Comments for Dion Cho - Oracle Performance Storyteller

Comment on Spying on the other session by Dion Cho

Dion Cho — Wed, 02 Dec 2009 07:13:37 +0000

Hi, Addie.

As far as I know, we have no way to get the current parameters and diagnostic events of the specific session from the dynamic performance views. But as of optimizer-related parameters, you might be able to rely on v$ses_optimizer_env view.

Comment on Logging in PL/SQL by jametonog

jametonog — Wed, 02 Dec 2009 07:02:30 +0000

log to table have a problem..

It need to commit the log transaction then can we read the log. So sometimes,we may need to use autonomous transactions to do the table logging.

Comment on Spying on the other session by Addie

Addie — Tue, 01 Dec 2009 22:59:20 +0000

Dion,

This is very useful. Thank you for sharing!

Do you know of any v$ views that will expose the current session settings in 10G?

Thanks,
Addie

Comment on Join Predicate Pushing and Group By by latch: cache buffers chains и настройка запросов с использованием cardinality feedback « Oracle mechanics

Tue, 01 Dec 2009 12:58:19 +0000

[...] раз, не зависит и не меняет план своего выполнения (механизм join predicate push-down не работает для обзоров с GROUP BY в…) в зависимости от условий основного запроса (query predicates) [...]

Comment on Parameter3 of gc buffer busy/gc current request wait event by Dion Cho

Dion Cho — Fri, 27 Nov 2009 03:12:35 +0000

Johnwoo.
It seems to depend on how we define the “1st”.
Tanel’s script defines the 1st as 0.
My calculation defines the 1st as 1.

When the first undo segment is #0, 123 is the 53th undo segment.
When the first undo segment is #1, 123 is the 54th undo segment.

Comment on Parameter3 of gc buffer busy/gc current request wait event by johnwoo

johnwoo — Fri, 27 Nov 2009 02:58:20 +0000

Tanel gave me a sql:
select class, r undo_segment_id from (
select class, null r
from (select class, rownum r from v$waitstat)
where r = bitand(&1,to_number(‘FFFF’,'XXXX’))
union all
select
decode(mod(bitand(&1,to_number(‘FFFF’,'XXXX’)) – 17,2),0,’undo header’,1,’undo data’, ‘error’) type
, trunc((bitand(&1,to_number(‘FFFF’,'XXXX’)) – 17)/2) undoseg_id
from
dual
)
where rownum = 1
/
It seems like the undo starts with 17.So from tanel’s sql,i get the 53th undo segment.

Comment on Parameter3 of gc buffer busy/gc current request wait event by Dion Cho

Dion Cho — Thu, 26 Nov 2009 08:32:47 +0000

Hi, Johnwoo.
Tanel might have meant following things:

UKJA@ukja1021> select rownum, class from v$waitstat;
    ROWNUM CLASS
---------- ------------------
         1 data block
         2 sort block
         3 save undo block
         4 segment header
         5 save undo header
         6 free list
         7 extent map
         8 1st level bmb
         9 2nd level bmb
        10 3rd level bmb
        11 bitmap block
        12 bitmap index block
        13 file header block
        14 unused
        15 system undo header
        16 system undo block
        17 undo header
        18 undo block

Undo header = 2*r + 15
Undo block = 2*r + 16
r = 0 means system undo, r >= 1 means user undo. Except the system undo, it starts with 17. 17 is the first undo segment.
So 123 is the undo block header of the 54th undo segment. (2*54 + 15)

Comment on Parameter3 of gc buffer busy/gc current request wait event by johnwoo

johnwoo — Thu, 26 Nov 2009 08:09:36 +0000

Thank you very much.But I get the answer from tanel that class numbers starting from 17 are regular undo segments.So is it 17 or 15?

Comment on Parameter3 of gc buffer busy/gc current request wait event by Dion Cho

Dion Cho — Thu, 26 Nov 2009 01:18:30 +0000

Hi, Johnwoo.

The value of 65659 means the block class “123″, which represents the undo header of the 54th(2*54+15=123) undo segment. This would mean that your nodes are fighting each other to get the same undo header(by any reason).

Comment on Parameter3 of gc buffer busy/gc current request wait event by johnwoo

johnwoo — Wed, 25 Nov 2009 09:37:28 +0000

It’s a good stuff.I encounterd a similar case.
From the gc buffer busy waits’s p1 and p2,i get the object’s UNDO datafile.But the p3’s 65659 that meaning “bitmap block”.The Undo tablespace isn’t assm.I confuse what the undo segment’s ‘bitmap block’.