Separate ASM HOME in RAC

Dear Experts,
I have created separate ORACLE & ASM home in our 2-NODE RAC environment. The question I have is when I add 3rd NODE to the Cluster I know I need to run 'addnode.sh' from ORACLE_HOME but since I have separate ASM_HOME also do I need to do the same from ASM_HOME?
thanks

Yes this needs to be done for both OHs (and from both OHs), to keep the oraInventory in sync as well as ensure the OCR has the consistent/approp OH created for each resource (rdbms and asm)

Similar Messages

Separating ASM home and DATABASE home?

I just installed a new 2 node cluster on linux Itanium and I have configured my CLUSTERWARE_HOME and DATABASE_HOME. We are using ASM for our database storage and I had used dbca to create the ASM instances and when I ran the dbca to create the new ASM instance I used the dbca from the DATABASE_HOME.
I have been told that I should have created an additional home for the ASM instance? Is this correct and is this a best practice for RAC and ASM?
Everything is up and running fine and I and pondering if it is worth reconfiguring so that we can have a separate ASM home?
Is there a good doc out there concerning how to reconfigure the home for ASM once it is up and runnig and my database is currently utilizing it for it's storage?
I know that I would just have to lay down the binariies into a new directory but what alse needs to change? Will this affect how the ASM instances are registered with the clusterware?
What is the real benefit of having them separate? I have been told that you can patch then separately but I don't see the benefit as the db would have to be down anyhow if the ASM instance would need to be patched?
Any suggestions are greatly appreciated.
Thanks.

Hi,
<< Oracle Documentation >>
Keep the following in mind when installing ASM:
When running more than one database instance on a single server or node, it is recommended that you install ASM in its own Oracle home on that server or node. This is advisable even if you are running only one database instance but plan to add one or more database instances to the server or node in the future.
With separate Oracle homes, you can upgrade and patch ASM and databases independently, and you can deinstall database software without impacting the ASM instance.
If an ASM instance does not already exist and you select the OUI option to install and configure ASM only, OUI installs ASM in its own Oracle home.
If you are running a single database instance on a server or node and have no plans to add one or more database instances to this server or node, ASM and the database can share a single Oracle home
If an ASM instance does not already exist and you select the OUI option to create a database that uses ASM for storage, OUI creates this single-home configuration.
Have a look at the following documents:
1) ML NOTE:265633.1 ASM Technical Best Practices
2) http://julian.dyke.users.btopenworld.com/com/Blog/2007_01_01_archive.html
3) http://www.dell.com/downloads/global/solutions/Oracle_DB_Deployment_with_Separate_ASM_and_Database_Home_Final_Web_Ready.pdf
4) http://download.oracle.com/docs/cd/B19306_01/server.102/b14231/storeman.htm#i1021337

ASM in it own asm home under separate asm user

Hi,
I would kindly ask those who are familiar with ASM on OEL/RHEL installing it
1. in its own asm home
2. under separate asm user
to look into this thread in ASM forum
Installing ASM single instance in its own home, separate from ORACLE_HOME
Any comment, suggestion or solution would be appreciated. To avoid duplicating threads, please make your comments there (on ASM forum).
Thanks in advance.
NJ

sergeko wrote:
I would use the same user id.Thanks. I know that. No problem with ASM home under the same user id as DB instance. But why did Oracle put this demo on OTN site?
http://www.oracle.com/technology/obe/demos/10gr2/asm_install_diff_home.viewlet/asm_install_diff_home_viewlet_swf.html
This is a guaranteed non-working example of ASM implementation and the central point of confusion.
NJ

Patching Strategy for CRS and ASM homes

I'm fairly new to RAC/ASM and haven't performed any patch set upgrades yet. Back in the simple days when I wanted to apply a patch set to a database, say from 10.2.0.4 to 10.2.0.5, I would create a brand new Oracle home ahead of time and apply the patch set to it. I'd name my homes like this:
/opt/oracle/product/10.2.0.4/db1
/opt/oracle/product/10.2.0.5/db1
During the maintenance window I would change /etc/oratab to point the database to the new 10.2.0.5 and complete the database upgrade scripts. The advantages of this strategy:
1 - Less risk installing software as nothing uses the new home yet. If something goes wrong in the install, no big deal. Research the problem and try again without being under the stress of a defined maintenance window.
2 - No need to backup old home for back-out purposes.
3 - Less time required for database to be down during actual patch window since Oracle Installer does not need to run.
Now with CRS and ASM, is there a way to pre-stage a new home for those, but not have them "active" to the node until later during the maintenance window?
For ASM, it seems like it would be possible to treat the same way as database and simply update ASM SID in /etc/oratab
+ASM1:/opt/oracle/product/10.2.0.5/asm1
but I'm not totally confident in that as I'm afraid the CRS home may already have references to the ASM home in the cluster registry.
For CRS, it seems like the home is pretty well hard-wired into the node startup scripts and installing a brand new CRS home will probably disrupt the running CRS home.
Any thoughts about this?

Hi,
user5448593 wrote:
I'm fairly new to RAC/ASM and haven't performed any patch set upgrades yet. Back in the simple days when I wanted to apply a patch set to a database, say from 10.2.0.4 to 10.2.0.5, I would create a brand new Oracle home ahead of time and apply the patch set to it.
Now with CRS and ASM, is there a way to pre-stage a new home for those, but not have them "active" to the node until later during the maintenance window?Although you have not mentioned the version you are actually on, it is a quite up-to-date question and dilemma.
Starting with 11.2 for Grid Infrastructure only "out-of-place" patchset upgrades are supported.
>
For ASM, it seems like it would be possible to treat the same way as database and simply update ASM SID in /etc/oratab
+ASM1:/opt/oracle/product/10.2.0.5/asm1
but I'm not totally confident in that as I'm afraid the CRS home may already have references to the ASM home in the cluster registry.
For CRS, it seems like the home is pretty well hard-wired into the node startup scripts and installing a brand new CRS home will probably disrupt the running CRS home.
Any thoughts about this?As of 11gR2 the ASM is part of the Grid Infrastructure, therefore it is running from the same home and not recommended to separate them. (although you can do that)
By the way, what is your upgrade path? It could be easier to answer your questions if we knew that as there has been a quite a few enhancements and changes in the upgrade/patching process from 10g to 11g. (even between 11gR1 and 11gR2)
Regards,
Jozsef

ASM install not RAC aware

SPARC:
installed 11g clusterware using crs user, oinstall and dba group,
then install 11g ASM using oracle user, but it does not show RAC option.
ASM install not RAC aware if not use same user?
Thanks.

I'm not sure if you're talking about creating an ASM instance? The ASM software install is the same as the Database install. Oracle recommends running ASM from a separate Home which would require a separate Oracle database install.

Listener starting out of ASM home?

Hello,
I have a fresh install of 10.2.0.3 RAC running on Solaris 2.9.
We have older 10.2.0.3 RAC installs where the ASM HOME and the RDBMS HOME(10.2.0.3) shared the same home and our Listener started under the .../10.2.0/network/admin directory like any other stand-alone database.
We separated the ASM HOME and RDBMS home on our new install and the Listener now automatically starts out the ASM HOME instead during bootup.
Is it okay to run the Listener out of the ASM HOME? It appears to work but our GRID is finding Listeners under both spots (RDBMS and ASM HOME)
Can I changed it to start out of the RDBMS home like all our other Listeners? If so, where does Oracle look during bootup of RAC to set this and how can I change it?

Thanks for the quick response Kevin.
I've ran your command and found:
NAME=ora.snapdragon1.LISTENER_SNAPDRAGON1.lsnr
TYPE=application
ACTION_SCRIPT=/u01/app/oracle/product/asm/bin/racgwrap
NAME=ora.snapdragon2.LISTENER_SNAPDRAGON2.lsnr
TYPE=application
ACTION_SCRIPT=/u01/app/oracle/product/asm/bin/racgwrap
So it appears that the variable ACTION_SCRIPT is being set to the ASM home during CRS startup which would explain why it's using that home.
Is there a way to change that variable? I'd like to start the listeners out of the RDBMS home like all my other databases.
I'll dig around and post back here if I find it first.
Thx,
Gary

Do I need to patch ASM home?

Hi,
I have a Linux 4 server which has ASM HOME and RDBMS HOME and both are on 10.2.0.2. I want to apply the latest CPU. Do I need to apply CPU on ASM home too. This is not a RAC environment.
--MM

I am sorry about the subject of this thread, its sounds confusing, Anyway. it should have said "Do I need to apply quaterly CPU patches to ASM_HOME too?"
--MM

Localconfig reset in ASM home

In our NON-RAC environment, we have 3 different database instance and one common ASM instance.
let assume if database homes on the same server below
1. /u01/app/prod/product/10.2.0/db_1
2. /u01/app/devel/product/10.2.0/db_1
3. /u01/app/test/product/11.1.0/db_1
Common ASM instance to all three databases
/u01/app/asm/product/10.2.0/asm_1
1. $ more /var/opt/oracle/ocr.loc
ocrconfig_loc=/u01/app/test/product/11.1.0/db_1/cdata/localhost/local.ocr local_only=TRUE
2. ps -ef |grep css
/u01/app/test/product/11.1.0/db_1/bin/ocssd.bin
From all the above i understand the ocssd.bin is pointing to one of the database instance but not to ASM instance
Do ocssd.bin should point to ASM instance? is ocssd.bin has been run under wrong ORACLE HOME instead of ASM home?
Does always localconfig reset needs to be run under $ORACLE_HOME/bin or $ASM_HOME/bin?
Thanks all in advance

Rakhi -
It seems when you executed localconfig add from root, you choose below home
/u01/app/test/product/11.1.0/db_1/bin/localconfig
I checked for localconfig file on my 11g installation, it seems ORACLE_HOME is hardcoded in this file.
<code>
[root@js bin]# more localconfig
#!/bin/sh
# localconfig.sbs for local CSS
# This script file should be run for following three cases:
# 1) add: When you are configuring the CSS for the first time
# 2) delete: When there is only one single instance Oracle home on a node and
# you are deleting/deinstalling that Oracle home
# 3) reset: When there are more than two single instance Oracle homes on a
# node and you are deleting/deconfiguring the Oracle home that
# currently runs CSS daemon but you want CSS to run from a
# different Oracle home
# NOTE: When using 'reset' option. This script can be run from
# source Oracle home or destination (Where CSS is to be reset to) Oracle
# home.
SILENT=%s_silent%
CSS_ORACLE_HOME=/u01/app/oracle/product/11g
CSS_ORACLE_OWNER=oracle
CSS_DBA_GROUP=dba
CSS_LANGUAGE_ID=AMERICAN_AMERICA.WE8ISO8859P1
CSS_PARAMS="CSS_ORACLE_HOME CSS_ORACLE_OWNER CSS_DBA_GROUP CSS_LANGUAGE_ID"
</code>
Regards

Pros and cons for sharing ASM home

Hello,
Had a question on RAC with ASM...the ASM home is using the DB home...i mean ASM is sharing DB home...how bad can this be ???...i do not have a ASM home or it was never created...So how can this be resolved...or can i still use the db home for my ASM home??? what are the pros and cons for this ??

So you are telling me that, if this 3 node RAC
cluster which only have 1 database should be
fine..and what about when its come to patching ???
any big problems on that ???If the ASM needs to be patched, the database probably needs to come down. If the database needs to be patched, the database probably needs to come down.
These are true whether the ASM is in the same home or not.
The challenge comes when you have 2 or more databases (2 DB instances). If the ASM goes down for patching, both DBs probably come down.
But if only one DB needs to be patched, and the Home is ALSO the ASM Home, both DBs come down.
So separating ASM home and DB home is irrelevant when one DB is used, but is very relevant when multiple DBs are used and each DB has it's own ORACLE_HOME.

Problem starting ASM on non-RAC environment on HP-UX 11.11

I am trying to start ASM on single node running multiple database as I am planning to test setting up on of the db files with ASM.
Wen I tried to start the ASM after I created the init+ASM.ora and exported the ORACLE _SID to +ASM; I got the error below;
SQL> startup
ORA-29701: unable to connect to Cluster Manager
I realized the css was down and I tried to start without much success as shown below; by the way, the ASM home is the same as the Oracle 10g (10.2.0.3) binaries home.
# $ORACLE_HOME/bin/localconfig add
clscfg: EXISTING configuration version 3 detected.
clscfg: version 3 is 10G Release 2.
Successfully accumulated necessary OCR keys.
clscfg: Arguments check out successfully.
NO KEYS WERE WRITTEN. Supply -force parameter to override.
-force is destructive and will destroy any previous cluster
configuration.
Configuration for local CSS has been initialized
Failed configure CSS. Fix the problem and rerun this script from destination Oracle home with 'reset' argument
# $ORACLE_HOME/bin/localconfig reset $ORACLE_HOME
Successfully accumulated necessary OCR keys.
Creating OCR keys for user 'root', privgrp 'sys'..
Operation successful.
Configuration for local CSS has been initialized
Adding to inittab
Startup will be queued to init within 30 seconds.
Checking the status of new Oracle init process...
Expecting the CRS daemons to be up within 600 seconds.
Giving up: Oracle CSS stack appears NOT to be running.
Oracle CSS service would not start as installed
Automatic Storage Management(ASM) cannot be used until Oracle CSS service is started
# $ORACLE_HOME/bin/crsctl set css misscount 600
Failure 32 in main OCR context initialization: PROC-32: Cluster Ready Services on the local node is not running Messaging error [9]
Please advice

This is what I get when I run the localconfig add
# $ORACLE_HOME/bin/localconfig add
clscfg: EXISTING configuration version 3 detected.
clscfg: version 3 is 10G Release 2.
Successfully accumulated necessary OCR keys.
clscfg: Arguments check out successfully.
NO KEYS WERE WRITTEN. Supply -force parameter to override.
-force is destructive and will destroy any previous cluster
configuration.
Configuration for local CSS has been initialized
Failed configure CSS. Fix the problem and rerun this script from destination Oracle home with 'reset' argument

SAN reboot for oracle DB at ASM in linux RAC

Hi Experts,
we use 10.2.0.4 database in ASM at oracle RAC in red hat 5 linux.
we use 3 directoey ( asm, crs, and database)
I got notes that SAN box (support SAM in database) will be reboot.
Under this condition, what do i need to do? shutdown instance? database? crs? or asm instance?
Thanks for your help?
JIM

user589812 wrote:
what means is about start all oracle related services in sequence?CRS will start the complete Oracle cluster s/w cluster for you - ASM, RAC, nodeapps, etc.
Usually, the only effort required is simply hitting the reset/power on button - as the o/s will boot, CRS will start and it will in turn bring up the s/w stack. No manual intervention required. (unless you on purpose configured it differently)
Based on Billy suggestion, can I use srvctl stop nodeapps -n all and #ORA_CRS_HOME/bin/crsctl stop crsNo - my suggestion is that before the SAN maintenance window period start, you do a "+shutdown -h now+" on all cluster nodes to halt/powerdown each and every RAC server.
And after the SAN maintenance period is over, and the SAN available again, ssh into the LoM (Lights Out Management) console of each server and do a "+start SYS+" (or equivalent) to powerup the server.
In other words, with the SAN down/busy rebooting/undergoing maintenance, I would not want to have my RAC servers up and running as there is no storage layer to run them on. IMO, it is a lot safer to have these servers powered down to during such a maintenance period.
PS. I have even had the odd case that during SAN maintenance power cables being pulled, Interconnect switches accidentally reset and so on - or you could have some bright spark also shutting down the aircon with the SAN and your RAC servers suffering heat problems and potential damage while running. So my question is - why should I take the risk of keeping my RAC servers up when the storage layer is not there and the cluster is broken and useless? Surely it makes a lot more sense to power down those servers too and then only power them on again when the maintenance period is over and the SAN (and data centre) is in a proper running state again.

How to clean the asm instance from RAC manually

for some reason i run crs_unregister asm and crs_unregister lsnr to remove the asm and listener resource from crs yesterday
and today i want rebuild the asm instance ,so i run dbca again ,but error
Error when starting ASM instance on node rac1: PRKS-1009 : Failed to start ASM instance "+ASM1" on node "rac1", [PRKS-1011 : Failed to check status of ASM instance "+ASM1" on node "rac1", [CRS-0210: Could not find resource ora.rac1.ASM1.asm.]]
[PRKS-1011 : Failed to check status of ASM instance "+ASM1" on node "rac1", [CRS-0210: Could not find resource ora.rac1.ASM1.asm.]]
DBCA could not startup the ASM instance on node: rac1. Manual intervention is required to recreate these instances. If you choose to proceed, ASM diskgroups will not be mounted on non-started remote ASM instances. Do you want to proceed with ASM diskgroup management?
and problem is how to do this "Manual intervention is required to recreate these instances", I already do 1, dd the asm disk, 2, remove the +ASM directory from $ORACLE_BASE, 3,clean ASM info from /etc/oratab, so what i can do next ?
i try restart crs ,now the error info is different !!
[oracle@rac1 ~]$ dbca -silent -responseFile /home/oracle/dbca.rsp
Look at the log file "/opt/ora/product/10.2.0/db_1/cfgtoollogs/dbca/silent6.log"
for further details.
[oracle@rac1 ~]$ cat /opt/ora/product/10.2.0/db_1/cfgtoollogs/dbca/silent6.log
ORA-00119: invalid specification for system parameter LOCAL_LISTENER
ORA-00132: syntax error or unresolved network name 'LISTENER_+ASM1'
ORA-00119: invalid specification for system parameter LOCAL_LISTENER
ORA-00132: syntax error or unresolved network name 'LISTENER_+ASM1'
Edited by: 859340 on 2011-7-8 下午11:01

Hi,
Can you post the dbca log?
and do you share the asm home with oracle home? if it is separated then add LISTENER_+ASM1 in tnsnames in ASMHOME, otherwise add it in ORACLEHOME
Cheers

Add ASM disk in RAC

Hello everyone,
We are on 10gR2 ASM.
The system admin has provided one disk as follows:
On node-1 (tphsmsd1)
/dev/rdisk/disk83 CANDIDATE
On node-2 (tphsmsd2)
/dev/rdisk/disk77 CANDIDATE
I've read that paths of ASM disks in RAC may be different.
When I give "ls -l /dev/rdisk/disk83" gives no such files and vice-versa for disk on node2.
I don't understand how ASM handles this ? If I add disk (disk83) on node 1 how it's going to get to that disk on node 2 ?
thanks
Jitu Keshwani

Jitu Keshwani wrote:
I've read that paths of ASM disks in RAC may be different.
When I give "ls -l /dev/rdisk/disk83" gives no such files and vice-versa for disk on node2.
I don't understand how ASM handles this ? If I add disk (disk83) on node 1 how it's going to get to that disk on node 2 ?The header label of the disk, identifies the ASM disk name and diskgroup it belongs to. You can hexdump or octaldump the 1st 128 bytes of the device as ascii chars to view the label. The string "+ORCLDISK+" in the header identifies it as an ASM disk. This is then followed by the ASM diskname and then the AM diskgroup it belongs to.
But there's no reason for not having a consistent and static device name layer across all cluster nodes. Each scsi device has a WWID (World Wide ID)- a unique identifier. This means that the kernel (and kernel drivers) can uniquely identify a device.
On Linux, Multipath is used to map a logical device name to a WWID - and using the same +/etc/multipath.conf+ configuration file on all cluster nodes, ensures that the same device names are used across the cluster.
But from the the device file entry you listed, you're likely not running Linux? In that case, depending on the Unix flavour used and the type of cluster storage, there can be similar options to Linux's Multipath. One such option is EMC's PowerPath - but that of course is specific to EMC SANs and requires additional licensing fees.

Help while creating ASM instance on rac environment on Linux

Help while creating ASM instance on rac environment on Oracle enterprise Linux 4 x84_64 Update 5 (10g2).
a) Installed ASMLib 2.0 Packages
b) Configured and Loaded the ASMLib 2.0 Packages
c) Created ASM Disks for Oracle
d) I can see, three volumes already defined for ASM file system.
By running following command (on both node), I get following that means disk are detected.
/etc/init.d/oracleasm listdisks
VOL1
VOL2
VOL3
But Disk members not displayed for creating disk groups. I even tried to "Change Disk Discovery Path" button and put "ORCL:VOL*" but still no luck.
Someone told me use, use ADD DISKS button but this button is disabled for me.
Can anyone, please help me if I missing something.

*****SOLVED******
Thanks for reply..
I set the Change Disk Discovery Path" to /dev/oracleasm/disk/* instead of "ORCL:VOL* and this worked for me.

Find ASM Home

Hi,
My question is :
To find the ORACLE HOME we use the "echo $ORACLE_HOME" and to find the CRS HOME we use the "echo $ORA_CRS_HOME"; In the same way, What variable i have to use to find ASM HOME ?
$ ps -ef |grep smon
oracle 7416 1 0 Apr08 ? 00:00:00 asm_smon_+ASM1
oracle 8218 1 0 Apr08 ? 00:06:20 ora_smon_ractest1
oracle 28210 15182 0 15:23 pts/0 00:00:00 grep smon
$ echo $ORACLE_HOME
/u01/app/oracle/product/10.2.0/db_1
$ echo $ORACLE_SID
ractest1
$ export ORACLE_SID=+ASM1
$ echo $ORACLE_SID
+ASM1
$ echo $ASM_HOME >>>>>>>>>>>>>>>>>>>>>>>>
$
$ sqlplus "/as sysdba"
SQL*Plus: Release 10.2.0.3.0 - Production on Sat Jul 3 15:25:11 2010
Copyright (c) 1982, 2006, Oracle. All Rights Reserved.
Connected to:
Oracle Database 10g Enterprise Edition Release 10.2.0.3.0 - 64bit Production
With the Partitioning, Real Application Clusters, OLAP and Data Mining options
SQL> select instance_name from v$instance;
INSTANCE_NAME
+ASM1
$ cat /etc/oratab
+ASM1:/u01/app/oracle/product/10.2.0/db_1:N
ractest:/u01/app/oracle/product/10.2.0/db_1:N
------------------------------------

First of all ASM_HOME ora ORA_ASM_HOME - there are no conventions or requirements by Oracle to use this variables.
When saying "oracle home" we do not mean the home directory for 'database' type instance or the home for 'asm' type instance. There is just one term, Oracle Home - which means oracle software installation directory - not regarding which type of instance is run from this directory(i.e. from oracle home).
So, for your ASM instance also you will have a term "oracle home" which means the home directory of database binaries and other files.
Regarding to your configuration, you yourself answered to your question in your first post(question). your database and asm instances both share the same oracle home, which is shown via entries in your oratab file
$ cat /etc/oratab
+ASM1:/u01/app/oracle/product/10.2.0/db_1:N
ractest:/u01/app/oracle/product/10.2.0/db_1:N

Separate ASM HOME in RAC

Similar Messages

Maybe you are looking for