Local pfile for all RAC instances

DB Version: 10Gr2
OS : Solaris 5.10
2 node RAC
In one of our RAC DBs, we use local pfile for both of our instances.
If we had used a common (shared) spfile, we need to change parameter only once and it will be persistent across reboots.
Can you think of any other disadvantages of using local pfile? Are there any advantages in using pfile in RAC enviroments ?

Yes .. you are right ..
Oracle recommends that you manage parameters using an SPFILE. If you use client-side parameter files, then Oracle does not preserve parameter changes made for self-tuning after shutdown.
Check Oracle Documentation :
http://download.oracle.com/docs/cd/B19306_01/install.102/b14203/params.htm#CIHFCHGH
Regards
Rajesh

Similar Messages

Running OEM for Other RAC Instances

Hi,
My RAC setup has 3 nodes and I know that more than one database exists (say Retail, Corporate, Private banking databases etc) on a single shared storage. The portlist.ini file from one of the node instance looks like below:
Enterprise Manager Console HTTP Port (a111) = 5500
Enterprise Manager Agent Port (a111) = 3938
Enterprise Manager Console HTTP Port (b111) = 1158
Enterprise Manager Agent Port (b111) = 3938
Enterprise Manager Console HTTP Port (c111) = 5501
Enterprise Manager Agent Port (c111) = 1831
Enterprise Manager Console HTTP Port (d111) = 5502
Enterprise Manager Agent Port (d111) = 1832
1. Is the above correct? Shouldn't there had been only one entry? Is there one entry each for every database?
2. The OEM is always defaulting to 1158. How can I make it run on other ports for all other databases?
Please help.
Thanks.

The concept behind dbconsole is, each database has its own dbconsole, which can only serve this particular database. Of course, in a RAC environment you have more than one instances per database, but dbconsole for database A does not know any other databases on the local or remote machines.
If you want to use dbconsole and you have x databases , you also have dboconsoles multiplied by x.
As an alternative you could use the old javabased EM, still available on the 10g client software. This EM can serve several databases.
Unlike dbconsole, which comes with the server software, Grid Control is independent of server/database installation. Mainly it consists of a Management Server and agents, which are installed on the target machines and collect all necessary informations for a centralized management.

CCMS - Configuring a Avg. for all server instances.

Is there any way to get this ????
We get the Data for the Individual server instance Production_01 where as I want it for the entire system.
As mentioned below in detail...
RZ20 - Transaction .
SAP CCMS Technical Expert Monitors
-> System / All Monitoring Segments / All Monitoring Contexts
    -> SAP_CCMS_Production
      -> Production_01
        -> R3BasisSystem
          -> MemoryManagement
            -> R3MemMgmtResources
              -> EsAct
Now for this we are currently monitoring only 1 instance . Something
like this is required .
Currently, for system MP1 (same problem for MP2 and MP3), I can use the
class MTE R3MemMgmtEsAct to monitor the Extended Memory.
R3MemMgmtEsAct checks the avaibility of Extended Memory for an AS.
I want get an alert only if the Extended Memory is not available for
all the AS in the system MP1, not only for an AS.
If I set a threshold value / 90% of Extended Memory used, I don't want
an alert if there are an AS with 95% of Extended Memory used and an
other AS with 20% used.
=> If there are 5 AS in my system : AS1=20% / AS2=20% / AS3=20%... =>
5x20=100% for the system.
Exemple1 / threshold value 90% :
Consumption / AS1 = 50% (-> 10% for the system)
Consumption / AS2 = 100% (-> 20% for the system)
Consumption / AS3 = 50% (-> 10% for the system)
Consumption / AS4 = 50% (-> 10% for the system)
Consumption / AS5 = 50% (-> 10% for the system)
=> Extended Memory used for the system = 60% => NO ALERT GENERATED
Exemple2 / threshold value 90% :
Consumption / AS1 = 89%
Consumption / AS2 = 100%
Consumption / AS3 = 89%
Consumption / AS4 = 89%
Consumption / AS5 = 89%
=> Extended Memory used for the system = 91,2% => ALERT GENERATED
=> I want check the system (several AS), not an AS.
Thanks to all in advance.
Regards,
Ravi Sharma

Hai,
First of all this is not possible at least with SAP tools.
Secondly why do you want to produce alert by calculating the average of the threshold values in %.
As per under SAP standard, if you have 5 application servers for a system then if any one application system goes beyond the set threshold a alert will be produced for the particular application server. This will help us to rectify the problem in that application server.
As per your query if at all you setup the alert management then if 3 application servers goes beyond the set threshold (say 95%) and still if the total average for all the 5 application servers does not go beyond the limit you had set (say 90%) then you will not know that 3 of the application servers has already gone beyond the threshold (as per SAP standards). This will cause the system to be overloaded and can cause some other consequences. Keeping the standard setup will help in good way to maintain your system.
Having the CCMS setup in the standard setup is better, which helps us to maintain systems individually.
Regards,
Yoganand.V

Where can i find start/stop scripts for all SAP instances

Hi,
I installed CI, DB and Dialog instance on a standalone S10 machine and i could see start scripts under /sapmnt/RAM/profile
START_DVEBMGS09_hostSAP
START_D24_hostSAP
But i cann't see stop scripts for the above and also i cann't see start and stop scripts for SAP DB instance. Could some one point me where i can find them when everything is installed on a single box.
Thanks,
Ram

USAGE: STARTSAP.EXE name=<SID> nr=<SYSNR> SAPDIAHOST=<host>
FURTHER INFORMATION:
- The executable sapstart.exe must be in the same directory
when you start sap instacne, it will check db instance first, if db instance not start , it will automatic start db instance then sap instance.
vice visa :
stopsap is alos can be used.

*** How do I put a program on my local network for all to use?

I would like to put a program we have downloaded for our accunting on our office network so that either of our 3 employees can access it and add the most recent information. Is there a way to do this through the network folder in the finder? I can see the other 2 computers and can connect to them, but I'm wondering if there is a folder that the person who has the program can put it in, so we can all access it.
Thanks in advance.

I'm not sure if you're asking about an application "program" that does something or a file that you put data in, and if it's an application how many licenses you have to use copies. Opening application across a newtwork can be very slow, depending on the size and complexity so that wouldn't be most preferred. Also sounds like you don't have a common fileserver?
Either way, how about this, assuming you want to keep everybody's personal data safe - On one of the computers, the one most likely to be running when needed, create a new Standard (not Admin) account that's only used to hold this program/data, and put the program/data in either the Public or Documents folder (depending on how much access control you need/want to have) in that new user's account. Then when people want to access the common data, they can log into that account via the Network folder, or park the icon for that account in their Dock for even easier access. Someone would have to remember to log that account on, I think, when ever the machine is started. Or for a little less security, you could just park the common data in the Public folder of the normal user on that computer.
I do essentially this regularly to move or back up data between my MacBook and wife's Powerbook - works just fine, subject to network speed of course.
Michael

Can't we start a RAC instance without local pfile ?

DB version : 11.1.0.7.0
OS : Solaris 10 SPARC
--Info about the cluster
2 Node RAC
Both instances using spfile stored in ASM Diskgroup
Passwordless SSH configuration is not working currently (i gather ssh is not mandatory after installation)
We have 7 RAC DBs running in our 11.1 CRS.
After a reboot of our 2 Node RAC servers, in Node2, 4 instances didn't come up.
None of these down instances have a local pfile. The ones that are up in Node2 have it.
So, When i manually tried to bring up the ones that are down
$ sqlplus / as sysdba
SQL*Plus: Release 11.1.0.7.0 - Production on Fri Mar 2 21:28:22 2012
Copyright (c) 1982, 2008, Oracle. All rights reserved.
Connected to an idle instance.
SQL> startup
ORA-01078: failure in processing system parameters
LRM-00109: could not open parameter file '/u01/app/oracle/product/11.1/db_1/dbs/initorcl2.ora'
SQL> exit
Disconnected-- Tried directly with spfile (No such option in oracle, but still tried ! )
SQL> startup spfile='+ORCL_DATA01/ORCL/spfileorcl.ora'
SP2-0714: invalid combination of STARTUP options-- Tried from Node1
-- This might have failed because ssh is not working currently. Not sure
$ srvctl start instance -d orcl -i orcl2
PRKP-1001 : Error starting instance orcl2 on node manhclust199
CRS-0215: Could not start resource 'ora.orcl.orcl2.inst'.
$ srvctl start database -d orcl
PRKP-1001 : Error starting instance orcl2 on node manhclust199
CRS-0215: Could not start resource 'ora.orcl.orcl2.inst'.Finally I had to create a local pfile on node2 for each instances in OH/dbs
The content of each local pfile is like
$ cat initorcl2.ora
spfile='+ORCL_DATA01/orcl/spfileorcl.ora'Now the startup worked without any issue for all 4 instances
SQL> startup
ORACLE instance started.
Total System Global Area 2137886720 bytes
Fixed Size                  2155464 bytes
Variable Size            1593838648 bytes
Database Buffers          536870912 bytes
Redo Buffers                5021696 bytes
Database mounted.
Database opened.So, local pfile is mandatory for all RAC instances ?

Hi,
My question is whether this instance specific local pfile must be there in OH/dbs directory ?YES
Didn't get a clear answer yet.
It seems that it is mandatory. And this is most probably the reason these instances didn't come up after the reboot of this node.Startup Sequence:
Startup of Cluster:
Using "srvctl" to start your database/instance:
Srvctl read the OCR and check the parameter SPFILE from your database registration:
If parameter SPFILE in OCR have a value the srvctl will use this value to start your database instance.
If parameter SPFILE in OCR don't have a value (i.e null) Oracle will try find the file SPFILE<ORACLE_SID>.ora on $ORACLE_HOME/dbs
If the file SPFILE<ORACLE_SID>.ora does not exists, Oracle will try use the file init<ORACLE_SID>.ora
If init<ORACLE_SID>.ora does not exist... the startup will fail.
From ORACLE_HOME of your database execute command above to get value of Spfile parameter on OCR:
srvctl config database -d <DB_NAME> |grep SpfileStartup manually from SHELL using SQLPLUS:
Using SQLPLUS to start a database Oracle always first use SPFILE<ORACLE_SID>.ora stored on $ORACLE_HOME/dbs/ if this file does not exist he will try use INIT<ORACLE_SID>.ora to start Database Instance.
If your spfile is stored on ASM you must create a INITFILE pointing to SPFILE stored on ASM.
e.g From instance 1
cat /oracle/oracle/product/11.2.0/db_1/dbs/initorcl1.ora
SPFILE='+DATA/ORCL/spfileorcl.ora'Regards,
Levi Pereira

RAC DB parameters where stored for all nodes

If I have 4 nodes (RACs) where database parameters stored for each instance.?
I know for single instance database all parameters are stored either in V$PARAMETER or V$SPPARAMETER depending on with which file database was started.

The v$spparameter has a column "SID" which reflects the SID for each RAC instance , e.g. :
SQL> select value from v$spparameter where sid='RAC1' and name='thread';
1
SQL> select value from v$spparameter where sid='RAC2' and name='thread';
2
Assuming that the RAC instance names are RAC1 and RAC2

Redo log groups/files on the local disk for RAC DB

Whether is it possible/supported to place redo log files/groups on the local disk for a RAC databases ?
Thank you,

Hi Mufalani,
are you sure about this one? I'd think that this only works as long as the nodes don't restart/crash. I can imagine this like this:
The nodes store redo logs on local disks which are NFS-exported to the other node. Everything should be fine. But when one node crashes/reboots, the other one has to perform crash recovery for the crashed instance but this won't work when the NFS mount (with redo logs) is unaccessible. So I would not want to do this.
Bjoern

Cleint tnsnames.ora for connecting to RAC instance

Dear all,
DB:10.2.0.1
o/s:redhat 5.6
2node rac cluster
I am getting ERROR: ORA-12170: TNS:Connect timeout occurred when trying to connect to rac instance.the timeout frequency is very less once in 10 times.
cleint tnsname.ora
RACTEST =
(DESCRIPTION =
(ADDRESS = (PROTOCOL = TCP)(HOST = 192.168.2.25)(PORT = 1521))
(ADDRESS = (PROTOCOL = TCP)(HOST = 192.168.2.27)(PORT = 1521))
(FAILOVER= TRUE)
(CONNECT_DATA =
(SERVER = DEDICATED)
(SERVICE_NAME = RACTEST)
Regards

Try to connect straight to the nodes. One node at the time. You can add new tnsnames entries for all RAC nodes so you can connect only one node not the whole RAC.
example:
RACTEST1 =
(DESCRIPTION =
(ADDRESS = (PROTOCOL = TCP)(HOST = 192.168.2.25)(PORT = 1521))
(CONNECT_DATA =
(SERVER = dedicated)
(SERVICE_NAME = RACTEST)
(INSTANCE_NAME = RACTEST1)
RACTEST2 =
(DESCRIPTION =
(ADDRESS = (PROTOCOL = TCP)(HOST = 192.168.2.27)(PORT = 1521))
(CONNECT_DATA =
(SERVER = dedicated)
(SERVICE_NAME = RACTEST)
(INSTANCE_NAME = RACTEST2)
If both nodes are working then problem is probably in network / firewall.
Usually this kind of problems comes because one node's database instance is somehow broken. And if that is the problem correction is restart failing instance.
Edited by: Simo Kemppinen on Jul 15, 2012 12:58 PM
Edited by: Simo Kemppinen on Jul 15, 2012 12:59 PM

How to open oracle listner for three database instances

please help me with following questions
I have 3 database instances
orcl
JIP02T
JIP04T
when i start the oracle database, does it start all 3 instances? when i start the oracle listner, does it start listner for all 3 instances?
when i echo $ORACLE_SID
[ora112@localhost admin]$ echo $ORACLE_SID
orcl
i can only see orcl and oracle listner is started only for one database instance.i can connect to only one database instance usign sql developer how can i start oracle database listner for other database instances as well

ShishirTekadeR wrote:
. oraenv
Note that is "dot" SPACE oraenv
You will be prompted for the value of ORACLE_SID (one of the databases). It looks like this:
oracle:orcl$ . oraenv
ORACLE_SID = [orcl] ? dwsbx
The Oracle base for ORACLE_HOME=/u01/app/oracle/product/11.2.0/db_1 is /u01/app/oracle
oracle:dwsbx$
HIGHLY APPRECIATED U R IMP Tips!!!!!
But it’s very difficult to remember & manually enter oracle SID & HOME details every time after run (. Oraenv)
.profile creation is simple method and no need to remember SID & HOME details.(for multiple DB in one server)You don't have to "remember & manually enter oracle SID & HOME details every time after run " oraenv sets all those details for you. All you have to remember is the name of the instance you want to work with. The same thing you'd have to remember if you if you created (as you suggested) three different "profiles" - actually three different shell scripts, each to set the environment for a specific instance.
Yes, I do place "default" settings in .bash_profile. But with multiple databases the "default" settings are often not going to be what you need so you'll still have to run some process to change them. Either a home-grown script or oraenv. And either way you will have to remember the names of the instances you have to chose from. So in general I see no advantage of reinventing the wheel.
>
>
>
As per user information
He is using different port number for different instanceWhich accomplishes absolutely nothing except extra configuration headaches. It accomplishes nothing for security. It accomplishes nothing for availability.
And also it is very simple method to manage three different DB in three different listener entry (for DB maintenance activity.)Not as easy as managing one default listener. By far the most frequent listener configuration issue I see on this forum is when people try to configure one listener per database.
Single listener in multiple instance (if listener down all DB connection will affect)And why would the listener be down? I've been working full time with Oracle since version 7.3 - somewhere around 1995. I've NEVER seen a listener fall over and die. I have seen many people on this forum whose db connections failed because they were trying (unsucessfully) to configure multiple listeners.
All of my servers are running multiple databases. I have one test server that is running close to a dozen separate databases. One listener. I've NEVER had a situation where I thought to myself, "Gee, it would really help if I had created a separate server for each database"
>
>
Best Regards,
Shishir Tekade.
My Blog: http://shishirtekade.blogspot.com
Edited by: ShishirTekadeR on Dec 27, 2012 10:29 PM
Edited by: ShishirTekadeR on Dec 27, 2012 10:30 PM
Edited by: ShishirTekadeR on Dec 27, 2012 10:31 PM

Retrieving the destination location of output for a specific instance

Hi,
I have reports that run daily and create output PDFs in Enterprise location. I am now trying to access these output PDFs using a C# program using SDK. I see that when i query for PDF instances, the query does not give me SI_SCHEDULEINFO. How do I access the destination location information for all output instances?
I am using CRS 2008. Thanks for the help.
Bin.

I just found it. It is in the SI_FILES property. Select SI_FILES in the query and infoobject.files would give the file information you need.

Creating a RAC standby database for a single instance database

Dear All,
I have a task of migrating a 500GB single instance database to a two-node RAC database with a little downtime at hand. My migration strategy is to:
1) Create a RAC physical standby for the Single Instance database
2) Switchover to RAC standby.
Primary and Standby OS and DB configurations:
OS: Windows Server EE 2003 (64-bit)
DB: Oracle 10g Database Release 2 (10.2.0.4)
Oracle 10g Clusterware Release 2 (10.2.0.4)
To create a RAC standby, I will:
a) Install Clusterware (10.2.0.1)
b) Install Database (10.2.0.1)
c) Patch both Clusterware and Database (10.2.0.4)
d) Create ASM instance for both the nodes (+ASM1 & +ASM2)
e) create standby controlfile on primary
f) Move standby controlfile, RMAN backup of primary, pfile, listener.ora, tnsnames.ora, password file to standby host-1
g) make necessary changes to the pfile on standby host-1 like cluster_database, instance_name, thread, ...
h) mount standby database and restore backup
Kindly validate my steps and if there already exists such a document then please do provide me with a link.
Regards

Please refer to MAA whitet paper :
[http://www.oracle.com/technology/deploy/availability/pdf/MAA_WP_10g_RACPrimaryRACPhysicalStandby.pdf]
[MAA website|http://www.oracle.com/technology/deploy/availability/htdocs/maa.htm]

Can I set an mx control property globaly for all instances of the control.

Is there a way of setting the value of a variable on a control for all instances of the control in the application?
For example, I have several instances of an mx:Label and I want to change the enabled property for all instances at once rather than refer to each one individually by id.

One way is to subclass Button and change its enabled getter to watch some static property. You don’t really want to change Button itself because Buttons exist in lots of places like ScrollBars.

Adding a domain user to the admin role within the local user management breaks all metro apps for all users!!

Hi,
I have posted this in another large thread under the "Windows 8 General" group but have not had any appropriate feedback from MS.
After hours of testing and working with other users I have managed to isolate a simple situation that breaks all metro ui applications within Windows 8 for all users on the machine. Here are my exact steps and notes.
Before continuing if you are running Avast then your solution may be to turn of the behaviour shield functionality as this also breaks metro apps. This is NOT the problem we are having!
I have performed 3 cleans installs after isolating the problem and am able to reproduce the issue every time using the same steps on two different machines.
First thing to say is that for us it has nothing to do with simply joining the domain, domain/group policies nor does it appear to have anything to do with the software we installed, the problem here is much more simple but the result is pretty terrible.
Here are my exact steps of what I did to reproduce our problem:
Complete format of HDD in preperation for a clean install
Clean install performed
Set up the machine initially with a local account
Test metro apps - all working fine
Open control panel from the desktop, click on System, change the system to join the domain, click reboot
Log into the system using my domain account
Test metro apps - all working fine
Here's were the problem starts. I need my domain account to have admin rights on the local machine so I can install programs without the IT men having to come over and enter their password every 5 mins.
I go to control panel via the desktop and click on User Accounts. From with here I then click on "Manage User Accounts". This requires the IT guys to enter their details to give me access to such functionality. This is fine
In the dialog box that opens I can only see the local user that was initially created during setup. The "Group" for this local account shows as "Administrators" - Image included below (important to note that metro apps are working at this point)
I click add and then add my domain account - also giving it administrator access
Sign off or reboot to ensure the new security is applied
Sign back in to the domain account
Test metro - ALL BROKEN
Sign out
Sign in as local account
Test Metro - NOW ALL BROKEN FOR THIS USER ALSO
So as soon as I add my domain account to the local user accounts and set it as admin it breaks all metro apps for all users. This is on a totally clean install with nothing at all installed other than the OS.
Annoyingly if I go back and change the domain account to a standard user or if I totally remove the domain account from the local account management system the problem does not go away for either user. basically it is now permanently broken. The only fix I
could fathom was a full re install and not giving the domain user admin access to the local machine.
Screen one - this is the local user accounts window AFTER joining the domain and logging in with my domain account (All metro apps working at this point)
Screen 2: User accounts AFTER joining the domain and AFTER adding domain account to local user management (METRO BROKEN)
I have isolated my machine from all group policies so nothing like that is affecting me. Users I have spoken to in different companies have policies that automatically add users to the local user management. This means that metro apps break as
soon as they join the domain which leads them to wrongly think it is group policies causing the error. Once they isolate themselves from this they can reproduce following my steps.
Thanks

Hi Juke,
Thank you for the response and apologies for the delay in getting back to you. My machine was running a long task so I couldn't try your suggested solution.
I had already tried running the registry merge suggested at the top of the thread to no avail. I had not tried deleting the OLE key totally so I did that and the problem still exists. I will post all the errors I see in event viewer below. For
your info, since posting my initial comment I have sent out my steps to 7 different people and we can all reproduce the problem. This comes to 10 different machines (3 of them mine then the other guys) in 3 different businesses / domains. We see the same errors
in event viewer.
Under "Windows Logs" --> "Application" : I get two separate error events the first reads "Activation of app winstore_cw5n1h2txyewy!Windows.Store failed with error: The app didn't start. See the Microsoft-Windows-TWinUI/Operational log for additional
information." The second arrives in the log about 15 seconds after the first and reads "App winstore_cw5n1h2txyewy!Windows.Store did not launch within its allotted time."
Under "Windows Logs" --> "System" : I get one error that reads "The server Windows.Store did not register with DCOM within the required timeout."
Under "Applications And Services Logs" --> "Microsoft" --> "Windows" --> "Apps" --> "Microsoft-Windows-TWinUI/Operational" : I get one error that reads "Activation of the app winstore_cw5n1h2txyewy!Windows.Store for the
Windows.Launch contract failed with error: The app didn't start."
If you require any further information just let me know and I will provide as much as I can.
Thanks

Generate one instance for all users in a Group

Hi All,
We have BO 3.1 with FP 2.6.
There is a report that needs to be schedule in such a way that it will generate only one instance for all the users belonging to u201CGlobalu201D community while for rest of the users wonu2019t see any instance of it. They will simply run the report on demand providing the prompt values.
I am able to run the report for whole u201CGlobalu201D group (using u2018schedule foru2019) option but the problem is it creates as many instances as the number of users I have in this group. As users of this group have same data level access, so primarily theyu2019ll get same report output. That is why it is not desirable to run this report for all of them instead there should be a way it creates one instance after each schedule for the whole community.
Please suggest.
Thanks,
Chandra

Hi Chandra,
I think there are still ways to use publications to achieve what you want.
For instance, I created the below scenario:
- Created 3 users (pub1, pub2, pub3) and put them in group (publication users)
- I created a publication and specified the "publication users" group as the enterprise recipients
- Set the destination to be Default & BusinessObjects Inbox. Set the "Send As" option to Shortcut instead of copy.
- Under Advanced, I set it to one db fetch for all recipients.
- Scheduled this and it ran the report once, and created a shortcut to that object in each users inbox (user in the group specified)
This meets most of your criteria. The rest of it should be handled by security on the objects/folders.
Another option may be Object Packages with the proper security setup.
Thanks
JB

Local pfile for all RAC instances

Similar Messages

Maybe you are looking for