Re: DBMS_STATS.GATHER_SYSTEM_STATS samples only BADSTATS


The only think I can think of is that the N4 and N5 values are the single block read count and single block read time (in milliseconds), and the average single block read from your figures is:

   N4                            : 138598982
   N5                            : 53294180

   n5/n4 = 53294180 / 138598982
       =  0.3845 m/s

Possibly a time less than 1m/s is not allowed, and
leaves you with 'BADSTATS'.

You have the same problem with the multiblock reads
(n6 and n7).   It looks like you may be getting a lot of
data from a file-system buffer.  (Or possibly your system
has been up for so long that you are getting an odd
32-bit wrap on some counters).


Regards

Jonathan Lewis

http://www.jlcomp.demon.co.uk/faq/ind_faq.html
The Co-operative Oracle Users' FAQ

http://www.jlcomp.demon.co.uk/cbo_book/ind_book.html
Cost Based Oracle: Fundamentals

http://www.jlcomp.demon.co.uk/appearances.html
Public Appearances - schedule updated 29th Nov 2005

----- Original Message ----- From: "Radoulov, Dimitre" <cichomitiko@xxxxxxxxx>
To: "Jonathan Lewis" <jonathan@xxxxxxxxxxxxxxxxxx>
Cc: <oracle-l@xxxxxxxxxxxxx>
Sent: Monday, December 12, 2005 10:56 PM
Subject: Re: DBMS_STATS.GATHER_SYSTEM_STATS samples only BADSTATS



What do you get from
   select * from sys.aux_stats$

I gathered the stats in a separate stats table, so:


STATID : OLTP TYPE : S VERSION : 4 FLAGS : 1 C1 : BADSTATS C2 : 12-09-2005 13:24 C3 : 12-10-2005 10:01 C4 : CPU_SERIO C5 : N1 : -1 N2 : -1 N3 : 588 N4 : 138598982 N5 : 53294180 N6 : 160604371 N7 : 46108340 N8 : 243873588 N9 : 421111191 N10 : 0 N11 : 15 N12 : 2619857577 D1 : R1 : R2 : CH1 : ----------------- STATID : OLTP TYPE : S VERSION : 4 FLAGS : 1 C1 : C2 : C3 : C4 : PARIO C5 : N1 : -1 N2 : -1 N3 : N4 : N5 : N6 : N7 : N8 : N9 : N10 : N11 : N12 : D1 : R1 : R2 : CH1 : -----------------

PL/SQL procedure successfully completed.





Best regards,
Dimitre Radoulov




----- Original Message ----- From: "Jonathan Lewis" <jonathan@xxxxxxxxxxxxxxxxxx>
To: <cichomitiko@xxxxxxxxx>; <oracle-l@xxxxxxxxxxxxx>
Sent: Monday, December 12, 2005 11:36 PM
Subject: Re: DBMS_STATS.GATHER_SYSTEM_STATS samples only BADSTATS





As far as I know, there is only one thing that
actually results in BADSTATS, and that's an
absence of any single block reads in the time
period.  (But there may be other oddities that
could be version dependent - like multiblock
read times being faster than single block read
times).

What do you get from
   select * from sys.aux_stats$

Regards

Jonathan Lewis

http://www.jlcomp.demon.co.uk/faq/ind_faq.html
The Co-operative Oracle Users' FAQ

http://www.jlcomp.demon.co.uk/cbo_book/ind_book.html
Cost Based Oracle: Fundamentals

http://www.jlcomp.demon.co.uk/appearances.html
Public Appearances - schedule updated 29th Nov 2005

----- Original Message ----- From: "Radoulov, Dimitre" <cichomitiko@xxxxxxxxx>
To: <oracle-l@xxxxxxxxxxxxx>
Sent: Wednesday, December 07, 2005 9:24 PM
Subject: DBMS_STATS.GATHER_SYSTEM_STATS samples only BADSTATS



The same 9.2.0.4 instance on Solaris 8 (do you remember the "hash join vs nested loops" thread :)) Trying to gather system statistics in a period with common workload, CPU load 30-70%, only one instance on this node. I get only BADSTATS.

Metalink says that we have BADSTATS when:

  => collection is completed but not pertinent, because there was no
workload
     that  could justify any statistics

  => start the collection again when the workload is relevant.

Tried with interval 5 min,15 min, 1h, 2h, 8h, and I get only BADSTATS.
Tried auto(interval)and manual(start/stop) mode, always BADSTATS.


I'll try with an interval > 8h, but, meanwhile, any suggestion is welcome.



Regards,
Dimitre Radoulov







--
http://www.freelists.org/webpage/oracle-l


Other related posts: