Re: Question about changing GI Home

From: Chad Cleveland <Chad.Cleveland_at_datavail.com>
Date: Tue, 10 Jun 2014 13:56:06 +0000
Message-ID: <CFBC6741.E3C3%chad.cleveland_at_datavail.com>

AGREED! The documentation is so sparse…I bet they’re thinking…no one would really move a cluster…it’s up, it’s running, don’t touch it! The MOVE states to run step 3 in the clone process and I found another small blurb that states “don’t run root.sh”. I ran through five or six different scenarios…including doing things Oracle says don’t do.

This is a live issue at my customer’s site and I needed to prove the method. They have 8 databases running on this cluster and a complete reinstall of the GI and restoration of the databases isn’t an option.

Per your suggestion, I’ve put my steps (truncated output) inline to this response. I will be willing to send my document to any parties interested…Please email me directly.

Thanks and have a great day!
Chad Cleveland

Attempting to migrate GI Home following documentation at:

http://docs.oracle.com/cd/E11882_01/install.112/e41961/rem_orcl.htm#CWLIN2955

Moving /u01/app/grid/11.2.0.4 to /u01/app/grid/11.2.0.3

PROCESS: Relocate Grid Infrastructure and bring all services online under new home.

Steps followed with results.

1. Detach Grid Infrastructure home

[grid_at_rac01 bin]$ ./detachHome.sh -silent -local -invPtrLoc /u01/app/grid/11.2.0.4/oraInst.loc

2) As Root, Unlock and move the Grid binaries from old GI Home Location to new GI Home Location

[root_at_rac01 ~]# cd /u01/app/grid/11.2.0.4/crs/install/

[root_at_rac01 install]# perl rootcrs.pl -unlock

 [root_at_rac01 /]# mv /u01/app/grid/11.2.0.4 /u01/app/grid/11.2.0.3

[root_at_rac01 ~]# cd /u01/app/grid/11.2.0.3/

3) Clone the Oracle Grid Infrastructure installation, using the instructions provided in "Creating a Cluster by Cloning Oracle Clusterware Step 3: Run the clone.pl Script on Each Destination Node," in Oracle Clusterware Administration and Deployment Guide<http://docs.oracle.com/cd/E11882_01/rac.112/e41959/clonecluster.htm#CWADD92126>.

[grid_at_rac01 bin]$ cd /u01/app/grid/11.2.0.3/clone/bin

[grid_at_rac01 bin]$ pwd

/u01/app/grid/11.2.0.3/clone/bin

[grid_at_rac01 bin]$ perl clone.pl -silent ORACLE_BASE=/u01/app/oracle ORACLE_HOME=/u01/app/grid/11.2.0.3 ORACLE_HOME_NAME=Ora11g_gridinfrahome1 INVENTORY_LOCATION=/u01/app/oraInventory -O'"CLUSTER_NODES={rac01, rac02}"' -O'"LOCAL_NODE=rac01"' CRS=TRUE

4) 4) Relink the Oracle Clusterware and Oracle ASM binaries

  [root_at_rac01 ~]# sudo su - grid
[grid_at_rac01 ~]$ cd $ORACLE_HOME/bin
[grid_at_rac01 bin]$ pwd

/u01/app/grid/11.2.0.3/bin
[grid_at_rac01 bin]$ ./relink

writing relink log to: /u01/app/grid/11.2.0.3/install/relink.log

5) Start up the new home location

[grid_at_rac01 bin]$ sudo su - root
[root_at_rac01 ~]# cd /u01/app/grid/11.2.0.3/rdbms/install/
[root_at_rac01 install]# ./rootadd_rdbms.sh
[root_at_rac01 install]#

[[root_at_rac01 install]# pwd

/u01/app/grid/11.2.0.3/crs/install
[root_at_rac01 install]# perl rootcrs.pl -patch -destcrshome /u01/app/grid/11.2.0.3
Using configuration parameter file: ./crsconfig_params /u01/app/grid/11.2.0.4/bin/crsctl does not exist to proceed stop Grid Infrastructure CRS-4123: Oracle High Availability Services has been started.
[root_at_rac01 install]#

6) Validate Services:

[grid_at_rac01 bin]$ crsctl stat res -init -t

At this time all services are backonline on Node1 and running from new grid location.

[grid_at_rac01 bin]$ ps -ef | grep crs

root     26043     1  1 13:13 ?        00:00:03 /u01/app/grid/11.2.0.3/bin/crsd.bin reboot
grid     26888 23323  0 13:16 pts/2    00:00:00 grep crs

[grid_at_rac01 bin]$

7) Repeat steps 1 through 6 on Second node.

      8) At this time the cluster is backonline and all services are running as we expect. I would like to issue a reboot to validate the cluster is indeed in order.

****Caution**** if you reboot now, you will be in trouble.

     This is the third execution I’ve attempted with this relocate process. I’ve rebooted the nodes and CRS, ASM, and Databases failed to start. No logs were written to the Grid Infrastructure home. An SR was opened with Oracle to no avail.

I was able to fix the issue by modifying the init processes used on startup and shutdown. Two files are used by the cluster and both were pointing to the old home. As root, make copies and edit each to reflect the correct location of GI Home. Make sure to do on both nodes.

cd /etc/init.d

[root_at_rac01 init.d]# ls -latr *ohasd*

-rwxr-xr-x 1 root root 6714 Jun 5 13:48 ohasd

-rwxr-xr-x 1 root root 8794 Jun 5 13:49 init.ohasd

 [root_at_rac01 init.d]# grep grid *ohasd*

init.ohasd:ORA_CRS_HOME=/u01/app/grid/11.2.0.4

init.ohasd:CLSECHO="/u01/app/grid/11.2.0.4/bin/clsecho"

init.ohasd:PERL="/u01/app/grid/11.2.0.4/perl/bin/perl -I${ORA_CRS_HOME}/perl/lib"

init.ohasd:CRSCTL=/u01/app/grid/11.2.0.4/bin/crsctl

init.ohasd: /u01/app/grid/11.2.0.4/bin/ohasd restart &

init.ohasd

: $SU root -c "/u01/app/grid/11.2.0.4/bin/ohasd restart &"

ohasd.:ORA_CRS_HOME=/u01/app/grid/11.2.0.4

ohasd.:CRSCTL=/u01/app/grid/11.2.0.4/bin/crsctl

[root_at_rac01 init.d]#

[root_at_rac01 init.d]# cp ohasd ohasd.chad

[root_at_rac01 init.d]# cp init.ohasd init.ohasd.chad

1. Reboot and validate services are online.

[root_at_rac01 init.d]# reboot

Broadcast message from root_at_rac01.localdomain

            (/dev/pts/1) at 8:15 ...

The system is going down for reboot NOW!

Process Complete.

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

Received on Tue Jun 10 2014 - 15:56:06 CEST