ALTER SYSTEM FLUSH SHARED_POOL

I am currently using oracle 8.1.5.0 standard edition and created a new index. Now to start using this index on some of my constant scripts that are already cached I need to flush the cache. What would be the effect of using this command while there are other sessions active?? Would it effect these in anyway if I used the above command?? Thank you in advance.

I don't think that the sql that is currently running will be affected, but all new sql launched by the sessions will have to be hard parsed (there is no prior sql anymore in the shared_pool)
I don't think you will hace to flush your shared pool since by creating an index on a table you have broken the breakable parse lock on that table. Oracle uses this lock to allow the parsed, cached statement to be invalidated (flushed) in the shared pool if the lock is broken (in case of a drop or alter).
so by adding the index, you break the parse lock and all cached statements that are referring to that table are flushed.
Greetings
Freek
I am currently using oracle 8.1.5.0 standard edition and created a new index. Now to start using this index on some of my constant scripts that are already cached I need to flush the cache. What would be the effect of using this command while there are other sessions active?? Would it effect these in anyway if I used the above command?? Thank you in advance.

Similar Messages

Alter system flushed shared pool in RMAN backup

Hi,
I am trying to take RMAN backup of 11.2.0.1 Database in IBM AIX 6.1 server.
The RMAN is hanging .
Though the backup gets completed, The channels allocated doesnt get released and the RMAN gets hanging.
In earlier RMAN backup Scripts,
the DBA was using alter system flush shared pool in RMAN backup script and the backup was getting succesful.
Now my question is , is using ALTER SYSTEM FLUSH SHARED POOL have any performance impact on the database.
Regards,
TEJAS

TEJAS_DBA wrote:
Hi,
I am trying to take RMAN backup of 11.2.0.1 Database in IBM AIX 6.1 server.
The RMAN is hanging .
Though the backup gets completed, The channels allocated doesnt get released and the RMAN gets hanging.Are you setting the large pool? If you don't, then rman uses the shared pool. Read about tuning rman performance in the docs.
>
In earlier RMAN backup Scripts,
the DBA was using alter system flush shared pool in RMAN backup script and the backup was getting succesful.
Now my question is , is using ALTER SYSTEM FLUSH SHARED POOL have any performance impact on the database.Yes, you are allowing the components in there to be loaded in the random order of whatever is called first. This may have a good impact if you had some fragmentation in there, or it could be mildly bad if everything was well sorted, or it could be very bad if you are unlucky or have some pattern of invalidations or should be pinning something or who-knows-what. It generally is considered not a good thing to do as a habit. You wind up with [url http://tkyte.blogspot.com/2012/05/another-debugging-story.html]rainy Monday scenarios.
Edit: I notice there are some bugs, including very slow performance when using a catalog. When you say hang, how long are you waiting? Have you considered current patches?
Edited by: jgarry on Aug 8, 2012 11:09 AM

Flush Shared Pool without Alter System Priv

I am trying to device a way to flush the shared pool by a user who doesn't have ALTER SYSTEM priv. I tried creating a stored proc in user's schema who has DBA role
with the following:
execute immediate 'alter system flush shared_pool';
and then granted execute on this to a normal user without explicit alter system priv, but the execution fails and seems that I need to grant alter system explicitly to this user. Is there any other workaorund for this?
Thank You
Kevin

it works for me! V10.2.0.4 too.
SQL> set time on
12:11:24 SQL> @privs
12:11:28 SQL> SET TERM ON ECHO ON
12:11:28 SQL> SET SERVEROUTPUT ON
12:11:28 SQL> CREATE USER PTST IDENTIFIED BY PTST DEFAULT TABLESPACE USERS QUOTA UNLIMITED ON users TEMPORARY TABLESPACE TEMP;
User created.
12:11:28 SQL> grant create session to ptst;
Grant succeeded.
12:11:28 SQL> grant alter system to dbadmin;
Grant succeeded.
12:11:28 SQL> connect dbadmin/admindb
Connected.
12:11:28 SQL> set term on echo on
12:11:28 SQL> create or replace procedure flush_pool
12:11:28   2 as
12:11:28   3 begin
12:11:28   4 execute immediate 'alter system flush shared_pool';
12:11:28   5 end flush_pool;
12:11:28   6 /
Procedure created.
12:11:28 SQL> grant execute on flush_pool to ptst;
Grant succeeded.
12:11:28 SQL> connect ptst/ptst
Connected.
12:11:28 SQL> SET TERM ON ECHO ON
12:11:28 SQL> execute dbadmin.flush_pool;
PL/SQL procedure successfully completed.
12:11:29 SQL> connect / as sysdba
Connected.
12:11:29 SQL> drop user ptst;
User dropped.

How to get the date/time of flush shared_pool commands?

I think there´s some script, which in been runned with a privileged user, that is recurring flushing the shared_pool.
Is there a way to know the last date/time when an 'alter system flush shared_pool' was issued?
Or, better, how can I tell Oracle RDBMS to log these commands on alert.log?
Best regards,
Luis Santos

You can audit the alter system commands:
audit alter system;
You probably want to ensure the following two system parameters check the doc for your version:
audit_trail=db_extended
audit_sys_operations=true
They require a system bounce, but by default sys operations are not audited so you'll need the latter.
Query the dba_audit_trail.

Alter system set shared_pool_size extremely slow.

I am trying to increase my shared_pool_size from 60m to 200m by simply typing alter system set shared_pool_size=200m;
I did alter system flush shared_pool; before to clear out the current shared_pool.
I'm on 9i and my SGA_MAX_SIZE is 3g and I have plenty room to grow.
show sga
Total System Global Area 3222769016 bytes
Fixed Size 744824 bytes
Variable Size 2382364672 bytes
Database Buffers 838860800 bytes
Redo Buffers 798720 bytes
My event in v$session_wait is background parameter adjustment.
Any help would be appreciated. I know that I can bounce the server and let it read it from the init.ora. However, I'm trying to fix some current 4031s in my shared pool. The flush did fix them temporarily.

Are you sure your instance was in the state you thought it was ? This is what you reported from show sga:
Total System Global Area 3222769016 bytes
Fixed Size                    744824 bytes
Variable Size             2382364672 bytes
Database Buffers           838860800 bytes
Redo Buffers                  798720 bytesYou say your shared_pool_size is 60M and you want to grow it to 200M, but the Variable Size in the output above suggests that your shared pool had already grown to about 2.3 GB, so your command would have been trying to shrink it quite dramatically. (And 9i isn't very good at shrinking the shared pool because of the problems of pinned and "KEEP"ed objects in the library cache.
Things to check (from another session):
v$sga_current_resize_ops to see if your resize operation is still running, and what it's trying to do.
v$sga_resize_ops to see i f your resize operation completed some time ago - leaving your session stuck somewhere else. (There's a convenient query I wrote against this view published here: http://jonathanlewis.wordpress.com/2007/04/16/sga-resizing/ )
v$session_wait (for the resize session) to see if the session is waiting or working
v$sesstat (for the resize session) to see if it gives you any clues about current activity
Regards
Jonathan Lewis
http://jonathanlewis.wordpress.com
http://www.jlcomp.demon.co.uk

Flush shared_pool hang

During the production when I try to flush the shared_pool with the command alter system flush shared_pool it hang.
What should I do to resolve this problem?
Any help

yours scenario depicting a clear picture that yours shared pool areas is not utilizing properly ,this situation occurs usually when yours application code not using bind varabiles ,there would be lot of hard parsing which require more CPU power as well more memory (shared pool).
For a quick go through one workaround to identify what are the queries and piece of code are not using propelry the shared pool is by turining on trace at system wide level and then review this trace file using tkrpof which will show up you exactly the hard coded queries.
for a detailed tkprof utility you can go through this link ,if possibke paste the tkprof result here.
http://download.oracle.com/docs/cd/B19306_01/server.102/b14211/sqltrace.htm#PFGRF01020Khurram

Locked Objects while flushing shared_pool?

Hello guys,
a little understanding about flushing shared pool and the stored objects in it.
What happens to objects (SQL Statements, DDIC cache, and so) that are needed at this time for execution.
For example an user is executing a select statement on a table and its is running while i am executing "alter system flush shared_pool".
What happens to the parsed sql statment which is used at this time by the sql query... and what is happens to the row cache (ddic cache) of the table which is queried at this time?
For my understanding this objects have to be locked (latch) by the user that is executing the query... are they invalidated after the query ends or what happens?
I know that this are internals but maybe someone of you knows it.
Regards
Stefan

Objects in the shared pool can be both "locked" and "pinned"; locks and pins can be held in null, shared, or exclusive mode.
If you are holding a cursor open you are locking it (with an item from the structure x$kgllk); if you are executing a cursor you will also be pinning it (with an iterm from the structure x$kglpn). Roughly speaking, if you hold a lock you are an interested party and want to know if the object gets changed or eliminated; if you hold a pin you are using the current version of the object and it must not change or disappear until you have finished with it.
If you flush the shared pool, pinned objects cannot be removed from the shared pool and (due to some oddities which are probably not deliberate design) there may be some objects which are in "the same chain" as a pinned object which are accidentally protected because the pinned object is "in the way".
I believe the same rationale of 'objects in use are protected' applies to the dictionary cache (row cache) - and there may be some objects there which are permanently pinned anyway. Normally you would only have a dictionary cache object pinned whilst optimising a statement, though, which might be quite rare.
Regards
Jonathan Lewis
http://jonathanlewis.wordpress.com
http://www.jlcomp.demon.co.uk

ALTER SYSTEM/DATABASE acts upon?

Hi,
I have a question inside my head since a long time now, I did lot of googling, but did'nt find a precise answer.
All DMLs will be served by Dedicated/Shared server processes as per the configuration. Selects/Updates/Deletes/Inserts will have their execution plans generated to/used from the Library Cache of shared pool; will have their corresponding buffers put into the RLB and from there into redo log files. This is found every where on the web.
But what happens when we issue an "ALTER SYSTEM ... ... ..." or an "ALTER DATABASE ... ... ..."?
Does the user process adjust the SGA_TARGET/PGA_AGGREGATE_TARGET or it just signals MMAN?
Does it Flush Buffer Cache/Shared Pool or does it just signal the MMAN again? What happens inside?
Does it kill the sessions itself or just signal the PMON?
Which process starts up first and takes on when "STARTUP;" is issued?
Which process gears up to Close, Dismount, and Shut down the database stagewise in the mode requested by "SHUTDOWN ....;"?
Any link that discusses these and other such internal Oracle algorithms/implementations would be greatly helpful.
Thanks,
Aswin.

Hi,
ice_cold_aswin wrote:
Does it kill the sessions itself or just signal the PMON?
Which process starts up first and takes on when "STARTUP;" is issued?
Which process gears up to Close, Dismount, and Shut down the database stagewise in the mode requested by "SHUTDOWN ....;"?I think at least these can be answered by using strace. For example, starting up an instance:
[oracle@localhost ~]$ strace -o trc.txt -f -t -e trace=process sqlplus / as sysdba
SQL*Plus: Release 11.2.0.1.0 Production on Tue Mar 30 00:20:49 2010
Copyright (c) 1982, 2009, Oracle. All rights reserved.
Connected to an idle instance.
SQL> startup
ORACLE instance started.
Total System Global Area            535662592 bytes
Fixed Size                            1337720 bytes
Variable Size                       335545992 bytes
Database Buffers                    192937984 bytes
Redo Buffers                          5840896 bytes
Database mounted.
Database opened.
18924 00:20:49 execve("/u01/app/oracle/product/11.2.0/dbhome_1/bin/sqlplus", ["sqlplus", "/", "as", "sysdba"], [/* 40 vars */]) = 0
18924 00:20:49 clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0xb7f81708) = 18925
18925 00:20:49 execve("/u01/app/oracle/product/11.2.0/dbhome_1/bin/oracle", ["oracletest11g2", "(DESCRIPTION=(LOCAL=YES)(ADDRESS"], [/* 41 vars */]) = 0
18925 00:20:51 clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0xae2708) = 18926
18926 00:20:51 execve("/bin/sh", ["/bin/sh", "-c", "/bin/df -k /u01/app/oracle/produ"], [/* 42 vars */]) = 0
18926 00:20:51 execve("/bin/df", ["/bin/df", "-k", "/u01/app/oracle/product/11.2.0/d"], [/* 42 vars */]) = 0
18926 00:20:51 exit_group(0)            = ?
18925 00:20:51 waitpid(18926, NULL, 0) = 18926
18925 00:20:51 --- SIGCHLD (Child exited) @ 0 (0) ---
18925 00:20:52 clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0xae2708) = 18927
18927 00:20:52 clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0xae2708) = 18928
18927 00:20:52 exit_group(0)            = ?
18925 00:20:52 --- SIGCHLD (Child exited) @ 0 (0) ---
18925 00:20:52 waitpid(18927, [{WIFEXITED(s) && WEXITSTATUS(s) == 0}], 0) = 18927
18928 00:20:52 execve("/u01/app/oracle/product/11.2.0/dbhome_1/bin/oracle", ["ora_pmon_test11g2"], [/* 46 vars */]) = 0
18925 00:20:52 clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0xae2708) = 18929
18929 00:20:52 clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0xae2708) = 18930
18929 00:20:52 exit_group(0)            = ?
18925 00:20:52 --- SIGCHLD (Child exited) @ 0 (0) ---
18925 00:20:52 waitpid(18929, [{WIFEXITED(s) && WEXITSTATUS(s) == 0}], 0) = 18929
18930 00:20:52 execve("/u01/app/oracle/product/11.2.0/dbhome_1/bin/oracle", ["ora_vktm_test11g2"], [/* 46 vars */]) = 0
18930 00:20:52 clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0x4b3708) = 18931
...

Urgent-how to run 'alter system kill session 'sid, serial#' in form 6i ?

I want to write a procedure in Form 6i so that user can kill the session by herself.
I know kill session sql is 'alter system kill session 'sid, serial#'', however, I fould that I can only run it it sql plus screen, how can I run it in Form or in Stored procedure?
Urgent....Please!

try using Forms_ddl('alter system......'); in the forms. it will execute the dml statements in the form.
zaibi.

ALLOW A USER TO KILL A SESSION WITHOUT ALTER SYSTEM PRIVILEGE.

Hi
I need a user to have permission to kill a session without having the ALTER SYSTEM privilege. I created a procedure on sys schema and granted the EXECUTE privilege to the user but it doesn't work, how can I do, help please.
CREATE OR REPLACE PROCEDURE SYS.PRC_SESSION_KILLER (P_SID IN NUMBER, P_SERIAL IN NUMBER)
AS
BEGIN
     EXECUTE IMMEDIATE 'GRANT ALTER SYSTEM TO SYSADMIN';
     EXECUTE IMMEDIATE 'ALTER SYSTEM KILL SESSION ''' || P_SID || ',' || P_SERIAL || ''' IMMEDIATE';
     EXECUTE IMMEDIATE 'REVOKE ALTER SYSTEM FROM SYSADMIN';
END;
Thank you very much.

Hi,
I second everything John said.
Are you sure the arguments are correct?
Below is the procedure I use. You may want to run it, just to see what the error is.
PROCEDURE     kill_internal
     s_id          IN     NUMBER,
     serial_num     IN     NUMBER,
     stat_out     OUT     VARCHAR2
IS
     alter_handle     INTEGER;
     ex_val          INTEGER;     -- Returned by dbms_sql.execute
BEGIN
     alter_handle := dbms_sql.open_cursor;
     dbms_sql.parse
          alter_handle,
          'ALTER SYSTEM     KILL SESSION '''     ||
               TO_CHAR (s_id, '999990')     ||
               ', '                    ||
               TO_CHAR (serial_num, '999990')     ||
          dbms_sql.native
     ex_val := dbms_sql.execute (alter_handle);
     dbms_sql.close_cursor (alter_handle);
     stat_out := 'Success: '                    ||
               TO_CHAR (s_id, '999990')     ||
               ', '                    ||
               TO_CHAR (serial_num, '999990');
EXCEPTION
     WHEN OTHERS
     THEN
          stat_out := 'Failure:'          ||
               SQLERRM;
--          dbms_output.put_line (stat_out);
          dbms_sql.close_cursor (alter_handle);
END     kill_internal
;

"Alter system set command" in a RAC database!!

Hi, all.
The database is (10.2.0.2.0) 2- node RAC database on 32-bit windows 2003
EE SP1.
I issued the following command on Node 1 database.
--> Alter system set db_block_buffers= xxx sid='rac1';
I was able to see "PE enqueue" in top 5 wait event section from an AWR report.
In addition, I was able to find "PZ99","PZ98" process dump file in BDUMP.
Soon later, I could find CKPT and DBWR hung.
Is there anyone who experienced this issue?
Thanks and Regards.
Message was edited by:
user507290

10.2.0.2 has some bug which is fixed 10.2.0.3
You check sequence cache. If it has less value, increase it to 10000.
select CACHE_SIZE from dba_sequences where SEQUENCE_OWNER='SYS' and SEQUENCE_NAME='AUDSES$';
SQL> alter sequence sys.audses$ cache 10000;
Ashok

Question about ALTER SYSTEM ARCHIVE LOG START

Good morning,
I'm trying (unsuccessfully) to get my database to be in archive log mode.
These are the steps I followed:
SQL> shutdown immediate;
Database closed.
Database dismounted.
ORACLE instance shut down.
SQL> startup mount;
ORACLE instance started.
Total System Global Area 535662592 bytes
Fixed Size                  1375792 bytes
Variable Size             377487824 bytes
Database Buffers          150994944 bytes
Redo Buffers                5804032 bytes
Database mounted.
SQL> select log_mode from v$database;
LOG_MODE
ARCHIVELOG
SQL> show parameter log_archive_start;
NAME                                 TYPE        VALUE
log_archive_start                    boolean     FALSE
SQL> alter system archive log start;
System altered.
SQL> show parameter log_archive_start;
NAME                                 TYPE        VALUE
log_archive_start                    boolean     FALSE
SQL>I've gone thru that process twice but, I don't seem to be able to get the ARCH process to start. (newbie mistake I'm sure...)
Thank you for your help (again!),
John.

The parameter log_archive_start is no more needed John (as suggested already) and the best way to check the archive options is through the archive log list command.
[oracle@edhdr2p0-orcl oui]$ sqlplus / as sysdba
SQL*Plus: Release 11.2.0.1.0 Production on Tue Aug 10 10:51:57 2010
Copyright (c) 1982, 2009, Oracle. All rights reserved.
Connected to:
Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - Production
With the Partitioning, Automatic Storage Management, OLAP, Data Mining
and Real Application Testing options
SQL> archive log list
Database log mode              No Archive Mode
Automatic archival             Disabled
Archive destination            USE_DB_RECOVERY_FILE_DEST
Oldest online log sequence     11
Current log sequence           13
SQL> shut immediate
Database closed.
Database dismounted.
ORACLE instance shut down.
SQL> startup mount
ORACLE instance started.
Total System Global Area 418484224 bytes
Fixed Size                  1336932 bytes
Variable Size             318769564 bytes
Database Buffers           92274688 bytes
Redo Buffers                6103040 bytes
Database mounted.
SQL> select process,status from V$archive_processes;
   PROCESS STATUS
         0 STOPPED
         1 STOPPED
         2 STOPPED
         3 STOPPED
         4 STOPPED
         5 STOPPED
         6 STOPPED
         7 STOPPED
         8 STOPPED
         9 STOPPED
        10 STOPPED
   PROCESS STATUS
        11 STOPPED
        12 STOPPED
        13 STOPPED
        14 STOPPED
        15 STOPPED
        16 STOPPED
        17 STOPPED
        18 STOPPED
        19 STOPPED
        20 STOPPED
        21 STOPPED
   PROCESS STATUS
        22 STOPPED
        23 STOPPED
        24 STOPPED
        25 STOPPED
        26 STOPPED
        27 STOPPED
        28 STOPPED
        29 STOPPED
30 rows selected.
SQL>
SQL> alter database archivelog;
Database altered.
SQL> alter database open;
archive log list;
Database altered.
SQL> Database log mode         Archive Mode
Automatic archival             Enabled
Archive destination            USE_DB_RECOVERY_FILE_DEST
Oldest online log sequence     11
Next log sequence to archive   13
Current log sequence           13
SQL>
SQL> select * from V$archive_processes where status <> 'STOPPED';
   PROCESS STATUS     LOG_SEQUENCE STAT
         0 ACTIVE                0 IDLE
         1 ACTIVE                0 IDLE
         2 ACTIVE                0 IDLE
         3 ACTIVE                0 IDLEHTH
Aman....

Question on alter system set cluster_database = true

11.2.0.2.0 windows 2008
doing a restore to a development RAC I noticed the following. I took the database out of cluster mode on one instance
alter system set cluster_database=false scope=spfile sid='PRD1';
shutdown, start up, do the restore, all good, I go to set cluster_Database=true but had issue instance would only start on one node, I reviewed my notes and found problem was when I set cluster_Database back to true I had typed
alter system set cluster_database=true scope=spfile sid='*';
instead of
alter system set cluster_database=true scope=spfile sid='PRD1';
So issue fixed.
I retested and confirmed '*' wont take. Ive seen many examples of '*' working online and I am sure I have used '*' many times so had just typed it off my head. Anyone explain? probably quotes or something.

Any chance there are multiple lines for cluster_database in the spfile ?I had thought maybe this, but confirmed no, it was a fresh build anyway that I was restoring to.
>
- Dump the spfile to a pfile ... edit ... recreate the spfile on all nodes.
>
Im not sure what this will achieve? the spfile is shared on the ASM.
>
As the cluster_database=true is a parameter to enable the RAC configuration. As you typed sid='*' it will try to enable the cluster_database parameter in all the instances(nodes) as it is RAC.
In your case as you did on a single node by specifiying the sid name it will work as it will look only for that sid and enable that particular node.
>
I find this hard to read but I think this misinterprets as well.
Thanks for the contribution. I think Ive found the solution
the asterisk doesnt apparently mean actually "all instances", from here
http://www.comp.dit.ie/btierney/oracle11gdoc/rac.111/b28254/admin.htm#BCEFICEE
and
"Setting SPFILE Parameter Values for Oracle Real Application Clusters"
-- this is a good example of what happened to me
>
*.OPEN_CURSORS=500
prod1.OPEN_CURSORS=1000
Note:
The value before the dot in an SPFILE entry identifies the instance to which the particular parameter value belongs. When an asterisk precedes the dot, the value is applied to all instances that do not have a subsequent, individual value listed in the SPFILE.
For the instance with the Oracle system identifier (SID) prod1, the OPEN_CURSORS parameter is set to 1000 even though it has a database-wide setting of 500. Parameter file entries that have the asterisk (*) wildcard character only affect the instances without an instance-specific entry. This gives you control over parameter settings for instance prod1. These two types of settings can appear in any order in the parameter file.
If another DBA runs the following statement, then Oracle updates the setting on all instances except the instance with SID prod1:
ALTER SYSTEM SET OPEN_CURSORS=1500 sid='*' SCOPE=MEMORY;
>
so when I was on node 1 and ran it for *, I was setting it for all the nodes except the node I was on which holds true from what Ive seen. I'll remember that one.

Alter system set nls_length_semantics

Hi all,
my question concerns the scope in the change of NLS_LENGTH_SEMANTICS can be performed.
The 10gR2 documentation only the
"Modifiable ALTER SESSION"
But what about altering the system and making your own setting to default for all sessions? With which scope?
I tried
alter system set nls_length_semantics='CHAR';
alter system set nls_length_semantics='CHAR' scope=spfile;
alter system set nls_length_semantics='CHAR' scope=both;
None had really any effect. Do I have to bounce the database?

Hello,
Do I have to bounce the database?Yes, you have to shutdown and startup the database.
Else the NLS_LENGTH_SEMANTICS change won't be effective.
You may have more details on the following thread:
nls_database_parameters->nls_length_semantics Help!
There's also an interesting Note from MOS:
Examples and limits of BYTE and CHAR semantics usage (NLS_LENGTH_SEMANTICS) [ID 144808.1]They give many information about NLS_LENGTH_SEMANTICS and the following Bug:
Bug 1488174
Problem: ALTER SYSTEM does not change the setting of NLS_LENGTH_SEMANTICS for the current and new (!) sessions.
Workaround: Don't use ALTER SYSTEM SET NLS_LENGTH_SEMANTICS scope=both; but set NLS_LENGTH_SEMANTICS as a init.ora parameter or issue ALTER SYSTEM SET NLS_LENGTH_SEMANTICS=CHAR scope=spfile; and bounce the database.Hope this help.
Best regards,
Jean-Valentin
Edited by: Lubiez Jean-Valentin on May 27, 2010 2:06 PM

Query on ALTER SYSTEM SET

Hi,
I was under impression that
"PFILE has the limitation to change the initialization parameters dynamically.If any init parameter is to be changed,we can edit the pfile and the parameter changes will come into effect with the next reboot of the system.
To overcome this limitation,Oracle has come up with the concept of SPFILE ,where in the initialization parameters can be changed dynamically using ALTER SYSTEM SET statement."
But,today I started my test database using PFILE and executed the following command to change the memory_target parameter from 300M to 290M.
ALTER SYSTEM SET memory_target=290M;
And to my surprise,this statement executed fine without any errors and memory_target parameter has been set to the new value(of course,in memory).
Isn't it changing the init parameters dynamically using PFILE itself??
Please clarify my doubt and correct my understanding.
Regards,
Bharath

bharathDBA wrote:
Hi,
I was under impression that
"PFILE has the limitation to change the initialization parameters dynamically.If any init parameter is to be changed,we can edit the pfile and the parameter changes will come into effect with the next reboot of the system.
To overcome this limitation,Oracle has come up with the concept of SPFILE ,where in the initialization parameters can be changed dynamically using ALTER SYSTEM SET statement."
But,today I started my test database using PFILE and executed the following command to change the memory_target parameter from 300M to 290M.
ALTER SYSTEM SET memory_target=290M;
And to my surprise,this statement executed fine without any errors and memory_target parameter has been set to the new value(of course,in memory).
Isn't it changing the init parameters dynamically using PFILE itself??
Please clarify my doubt and correct my understanding.
Regards,
Bharath
You have a wrong interpenetration about the genesis of the SPFILE. Yes it is a binary file and all that but remember, its not the SPFILE that makes a parameter dynamic or a PFILE that makes it static but it's the very nature of the parameter itself that makes it happen. Yes, the difference in the PFILE and SPFILE parameter would be evident when you would change a dynamic parameter and would give a bounce to the db. Since teh PFILE wasn't updated (its a text file so has to be updated manually) , with the next restart, the parameter would take the same value as it has in the PFILE. But with the SPFILE, depending on what you have set in the SCOPE, the parameter would behave accordingly.
HTH
Aman....

ALTER SYSTEM FLUSH SHARED_POOL

Similar Messages

Maybe you are looking for