Re: Oracle Stats
From: joel garry <joel-garry_at_home.com>
Date: Mon, 18 May 2009 11:16:06 -0700 (PDT)
Message-ID: <d59bb2ca-58d5-46a2-9317-e044a2d9a462_at_k9g2000pra.googlegroups.com>
On May 18, 10:36 am, a..._at_unsu.com wrote:
> On May 17, 5:08 pm, ddf <orat..._at_msn.com> wrote:
>
>
>
> > On May 17, 12:51 am, Ind-dba <oraclear..._at_googlemail.com> wrote:
>
> > > On May 16, 2:30 am, "stevedhow..._at_gmail.com" <stevedhow..._at_gmail.com>
> > > wrote:
>
> > > > On May 15, 11:47 am, a..._at_unsu.com wrote:
>
> > > > > Statistic Total
> > > > > -------------------------------- --------------------
> > > > > BUSY_TIME 14,954,028
> > > > > IDLE_TIME 121,666,971
> > > > > NICE_TIME 0
> > > > > SYS_TIME 6,017,299
> > > > > USER_TIME 8,936,729
> > > > > LOAD 3
> > > > > RSRC_MGR_CPU_WAIT_TIME 0
> > > > >>>>>>>>>>>> PHYSICAL_MEMORY_BYTES 77,192 <<<<<<<<<<
> > > > > NUM_CPUS 16
> > > > > NUM_CPU_SOCKETS 4
>
> > > > This section is puzzling. How much RAM do you have?
>
> > > There is an issue with PHYSICAL_MEMORY_BYTES column in v$osstat view.
> > > This is well explained in following link.http://phlonx.com/blog/fred/index.php/2009/03/20/vosstat-anomaly/-Hidequoted text -
>
> > > - Show quoted text -
>
> > One person found this problem in one installation of Oracle, on Linux
> > RHEL 4 64-bit. One person. No mention of this on Metalink. One
> > instance of a condition doesn't constitute an 'issue'.
>
> > You should really do more research before you post such things.
>
> > David Fitzjarrell
>
> David,
>
> If I can add, documentation says that there is an automatic job that
> runs and keeps stats on tables with missing or stale stats. If this
> is true, what happened here? A table changed hundreds of time daily
> and no stats updated since 11/08??
>
> SQL> select table_name, to_char(last_analyzed,'dd-mon-YYYY
> hh24:mi:ss'),
> monitoring from user_tables where table_name = 'CUSTOMER_ORDER'; 2
>
> TABLE_NAME TO_CHAR(LAST_ANALYZE MON
> ------------------------------ -------------------- ---
> CUSTOMER_ORDER 02-nov-2008 01:05:30 YES
Date: Mon, 18 May 2009 11:16:06 -0700 (PDT)
Message-ID: <d59bb2ca-58d5-46a2-9317-e044a2d9a462_at_k9g2000pra.googlegroups.com>
On May 18, 10:36 am, a..._at_unsu.com wrote:
> On May 17, 5:08 pm, ddf <orat..._at_msn.com> wrote:
>
>
>
> > On May 17, 12:51 am, Ind-dba <oraclear..._at_googlemail.com> wrote:
>
> > > On May 16, 2:30 am, "stevedhow..._at_gmail.com" <stevedhow..._at_gmail.com>
> > > wrote:
>
> > > > On May 15, 11:47 am, a..._at_unsu.com wrote:
>
> > > > > Statistic Total
> > > > > -------------------------------- --------------------
> > > > > BUSY_TIME 14,954,028
> > > > > IDLE_TIME 121,666,971
> > > > > NICE_TIME 0
> > > > > SYS_TIME 6,017,299
> > > > > USER_TIME 8,936,729
> > > > > LOAD 3
> > > > > RSRC_MGR_CPU_WAIT_TIME 0
> > > > >>>>>>>>>>>> PHYSICAL_MEMORY_BYTES 77,192 <<<<<<<<<<
> > > > > NUM_CPUS 16
> > > > > NUM_CPU_SOCKETS 4
>
> > > > This section is puzzling. How much RAM do you have?
>
> > > There is an issue with PHYSICAL_MEMORY_BYTES column in v$osstat view.
> > > This is well explained in following link.http://phlonx.com/blog/fred/index.php/2009/03/20/vosstat-anomaly/-Hidequoted text -
>
> > > - Show quoted text -
>
> > One person found this problem in one installation of Oracle, on Linux
> > RHEL 4 64-bit. One person. No mention of this on Metalink. One
> > instance of a condition doesn't constitute an 'issue'.
>
> > You should really do more research before you post such things.
>
> > David Fitzjarrell
>
> David,
>
> If I can add, documentation says that there is an automatic job that
> runs and keeps stats on tables with missing or stale stats. If this
> is true, what happened here? A table changed hundreds of time daily
> and no stats updated since 11/08??
>
> SQL> select table_name, to_char(last_analyzed,'dd-mon-YYYY
> hh24:mi:ss'),
> monitoring from user_tables where table_name = 'CUSTOMER_ORDER'; 2
>
> TABLE_NAME TO_CHAR(LAST_ANALYZE MON
> ------------------------------ -------------------- ---
> CUSTOMER_ORDER 02-nov-2008 01:05:30 YES
Look at the definition again. It depends how big the table is, too. Also, if you lock stats, it won't be evaluated automatically.
jg
-- _at_home.com is bogus. http://www.darkreading.com/insiderthreat/security/privacy/showArticle.jhtml?articleID=217400831Received on Mon May 18 2009 - 13:16:06 CDT