RE: rac reboots

Long answer: It really depends on what is configured to be done by the /etc/rc scripts. For example, in one 10.2.0.4 RAC environment here, the total sequence is automated on shutdown, but we chose to omit automatically start the database instance itself on restart. This was because a RAC node restart for us on this system is a "big deal" and when it occurs, we may want to do some diagnostics before turning the world loose in the instance. (Your mileage will likely vary significantly!)

In a 10g RAC system, we do this (or its automated equivalent):

As Oracle:
Stop database instance

As root:
## Stop the ASM instance and the listener - primarily: $CRS_HOME/bin/srvctl stop nodeapps -n <your_node_name> $CRS_HOME/bin/crsctl stop crs

## If the following command returns nothing, the clusterware is down: ## ps -ef |egrep "crsd.bin|ocssd.bin|evmd.bin|oprocd"| fgrep -v "ps -ef |egrep"

On startup, the first step would be:
As root:
$CRS_HOME/bin/oprocd start
Then just reverse the steps above - with start instead of stop.

For us, almost all of this is automated, but I don't remember now how much of it was "out of the box". Most if not all I suspect.

The relevant crs master scripts (for 10.2.0.4 on Linux) are:
/etc/init.d/init.crs - which calls (in different order depending on for startup or shutdown):
/etc/init.d/init.crsd
/etc/init.d/init.cssd
/etc/init.d/init.evmd

There are other, perhaps better, ways - using srvctl for database instance start/stop.

In 11g, there are more options - using ohasd for example...

Don Granaman | Phone: 402-361-3073 | Cell: 402-960-6955 | Fax: 402-361-3173 | Solutionary | Relevant . Intelligent . Security

-----Original Message-----
From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of ed lewis Sent: Saturday, December 10, 2011 7:37 AM To: oracle-l_at_freelists.org
Subject: rac reboots

Hi,

    I have some questions in regard to a reboot of a rac node.     In our environment, when a normal shutdown of a cluster node is done, I do not shut down anything beforehand. I feel that the scripts executed in /etc/rc directories will handle the shutdown and startups. I also think that the scripts will do this in a logical, and safe manner. I have yet to see otherwise.

In a normal system shutdown/reboot , is it necessary to manually shut down any of the oracle components: CRS, ASM and the database ?

Outside of the shutdown/startup scripts in /etc/rc that are provided with a cluster install, should any other scripts be implemented, or will the scripts provided be enough ?

I appreciate your input.

ed

Solaris 10
Oracle 10.2.0.3, 11.2.0.3

--
http://www.freelists.org/webpage/oracle-l


--
http://www.freelists.org/webpage/oracle-l