Performing a Reduced Downtime Upgrade
If you are upgrading from Fusion Middleware 12c (12.2.1.3) release, you can use this process to upgrade your multi-node domain without shutting down all of the servers at the same time.
The procedures described in this section are based on the Oracle Fusion Middleware Standard Installation Topology (SIT) and require that you have a multi-node environment. The standard installation topology for Oracle Fusion Middleware Infrastructure has a standard WebLogic Server domain that contains an Administration Server and a cluster containing two Managed Servers. Host 1 is used to describe the procedures performed on the host with the Administration server and Host 2 is used to describe the procedures performed on the other managed server host(s). If you have more than two hosts in your environment, be sure to complete the procedures on each additional node.
Required Tasks that Must be Completed Before You Begin
Review the following before you begin a reduced downtime upgrade:
- Review the preupgrade checklists for the components in your deployment. The checklists are found in each of the component-specfic upgrade guides. Some products may require additional steps before performing the upgrade.
- Create a complete backup of the Oracle home (on all of the nodes), the entire domain directory (on all of the nodes) and component schemas before performing the upgrade. In addition, Oracle recommends that you create a backup of UI customizations and the applications directory, in addition to domain directory. See Creating a Complete Backup.
- Make sure that you are using the appropriate JDK version for this release. For this release the correct version is jdk1.8.0_211
- If you are upgrading a shared component directory, back up the contents of the shared directory before the upgrade. The configuration upgrade makes changes to these directories.
- Make sure that your backups include any modified scripts, such as
setStartupEnv.sh, for example. The upgrade will overwrite any customized files and you will lose your changes.
Parent topic: About the Reduced Downtime Upgrade Process
Performing the Upgrade on Host 1
Perform the following tasks on the machine that hosts the Administration server and serves as the primary machine for your deployment.
- Stopping Components, Servers and Processes on Host 1
You must shut down all of the system components, processes, servers (including the Administration Server and any managed servers), and the node manager (if running). - Uninstalling the Software
When performing a rolling upgrade, an empty directory is required for installing the new binaries prior to upgrading. - Installing the Product Distributions
Before starting your upgrade, uninstall the software from the existing Oracle home, then use the Oracle Universal Installer to install the 12c (12.2.1.4.0) product distributions into the same Oracle home on the target system. You must install the product distributions on each host during the upgrade. - Restarting Node Manager, Administration Server, Managed Servers and Components on Host 1
After the upgrade, you must restart the components, servers, and processes in the correct order.
Parent topic: Performing a Reduced Downtime Upgrade
Stopping Components, Servers and Processes on Host 1
You must shut down all of the system components, processes, servers (including the Administration Server and any managed servers), and the node manager (if running).
Note:
The procedures in this section describe how to stop components, servers, and processes using the WLST command line or a script. You can also use the Oracle Fusion Middleware Control and the Oracle WebLogic Server Administration Console.- System Components (if any)
- Managed Server(s)
- Administration Server
- Node Manager
Parent topic: Performing the Upgrade on Host 1
Stopping System Components
To stop system components, such as Oracle HTTP Server, use the stopComponent script:
DOMAIN_HOME\bin directory and execute the following script for each component:./stopComponent.sh component_name Stopping the Managed Servers
stopManagedWebLogic script:
Uninstalling the Software
When performing a rolling upgrade, an empty directory is required for installing the new binaries prior to upgrading.
Note:
This task applies only to standalone Oracle HTTP Server. You cannot uninstall Oracle HTTP Server when it is collocated in a WebLogic Server domain, as partial deconfiguration of a configured domain is not supported.Note:
You must deinstall the upperstack components first, then deinstall JRF. After deinstalling the JRF, back up any remaining files and then delete all files in the directory. The installation directory must be empty.Follow the instructions in this section to remove the software from the existing ORACLE_HOME. You will reinstall the new software into this same directory.
To start the Oracle Universal Installer in deinstallation mode, execute the following command:
UNIX: ORACLE_HOME/oui/bin/deinstall.sh
Windows: ORACLE_HOME\oui\bin\deinstall.cmd
If you want to uninstall the product in a silent (command-line) mode, see Running the Oracle Universal Installer for Silent Uninstallation in Installing Software with the Oracle Universal Installer.
Parent topic: Performing the Upgrade on Host 1
Installing the Product Distributions
Before starting your upgrade, uninstall the software from the existing Oracle home, then use the Oracle Universal Installer to install the 12c (12.2.1.4.0) product distributions into the same Oracle home on the target system. You must install the product distributions on each host during the upgrade.
Note:
When Infrastructure is required for the upgrade, you must install the Oracle Fusion Middleware distribution first before you install other Fusion Middleware products.-
Make sure that you have uninstalled the existing product distributions before installing the 12c (12.2.1.4.0) distributions.
- Use the same install path (Oracle home) for the installation of the new distributions that you used for the previous distributions that you uninstalled.
Parent topic: Performing the Upgrade on Host 1
Restarting Node Manager, Administration Server, Managed Servers and Components on Host 1
After the upgrade, you must restart the components, servers, and processes in the correct order.
Note:
The procedures in this section describe how to start servers and processes using the WLST command line or a script. You can also use the Oracle Fusion Middleware Control and the Oracle WebLogic Server Administration Console. See Starting and Starting Administration and Managed Servers and Node Manager in Administering Oracle Fusion Middleware.The components must be started in the following order:
- Node Manager
- Administration Server
- Managed Server(s)
- System Components
Note:
If you are unable to successfully start any of the following components on Host 1, do not proceed with the upgrade on the remaining Hosts. You must first resolve the issues with the components on Host 1.Note:
Windows Users Only: When restarting the servers on a Windows operating system, the upgraded domain might fail with a parsing exception. To fix this parsing error, add the property "-Doracle.xml.schema/Ignore_Duplicate_components=true" to the server startup script setDomainEnv.cmd.
Parent topic: Performing the Upgrade on Host 1
Performing the Upgrade on Host 2
Once you have completed the upgrade on host 1, perform the following steps on each additional host in your environment. Our standard topoloy example includes only two hosts, but you may have more.
- Stopping the Components, Servers and Processes on Host 2
You must stop the system components, managed servers and node manager running on Host 2. - Uninstalling the Software
When performing a rolling upgrade, an empty directory is required for installing the new binaries prior to upgrading. - Installing the Software into the Existing Middleware Home
Install the software into the empty Middleware home. - Restarting the Managed Servers and Processes
After the upgrade is complete on Host 2, restart the managed servers.
Parent topic: Performing a Reduced Downtime Upgrade
Stopping the Components, Servers and Processes on Host 2
You must stop the system components, managed servers and node manager running on Host 2.
Parent topic: Performing the Upgrade on Host 2
Stopping System Components
To stop system components, such as Oracle HTTP Server, use the stopComponent script:
DOMAIN_HOME\bin directory and execute the following script for each component:./stopComponent.sh component_name Uninstalling the Software
When performing a rolling upgrade, an empty directory is required for installing the new binaries prior to upgrading.
Note:
This task applies only to standalone Oracle HTTP Server. You cannot uninstall Oracle HTTP Server when it is collocated in a WebLogic Server domain, as partial deconfiguration of a configured domain is not supported.Note:
You must deinstall the upperstack components first, then deinstall JRF. After deinstalling the JRF, back up any remaining files and then delete all files in the directory. The installation directory must be empty.Follow the instructions in this section to remove the software from the existing ORACLE_HOME. You will reinstall the new software into this same directory.
To start the Oracle Universal Installer in deinstallation mode, execute the following command:
UNIX: ORACLE_HOME/oui/bin/deinstall.sh
Windows: ORACLE_HOME\oui\bin\deinstall.cmd
If you want to uninstall the product in a silent (command-line) mode, see Running the Oracle Universal Installer for Silent Uninstallation in Installing Software with the Oracle Universal Installer.
Parent topic: Performing the Upgrade on Host 2
Installing the Software into the Existing Middleware Home
Install the software into the empty Middleware home.
Uninstall the software from the existing Oracle home before installing the 12c (12.2.1.4.0) product distributions. Install the software using the same process you used to install on Host 1.
Parent topic: Performing the Upgrade on Host 2
Restarting the Managed Servers and Processes
After the upgrade is complete on Host 2, restart the managed servers.
Parent topic: Performing the Upgrade on Host 2
Validating the Upgrade
After you have completet the upgrade on all hosts, complete the standard upgrade verification tasks to ensure that all components will continue to work as expected.
See Tasks to Perform After Upgrade.
Note:
Only perform those tasks that pertain to your environment, configuration and preferences. These tasks are meant to assist you in verfiying that the upgrade was successful. You may need to perform additional testing based on your configuration.Recovering from a Failed Upgrade
If your upgrade is unsuccessful, you will need to restore your environment from backup. Make sure that you include your backed up configuration and script files. Restore the backup of the Oracle home (on all of the nodes), the entire domain directory (on all of the nodes) and component schemas. In addition, you will need to restore any UI customizations and the applications directory, in addition to domain directory.
Parent topic: Performing a Reduced Downtime Upgrade