Create your own Oracle Staging Environment, Part 4: Staging the GI

Lets continue this series of posts:

Create your own Oracle Staging Environment, Part 1: The Staging Server
Create your own Oracle Staging Environment, Part 2: Kickstart File(s) and Profiles
Create your own Oracle Staging Environment, Part 3: Setup the lab environment

In Part 3 we had two (or more) VMs up and running with the following specs:

• eth0 configured as the public network
• eth1 configured as the private interconnect
• three shared storage devices for the ASM disks
• oracle users and groups
• a sample source file deployed on all Vms

this might be sufficiant for starting a lab if you want to demonstrate the installation of a grid infratructure. but we can even go further and automate the setup of the GI. what else would be required?

• the source file for the grid infrastructure (stored in /var/www/html/sources). in this case the 11.2.0.4 files downloaded from mos
• kernel parameters
• partitions on the disks
• limits for the oracle users
• setting the io scheduler
• the cluster configuration file
• the gi software installation
• password-less ssh connectivity for the grid user

here are the updated scripts (as pdfs to reduce the length of this post):

• create_gi_rac_lab.sh
• finish_gi_setup.sh (this needs to placed in the /var/www/html directory of the cobbler server)
• oracledatabaseserver.ks

place these on the cobbler server in same locations as in the previous posts, then:

./create_gi_rac_lab.sh mylab 2

fetch one of the vm setup scripts:

ls -la /var/www/html/mylab/setup*.sh
-rw-r--r--. 1 root root 3821 Jan 11 09:32 /var/www/html/mylab/setup_p_1.sh
-rw-r--r--. 1 root root 3821 Jan 11 09:32 /var/www/html/mylab/setup_p_2.sh

… and fire up the VMs on your workstation.

give each vm some minutes before starting the next one. the reason is that the last vm will do the configuration of the GI and therefore the other vms should be up and running when the configuration starts. once the last one completed you have the GI up and running on all the nodes:

[root@mylabp1vm1 ~]# /opt/oracle/product/crs/11.2.0.4/bin/crsctl stat res -t
--------------------------------------------------------------------------------
NAME           TARGET  STATE        SERVER                   STATE_DETAILS       
--------------------------------------------------------------------------------
Local Resources
--------------------------------------------------------------------------------
ora.CRS_DG.dg
               ONLINE  ONLINE       mylabp1vm1                                   
               ONLINE  ONLINE       mylabp1vm2                                   
ora.asm
               ONLINE  ONLINE       mylabp1vm1               Started             
               ONLINE  ONLINE       mylabp1vm2               Started             
ora.gsd
               OFFLINE OFFLINE      mylabp1vm1                                   
               OFFLINE OFFLINE      mylabp1vm2                                   
ora.net1.network
               ONLINE  ONLINE       mylabp1vm1                                   
               ONLINE  ONLINE       mylabp1vm2                                   
ora.ons
               ONLINE  ONLINE       mylabp1vm1                                   
               ONLINE  ONLINE       mylabp1vm2                                   
--------------------------------------------------------------------------------
Cluster Resources
--------------------------------------------------------------------------------
ora.LISTENER_SCAN1.lsnr
      1        ONLINE  ONLINE       mylabp1vm2                                   
ora.cvu
      1        ONLINE  ONLINE       mylabp1vm1                                   
ora.mylabp1vm1.vip
      1        ONLINE  ONLINE       mylabp1vm1                                   
ora.mylabp1vm2.vip
      1        ONLINE  ONLINE       mylabp1vm2                                   
ora.oc4j
      1        ONLINE  ONLINE       mylabp1vm1                                   
ora.scan1.vip
      1        ONLINE  ONLINE       mylabp1vm2                                   

there still is much room for improvement, but for now it at least works … takes around 20 minutes on my workstation until both nodes are up and configured.