Re: Flashback log file sync

"Mladen Gogala" <gogala.mladen_at_gmail.com> wrote in message news:pan.2012.05.12.19.03.30_at_gmail.com...
| On Sat, 12 May 2012 19:49:36 +0100, Jonathan Lewis wrote:
|
|
| > Did you check the session stats to see what your reads were ?
|
| I turned on 10046 tracing. Reads were from the TEST_LOB tablespace, the
| LOB segmentL
|
|

The 10046 trace tells us what they were, not why they happened. It's important to see (for example) things like "undo blocks read for flashback new".

|
| =126792 block cnt=4 obj#=-1 tim=1336849142292618
| WAIT #140145345192712: nam='direct path read' ela= 280 file number=7
| first dba=127180 block cnt=4 obj#=-1 tim=1336849142302190
| WAIT #140145345192712: nam='direct path read' ela= 359 file number=7
| first dba=126352 block cnt=4 obj#=-1 tim=1336849142303526
| WAIT #140145345192712: nam='direct path read' ela= 136 file number=7
| first dba=126416 block cnt=4 obj#=-1 tim=1336849142341101
| WAIT #140145345192712: nam='direct path read' ela= 38497 file number=7
| first dba=126480 block cnt=4 obj#=-1 tim=1336849142379976
| WAIT #140145345192712: nam='direct path read' ela= 122 file number=7
| first dba=126612 block cnt=4 obj#=0 tim=1336849142445429

Interesting that your reads are often about 64 blocks apart, even though waiting for 4 blocks.

-- 
Regards

Jonathan Lewis
http://jonathanlewis.wordpress.com
Oracle Core (Apress 2011)
http://www.apress.com/9781430239543