Skip to Main Content

Oracle Database Discussions

Announcement

For appeals, questions and feedback about Oracle Forums, please email oracle-forums-moderators_us@oracle.com. Technical questions should be asked in the appropriate category. Thank you!

"local write wait" at awr top 10 event

YoavAug 4 2016 — edited Aug 4 2016

Hello,

On oracle 12.1.0.2 runing on oracle linux 6 connect to oracle ZFS3 storage we are constatly see "local write wait" on the top wait evet.

We have also notice that the whis wait event is come togther with truncate operations.

Awr report shows :

Top 10 Foreground Events by Total Wait Time

EventWaitsTotal Wait Time (sec)Wait Avg(ms)% DB timeWait Class
rdbms ipc reply2872414.68413.0726.1Other
free buffer waits167217613030.1623.6Configuration
local write wait1132033.817997.8622.0User I/O
write complete waits25942.937716.8210.2Configuration
enq: RO - contention7635.690800.076.9Application
row cache lock2507.2253622.165.5Concurrency
DB CPU 285.8 3.1
log file sync1,04583.980.31.9Commit
log buffer space15876.1481.74.8Configuration
db file sequential read15,135342.25.4

Also ASH report shows that the event "local write wait" is due to truncate operations:

Top SQL with Top Events

SQL IDPlan HashExecutions% ActivityEvent% EventTop Row Source% Row SourceSQL Text
8jdnp4xqz46jv7543346734.94local write wait3.05DDL STATEMENT3.05truncate table TMP_SP_REFRESHC...
rdbms ipc reply1.82DDL STATEMENT1.82
0ajcttwqv0q5h2518438744.65local write wait2.49DDL STATEMENT2.49TRUNCATE TABLE TMP_CDRS_FULL
rdbms ipc reply2.12DDL STATEMENT2.12
0kbpwn4f4pngq70098948024.35free buffer waits4.31LOAD TABLE CONVENTIONAL4.31INSERT INTO TMP_IMPORT_PLIST T...
89wa9tmwk3unf383741385643.35enq: RO - contention2.01DDL STATEMENT2.01truncate table D$IMPORT_DIALCO...
rdbms ipc reply1.15DDL STATEMENT1.15
7qdt98x5jqp8n170489939312.97local write wait1.97DDL STATEMENT1.97** SQL Text Not Available **

I have found some information here :

https://docs.oracle.com/cd/B16240_01/doc/doc.102/e16282/oracle_database_help/oracle_database_wait_bottlenecks_local_writ…

which explain that :

"The wait event can be caused by truncate operations. Truncate operations cause the DBWR to be posted to flush out the space header."

I have found that in 10g there was a bug related to truncate operation.

I have tried the suggested work around set the hidden parameter : _db_fast_obj_truncate=FALSE . It helped to reduce the wait on event  "enq: RO - fast object reuse" , but we still see high "local write wait"

Please advise how to handle such performance issue.

Thanks

Yoav

Comments
Locked Post
New comments cannot be posted to this locked post.
Post Details
Locked on Sep 1 2016
Added on Aug 4 2016
11 comments
3,458 views