Ivan Kartik - Oracle and Linux Blog - - Fun: How is it possible?

Purpose of this post is more for fun than practice. Consider this output:

$ ps -aef|grep pmon|grep -v grep
oracle   25330     1  0 Apr16 ?        00:00:02 ora_pmon_RMANDELL
oracle   26084     1  0 Apr16 ?        00:00:01 ora_pmon_RMANDELL
oracle   26128     1  0 Apr16 ?        00:00:01 ora_pmon_RMANDELL
oracle   26295     1  0 Apr16 ?        00:00:01 ora_pmon_RMANDELL

Four PMONs for the same ORACLE_SID? How did I do it? To ask some additional question or to post answer simply add a comment. Deadline is set on 24.4.2008 23:59 (Thursday). More info:


$ ls $ORACLE_HOME/dbs
hc_RMANDELL.dat  initRMANDELL.ora  lkRMANDELL  orapwRMANDELL

- ps shows real Oracle PMON processes - one (S)PFILE - one Oracle installation - instances are started under the same user and group - version of Oracle DB does not matter Waiting for Your answers... UPDATE And the winner is Jakub Wartak who posted correct solution although I was expected little different solution (he had used symlink for it) but the main point is the same - you need to set "different" ORACLE_HOME. So how i did start more (for instances) with the same ORACLE_SID? My ORACLE_HOME for 11g is located at /opt/oracle/111.


$ ps -aef|grep pmon|grep -v grep
$ env|grep ORA
ORACLE_SID=RMANDELL
ORACLE_BASE=/opt/oracle
ORACLE_HOME=/opt/oracle/111
$ sqlplus / as sysdba

SQL*Plus: Release 11.1.0.6.0 - Production on Mon Apr 21 08:46:23 2008

Copyright (c) 1982, 2007, Oracle.  All rights reserved.

Connected to an idle instance.

SQL> startup nomount
ORACLE instance started.

Total System Global Area  538656768 bytes
Fixed Size                  2146112 bytes
Variable Size             150995136 bytes
Database Buffers          381681664 bytes
Redo Buffers                3833856 bytes
SQL> exit
Disconnected from Oracle Database 11g Enterprise Edition Release 11.1.0.6.0 -
64bit Production
With the Partitioning, Oracle Label Security, OLAP, Data Mining
and Real Application Testing options

Now I change the ORACLE_HOME and add "/" at the end of path which change the ORACLE_HOME but in fact path reamains valid. So my new ORACLE_HOME is /opt/oracle/111/.


$ export ORACLE_HOME=/opt/oracle/111/
$ sqlplus / as sysdba

SQL*Plus: Release 11.1.0.6.0 - Production on Mon Apr 21 08:46:46 2008

Copyright (c) 1982, 2007, Oracle.  All rights reserved.

Connected to an idle instance.

SQL> startup nomount
ORACLE instance started.

Total System Global Area 538656768 bytes Fixed Size 2146112 bytes Variable Size 150995136 bytes Database Buffers 381681664 bytes Redo Buffers 3833856 bytes SQL> exit Disconnected from Oracle Database 11g Enterprise Edition Release 11.1.0.6.0 - 64bit Production With the Partitioning, Oracle Label Security, OLAP, Data Mining and Real Application Testing options Again I modify ORACLE_HOME and I put another "/".


$ export ORACLE_HOME=/opt/oracle/111//
$ sqlplus / as sysdba

SQL*Plus: Release 11.1.0.6.0 - Production on Mon Apr 21 08:47:12 2008

Copyright (c) 1982, 2007, Oracle.  All rights reserved.

Connected to an idle instance.

SQL> startup nomount
ORACLE instance started.

Total System Global Area  538656768 bytes
Fixed Size                  2146112 bytes
Variable Size             150995136 bytes
Database Buffers          381681664 bytes
Redo Buffers                3833856 bytes
SQL> exit
Disconnected from Oracle Database 11g Enterprise Edition Release 11.1.0.6.0 -
64bit Production
With the Partitioning, Oracle Label Security, OLAP, Data Mining
and Real Application Testing options

I can also add "/" at begin of path and path of course remains valid.


$ export ORACLE_HOME=///opt/oracle/111
$ sqlplus / as sysdba

SQL*Plus: Release 11.1.0.6.0 - Production on Mon Apr 21 08:48:01 2008

Copyright (c) 1982, 2007, Oracle.  All rights reserved.

Connected to an idle instance.

SQL> startup nomount
ORACLE instance started.

Total System Global Area  538656768 bytes
Fixed Size                  2146112 bytes
Variable Size             150995136 bytes
Database Buffers          381681664 bytes
Redo Buffers                3833856 bytes
SQL> exit
Disconnected from Oracle Database 11g Enterprise Edition Release 11.1.0.6.0 -
64bit Production
With the Partitioning, Oracle Label Security, OLAP, Data Mining
and Real Application Testing options

Now I have started four RMANDELL instances.


$ !ps
ps -aef|grep pmon|grep -v grep
oracle   26459     1  0 08:46 ?        00:00:00 ora_pmon_RMANDELL
oracle   26501     1  0 08:46 ?        00:00:00 ora_pmon_RMANDELL
oracle   26541     1  0 08:47 ?        00:00:00 ora_pmon_RMANDELL
oracle   26583     1  0 08:48 ?        00:00:00 ora_pmon_RMANDELL

Maybe you are asking what is the meaning of this post. As I wrote in title in this post it is for fun. Even this post is for fun it brings one lesson. If you've got following message


ORA-01034: ORACLE not available
ORA-27101: shared memory realm does not exist

or you've got "Connected to an idle instance.", your instance is up and running and your enviroment seems to be correct. Then you did (probably) set the enviroment manually (using a "Tab key" in Bash) and that created "/" at the end of the path. To find out which ORACLE_HOME is used just take a look on any trace file genereated in background_dump_dest directory.

Comments

Nicolas Gasparotto18 April, 2008Good question. Is it a bug or voluntary ? Oracle version ? One listener ? Does your instance run fine ?

Ivan Kartik19 April, 2008>> Is it a bug or voluntary ? Well I'm not sure :-D but probably voluntary. >> Oracle version ? RMANDELL is 11gR1 but in fact version does not matter. I tested this also on 9i and 10g. >> Does your instance run fine ? AFAIK yes. But note there is just one instance with mounted database. The other instances are just in nomount state. But question is how could I start so many instances with the same ORACLE_SID.

Jakub Wartak20 April, 2008By using different ORACLE_HOME, eg. # there is one running ORACLE_SID=test instance [oracle@xeno ~]$ env | grep ^ORA ORACLE_SID=test ORACLE_HOME_LISTNER=/u01/app/oracle/product/10.2.0 ORACLE_BASE=/u01/app/oracle ORACLE_HOME=/u01/app/oracle/product/10.2.0 [oracle@xeno ~]$ cd /u01/app/oracle/product/ [oracle@xeno product]$ ln -s 10.2.0 bleh [oracle@xeno product]$ export ORACLE_HOME=`pwd`/bleh [oracle@xeno product]$ !env env | grep ^ORA ORACLE_SID=test ORACLE_HOME_LISTNER=/u01/app/oracle/product/10.2.0 ORACLE_BASE=/u01/app/oracle ORACLE_HOME=/u01/app/oracle/product/bleh [oracle@xeno product]$ sqlplus / as sysdba SQL*Plus: Release 10.2.0.3.0 - Production on Sun Apr 20 21:40:15 2008 Copyright (c) 1982, 2006, Oracle. All Rights Reserved. Connected to an idle instance. SQL> startup nomount; ORACLE instance started. Total System Global Area 327155712 bytes Fixed Size 1261612 bytes Variable Size 201330644 bytes Database Buffers 121634816 bytes Redo Buffers 2928640 bytes SQL> host ps -aef | grep pmon | grep -v grep oracle 3910 1 0 12:03 ? 00:00:00 ora_pmon_test oracle 14699 1 0 21:40 ? 00:00:00 ora_pmon_test

Ivan Kartik21 April, 2008Congratulations, your answer is correct. If you like "Slivovica" please leave me a message (use "contact me" link on right menu) with your correct (snail) mail address and I'll send you a bottle :-D

Nicolas21 April, 2008Nice to know, thanks Ivan, and congrat to Jakub. But it's not fair, you close the thread before the deadline... ;-)

Nicolas21 April, 2008In addition, you could complains on metalink note 435044.1, that must be updated as you check same behaviour on 11g. Have fun !

Ivan Kartik21 April, 2008>> But it's not fair, you close the thread before the deadline... Well you know solution was posted so there wasn't a sense to continue... :-D >> In addition, you could complains on metalink note 435044.1, that must be updated as you check same behaviour on 11g. Thank you Nicolas. Good to know that there exists note for this. I saw this behaviour for first time on 8i (8.1.6) on Linux some years ago. My colleague was facing to this problem last week and fortunately I could remember the reason :-D Problem of my colleague had inpired me to write this post...

Ivan Kartik21 April, 2008>> that must be updated as you check same behaviour on 11g. I have sent them e-mail, we'll see how many time will take this update :-D

Nicolas05 May, 2008>>I have sent them e-mail, we'll see how many time will take this update :-D Amazing, the note has already been updated ! Thanks to you to be the originator.

About

Categories

Latest posts

Pages

Fun: How is it possible? - updated (and the winner is...)

Comments

New comment