Oracle cached sequences

When dealing with Oracle sequences it is well known that cached sequences values are not lost following a tidy shutdown of a single database instance whereas a brut shutdown will generate a loss of sequences values. I did the experiment under Oracle-Linux and Oracle-Windows and results are shown below:

1.Oracle-Linux Fedora 16

SQL> create sequence mho_seq;

Sequence created.

SQL> select sequence_name, cache_size from all_sequences where sequence_name = 'MHO_SEQ';

SEQUENCE_NAME                  CACHE_SIZE
------------------------------ ----------
MHO_SEQ                                20

SQL> select mho_seq.nextval from dual;

   NEXTVAL
----------
         1

SQL> select mho_seq.nextval from dual;

   NEXTVAL
----------
         2

SQL> select mho_seq.nextval from dual;

   NEXTVAL
----------
         3

SQL> shutdown
Database closed.
Database dismounted.
ORACLE instance shut down.
SQL> startup
ORACLE instance started.

Total System Global Area 1673965568 bytes
Fixed Size                  1336932 bytes
Variable Size            1090521500 bytes
Database Buffers          570425344 bytes
Redo Buffers               11681792 bytes
Database mounted.
Database opened.
SQL> select mho_seq.nextval from dual;

   NEXTVAL
----------
         4

After a normal shutdown we didn’t loss any sequence value. Let’s now stop the database abruptly. One way to do this is to kill an internal Oracle process like SMON for example

[oracle@localhost mho]$ ps -ef | grep smon
oracle    4470     1  0 08:54 ?        00:00:00 ora_smon_DB11G
oracle    4597  4541  0 08:55 pts/3    00:00:00 grep --color=auto smon
[oracle@localhost mho]$ kill -9 4470
[oracle@localhost mho]$

SQL> select mho_seq.nextval from dual;
select mho_seq.nextval from dual
*
ERROR at line 1:
ORA-03135: connection lost contact
Process ID: 4516
Session ID: 125 Serial number: 5

SQL> startup
ORACLE instance started.

Total System Global Area 1673965568 bytes
Fixed Size                  1336932 bytes
Variable Size            1090521500 bytes
Database Buffers          570425344 bytes
Redo Buffers               11681792 bytes
Database mounted.
Database opened.
SQL> select mho_seq.nextval from dual;

   NEXTVAL
----------
        24

And see now how we went from value 4 to value 24!. We lost 20 cached values.

2.Oracle-Windows

I was not going to write this because it brings nothing new per regard to what happened under Linux operation system but I thought it is worth saying few words on the brut manner I used to simulate the oracle database crash and its consequences. You may have already guessed that I used the famous and easy windows ‘emergency exit’: ctlr+alt+delete to kill the process named Oracle. So I did, and, unfortunately when I wanted to restart the database the following internal error kicks off:

mohamed: mhouri> startup
Instance ORACLE lancÚe.

Total System Global Area 535662592 bytes
Fixed Size 1375792 bytes
Variable Size 289407440 bytes
Database Buffers 239075328 bytes
Redo Buffers 5804032 bytes
Base de donnÚes montÚe.
ORA-00600: code d'erreur interne, arguments : [kcratr_nab_less_than_odr], [1], [1410], [10771], [11555], [], [], [],
[], [], [], []

Thanks to Nassyam Basha who pointed me to the this blog article, I have recovered up my Oracle-Windows database using the following steps;

mhouri> Startup mount ;
Instance ORACLE lancÚe.

Total System Global Area  535662592 bytes
Fixed Size                  1375792 bytes
Variable Size             289407440 bytes
Database Buffers          239075328 bytes
Redo Buffers                5804032 bytes
Base de donnÚes montÚe.

mhouri> Show parameter control_files

VALUE
-----------------------------------------------
C:\APP\MOHAMED\ORADATA\MHOURI\MHOURI\CONTROL01.CTL,
C:\APP\MOHAMED\FLASH_RECOVERY_AREA\MHOURI\ARCHIVELOG\MHOURI\CONTROL02.CTL

mhouri> select a.member,a.group#,b.status from v$logfile a ,v$log b where a.group#=b.group# and b.status='CURRENT';

MEMBER                                          GROUP# STATUS
----------------------------------------------- ------- -------
C:\APP\MOHAMED\ORADATA\MHOURI\MHOURI\REDO03.LOG  3      CURRENT

mhouri> Shutdown abort ;
Instance ORACLE arrÛtÚe.
mhouri> Startup mount ;
Instance ORACLE lancÚe.

Total System Global Area  535662592 bytes
Fixed Size                  1375792 bytes
Variable Size             289407440 bytes
Database Buffers          239075328 bytes
Redo Buffers                5804032 bytes
Base de donnÚes montÚe.

mhouri> recover database using backup controlfile until cancel ;
ORA-00279: changement 32458110 gÚnÚrÚ Ó 02/02/2013 17:00:53 requis pour thread 1
ORA-00289: suggestion :
C:\APP\MOHAMED\FLASH_RECOVERY_AREA\MHOURI\ARCHIVELOG\MHOURI\ARCHIVELOG\2013_02_03\O1_MF_1_1410_%U_.ARC
ORA-00280: le changement 32458110 pour le thread 1 se trouve au no de sÚquence 1410

Indiquer le journal : {<RET>=suggÚrÚ | nomfichier | AUTO | CANCEL}
C:\APP\MOHAMED\ORADATA\MHOURI\MHOURI\REDO03.LOG
Fichier journal appliquÚ.
RÚcupÚration aprÞs dÚfaillance matÚrielle terminÚe.
mhouri> Alter database open resetlogs ;

Base de donnÚes modifiÚe.

This finally brings my windows database up.

The bottom lines:

1. Loosing oracle sequence values is more an effect of stress on the shared pool and excessive rollbacks then anything else.

2. I went from proofing to myself that a cached sequence values are not lost following a tidy database shutdown to stuff related to recovering my database following an ora-6000 error