ORA-01034 / ORA-27101 / Linux-x86_64 Error

[bapocalypse]

Environment Information:
* Hardware: Dell PowerEdge 12GB RAM / 2TB disk
* version of Oracle: 10g R2
* Operating system: CentOS 5.4

] Hello, I need help. I never acted like a DBA, much less I had already used Linux in life. I was prompted to install centos 5.4 with Oracle 10G R2.

Unfortunately it was not possible to create 3 virtual servers for each need (development, approval and production). All environments should be installed in the same instance.

I was able to install Linux and Oracle without problems. I created 03 (three) distinct bases. (Dev, Homolg, PROD).

When restarting the server, do the following steps:

Select all

#lsnrctl start (sem erros) 
# dbstart (também sem erros) 
# sqlplus /nolog 
SQL> connect system/senha@desenv 
connected. 
SQL> connect system/senha@homolog 
connected. 
[b]SQL> connect system/senha@prod 
ERROR: 
ORA-01034: ORACLE not available 
ORA-27101: shared memory realm does not exist 
Linux-x86_64 Error: 2: No such file or directory[/b]

Does anyone have any idea what it could be? I can not find any help. I have already reinstalled Linux and Oracle and nothing.

Follow Startup.log

Select all

 
/oracle/ora10g/bin/dbstart: Starting up database "dboradsv" 
Mon May 17 09:02:24 BRT 2010 
 
 
SQL*Plus: Release 10.2.0.1.0 - Production on Mon May 17 09:02:24 2010 
 
Copyright (c) 1982, 2005, Oracle.  All rights reserved. 
 
SQL> Connected to an idle instance. 
SQL> ORACLE instance started. 
 
Total System Global Area 3758096384 bytes 
Fixed Size                  2025232 bytes 
Variable Size             738199792 bytes 
Database Buffers         3003121664 bytes 
Redo Buffers               14749696 bytes 
Database mounted. 
Database opened. 
SQL> Disconnected from Oracle Database 10g Enterprise Edition Release 10.2.0.1.0 - 64bit Production 
With the Partitioning, OLAP and Data Mining options 
 
/oracle/ora10g/bin/dbstart: Database instance "dboradsv" warm started. 
 
/oracle/ora10g/bin/dbstart: Starting up database "dborahml" 
Mon May 17 09:02:35 BRT 2010 
 
 
SQL*Plus: Release 10.2.0.1.0 - Production on Mon May 17 09:02:35 2010 
 
Copyright (c) 1982, 2005, Oracle.  All rights reserved. 
 
SQL> Connected to an idle instance. 
SQL> ORACLE instance started. 
 
Total System Global Area 3758096384 bytes 
Fixed Size                  2025232 bytes 
Variable Size             721422576 bytes 
Database Buffers         3019898880 bytes 
Redo Buffers               14749696 bytes 
Database mounted. 
Database opened. 
SQL> Disconnected from Oracle Database 10g Enterprise Edition Release 10.2.0.1.0 - 64bit Production 
With the Partitioning, OLAP and Data Mining options 
 
/oracle/ora10g/bin/dbstart: Database instance "dborahml" warm started. 
 
/oracle/ora10g/bin/dbstart: Starting up database "dboraprd" 
Mon May 17 09:02:43 BRT 2010 
 
 
SQL*Plus: Release 10.2.0.1.0 - Production on Mon May 17 09:02:43 2010 
 
Copyright (c) 1982, 2005, Oracle.  All rights reserved. 
 
SQL> Connected to an idle instance. 
SQL> ORA-27102: out of memory 
Linux-x86_64 Error: 28: No space left on device 
SQL> Disconnected 
 
/oracle/ora10g/bin/dbstart: Database instance "dboraprd" warm started. 
 
 

[Dulima-SP]

Friend,
how did you dye the memory for these bases ???

Have you tried to rise from production ??

Up the bases on different orders ??

[bapocalypse]

Dulima, thanks for the quick response.

then, in fact no matter the "order". It's always the third that gives the mistake.

And in relation to scaling memories for the bases, you really got me, because I have no idea how to do this. And I do not even know what it's about. If it is in the creation of the bases, it was at the base of the NNF (Next, Next, Finish).

[Dulima-SP]

Let's see what we got !!!
já já The Masters appear !!!


in Linux
Digital

Select all

df -h

and oralce

Select all

 show parameters sga

in 3 instances !!!

[bapocalypse]

Come on:

Select all

df-h

Select all

 
[oracle@bspolnxora1 ~]$ df -h 
Filesystem            Size  Used Avail Use% Mounted on 
/dev/mapper/vgroot-lvroot 
                      3.0G  492M  2.3G  18% / 
/dev/mapper/vgroot-lvvar 
                      3.0G  181M  2.6G   7% /var 
/dev/mapper/vgroot-lvhome 
                      2.0G   72M  1.8G   4% /home 
/dev/mapper/vgroot-lvusr 
                      4.9G  2.3G  2.4G  49% /usr 
/dev/mapper/vgroot-lvtmp 
                      4.9G  218M  4.4G   5% /tmp 
/dev/sda1             190M   19M  162M  11% /boot 
tmpfs                 5.9G     0  5.9G   0% /dev/shm 
/dev/mapper/vgroot-lvoracle 
                      9.9G  2.9G  6.5G  31% /oracle 
/dev/mapper/vgoradata-lvoradata 
                      985G  2.9G  932G   1% /oradata 

base dboradsv

Select all

 
SQL> show parameters sga 
 
NAME                                 TYPE        VALUE 
------------------------------------ ----------- ------------------------------ 
lock_sga                             boolean     FALSE 
pre_page_sga                         boolean     FALSE 
sga_max_size                         big integer 3584M 
sga_target                           big integer 3584M 
 

DBORAHML base

Select all

 
SQL> show parameters sga 
 
NAME                                 TYPE        VALUE 
------------------------------------ ----------- ------------------------------ 
lock_sga                             boolean     FALSE 
pre_page_sga                         boolean     FALSE 
sga_max_size                         big integer 3584M 
sga_target                           big integer 3584M 
 

base dboraprd (where error happens, but I believe to be the same value of others)

Select all

 
ERROR: 
ORA-01034: ORACLE not available 
ORA-27101: shared memory realm does not exist 
Linux-x86_64 Error: 2: No such file or directory 

[diegolenhardt]

I think you'vê created the 3 bases like dedicated, this allocated a memory value for SGA larger than what you have of RAM.

Homologation and Develops 3.5GB?

You will have to give Alter System, to do also for the graphical interface, with Oracle Enterprise Manager,

first has to lower the SGA_TARGET, SCOPE = spfile ..., so that the other decrease,

[diegolenhardt]

I traveled, the fact that being dedicated or shared does not change anything to sga, only PGA,

but your problem is this same, it's very high value for sga, the solution is to decrease it there For some instance,

If you did for DBCA, you were questioned about memory in the fifth step I think, could put in percentage or fixed values ??for each component including,

[Dulima-SP]

Leave the production base with more memory !!!

makes the changes and put there for us !!

[bapocalypse]

So ... I made a radical decision. I will have only one single basis with the Dev / Homolog / Prod schemas. This is the worst possible scenario.

although I will not have production in reality. Only one bank with the production data. Another machine will be the database server for the application.

In this way, I am creating the database with the DBCA and allocating 80% for Oracle. That was like this:
SGA: 8GB
PGA: 2GB
Oracle Process: 40MB

In this hairy scenario, it would be the Ideal this way?

Our biggest problem will be the large volume of data to be processed.

[diegolenhardt]

Look, man, I do not know your case, finally, I would make the separate production, with a good so much more than memory, and homologation / devised in schemas, but in a different instance,
you want to download the bank for some reason, go for everything, I do not know there are many reasons,

inclusive, the main thing is that, so you can avoid disk competition between the bases, allocned datafiles Separately, I would even give too the way you said,

this succession, requires what exactly, processing, lots of insertion? calculations, etc.

more details ...

[bapocalypse]

Well, I know it's a pig scenario. But you'll try to convince the people here. As I said in the beginning, I'm not DBA but I have my concepts and are correct, as you said too. Three separate instances would be ideal. Each in your VMware for example ... but ....

and will only be for great processing of data volume overall. And they want everything together, just to use the maximum server. Even with the competition.

Hardness ... Everything you learn from the years has to do wrong afterwards.

sends who can, obey who has judgment.

: OOPS:

[diegolenhardt]

Good luck, rs

Dai goes a head on the develops and runs an instruction that takes 100% CPU from the server, VMware ESXi 4.0 would be ball show, at least two VMs, production and another HOMOG / DESS,

Are you using Storage or normal SCSI discs?

has a nice bid that is the ASM, it manages the data files, not having an extra layer that is the disk mounted on the operating system, is a bit more advanced, but I believe it would win a little performance,

but was solved, closed it was,

Any doubt put there ..

[bapocalypse]

So I'm using Dell with VMware Esxi 4.0. And should create VMs to resolve. But it was not possible because I did not have a white card for it.

and will be fucking a process to disrupt the other. But, I hope it occurs soon and changed as soon as possible.

and on the server, are SAS 15K discs in RAID5.

As for asm, I will search for what it is, because I have no idea.

Now I am seeing about the use of Bigfile on Tablespace or not, since I will have more than 100GB.

Thanks.

[diegolenhardt]

If you create the tablespaces by DBCA, there is a check for you to score, to use bigfile,


Ideal is always take a look at the documentation, If it helps ..
http://download.oracle.com/docs/cd/B193 ... m#ADMIN002
If I am not mistaken Bigfile goes up to 32TB,

RAID5 is not recommended for the database, in question of Performance, from one surveyed on the subject you will see, the bank will rotate in archivelog mode?

in relation to VMware has to take care of, as far as I know, Oracle does not support virtualized banks in VMware technology, only from hers, of course, Oracle VM,

and in relation to licensce, for Oracle tells the total color number of the machine, see the machine, even if you allocated for example only two 2 color processors each,

99] is expensive, they are small things, that at the end of a difference ...