redo partition, and the hard disk occupied by the arch partition space should be RAID 1 as much as possible.
To be continued