Solaris

Hi,
I've this Oracle database version:
Connected to:
Oracle Database 11g Enterprise Edition Release 11.2.0.4.0 - 64bit Production

Oracle is install in a SunOS DBSERVER1 5.11 11.2 sun4v sparc sun4v cluster (Not RAC environment)
DBSERVER1 is node active and DBSERVER2 is node inactive


Now I need to shutdown Oracle DB and I'd like to Know if I have to put in
before the freeze cluster or I can do the shutdown normally issue the command "shutdown immediate".
I have another cluster redhat and I follow the below sequence to shutdown oracle DB:

Code:

1) clusvcadm -Z <service_name>
2) sqlplus / as sysdba
shutdown immediate
startup
3) clusvcadm -U <service_name>

In SunOS cluster What is the correct set of sequence to shutdown oracle DB?

Thanks in advance!

I'm confused - the example you gave restarted the oracle instance.

Generally during an oracle install, system startup and shutdown routines:
/var/opt/oracle/oratab are created to handle this. sysadmins do nothing except cluster and system shutdown/reboot stuff.

Is your oracle instance running on a Solaris cluster - it looks that way.

If what you showed is how you handle a reboot, then remove the sqlplus restart command and the clusvcadm -U command in order to completely power down.

I am not that well versed with SunOS clusters, but we use a lot of Oracle on AIX too. Doesn't the listener have to be shut down too when you want to completely bring the DB down, like this:

Code:

$ lsnrctl stop <listener_name>
$ sqlplus / as sysdba
sql> shutdown immediate
sql> ^D
$

Furthermore, in AIX both the "lsnrctl" and the "sqlplus" has to be issued as a specific user, which is NOT "root" usually. You will have to prepend all these commands be either an su - <ora_adm> -c or even by sudo su - <ora_adm> -c if the situation is the same in SunOS.

I hope this helps.

bakunin

One listener is all that is needed for a single db server with lots of other oracle instances. Bakunin is correct about the listener if the oracle db is the only one. Otherwise if you stop the listener no application can run against other instances of oracle. But there is not enough information in the question to tell.

IF you are actually freezing a single resource, the oracle db, I'm not sure.

You should not issue shutdown commands from instance on Solaris Cluster.

The clusterware will see this as failed resource group and will failover on another node.

What you need to do is :

Identify the cluster group resource which is responsible for resources.
clrg show and clrs show will tell you that.

After you have identifed the cluster group(s) responsible for database you will just need to do a clrg offline group-name

This will stop everything related to database (every resource in resource group[s]), umount filesystems if used, stop ASM etc. -> depending on the resources configuration and database resources setup).

If you want to tinkle around the database while resource group is online you will need to put its resources in unmonitor state clrg unmonitor group-name .
Be sure to return them to monitor after you are done, or risk failover not happening.

Test this on virtualbox cluster or test enviroment, setup a small cluster and get into resource groups and resources

I haven't done this with oracle databases, but i have with NFS and HA filesystem on sun cluster.


Hope that helps.
Regards
Peasant.