Data Preserving Reprovisioning: General

Understand general FAQs on Data Preserving Reprovisioning.

My Oracle Database Appliance deployment is running Oracle Database Appliance release 18.8. Can I directly upgrade my system to Oracle Linux 8 and Oracle Database Appliance release 19.21?

No. You must upgrade the appliance, that is on a release earlier than 18.8 to either Oracle Database Appliance release 19.17 or later first, using the steps described in the Upgrading Oracle Database Appliance Using Data Preserving Reprovisioning chapter in the Oracle Database Appliance Deployment and User's Guide in the Oracle Database Appliance release 19.20 documentation library. After you upgrade the system to Oracle Database release 19.17 or later which runs Oracle Linux 7, you can upgrade to Oracle Linux 8.

Why is Oracle Database Appliance upgrading to Oracle Linux 8 using Data Preserving Reprovisioning instead of doing an in-place upgrade?

The Linux operating system upgrade utility, Leapp, is a framework to perform in-place upgrade of Oracle Linux 7 systems to Oracle Linux 8. Leapp runs a set of preupgrade checks on the system to be upgraded before running the upgrade. These preupgrade checks impose specific requirements on the system. For example, Leapp expects access to YUM repository to update to the latest Oracle Linux 7 release before performing the actual upgrade, clearing of locked packages, disabling of secure boot and FIPS mode, kernel on Unbreakable Enterprise Kernel 6, unmounting NFS exports, and several others. Oracle Database Appliance, being a secured engineered system, employs design choices that would require extending the Leapp software to integrate with the Oracle Database Appliance software stack. For example, Oracle Database Appliance uses a very specific set of RPMss locks RPM package versions, does not assume access to YUM repositories, has FIPS support enabled, is not on Unbreakable Enterprise Kernel 6, and exports Oracle Database Appliance repository to DB systems through NFS. In addition, each Oracle Database Appliance system may have a different set of Leapp issues (also called inhibitors) which makes it difficult to support a consistent upgrade experience for all users. Conversely, with the clean install approach that Data Preserving Reprovisioning employs, you experience a uniform upgrade while continuing to support the appliance principle of Oracle Database Appliance.

For information about LEAP, see the topic Upgrading Systems With Leapp at https://docs.oracle.com/en/operating-systems/oracle-linux/8/leapp/leapp-AboutLeapp.html

What are minimum supported versions for the databases before upgrading Oracle Database Appliance bare metal system and DB system to Oracle Linux 8?

The databases must be at the following minimum versions at the minimum to upgrade to Oracle Linux 8:
  • 19.17.0.0.221018 or later: Databases can be restarted in Oracle Linux 8 environment.
  • 21.8.0.0.221018 (DB system only): DB systems with 21c databases can be restarted on Oracle Linux 8 bare metal Oracle Linux 7. DB systems with 21c databases can be upgraded to Oracle Linux 8 DB systems and retain 21c databases.

My deployment is OVM and I am on Oracle Database Appliance release 19.13. Can I upgrade my system to Oracle Linux 8 using the Data Preserving Reprovisioning process?

No. You must upgrade the appliance to either Oracle Database Appliance release 19.18 or later first, using the steps described in the Migrating from Virtualized Platform to KVM on Oracle Database Appliance chapter in the Oracle Database Appliance Deployment and User's Guide in the Oracle Database Appliance release 19.20 documentation library. After you upgrade the system to Oracle Database release 19.18 or later which has DCS software and Oracle Linux 7, you can upgrade to Oracle Linux 8.

Can I use ODABR when upgrading my system from Oracle Linux 7 to Oracle Linux 8?

See My Oracle Support Note ODA (Oracle Database Appliance): ODABR a System Backup/Restore Utility at https://support.oracle.com/rs?type=doc&id=2466177.1. The note states that ODABR may not be useful for systems upgrading from Oracle Linux 7 to Oracle Linux 8. The primary use of ODABR is to create LVM snapshots and backup data blocks that are modified while patching the system. Data Preserving Reprovisioning orchestrates an operating system upgrade of the Oracle Database Appliance system from Oracle Linux 7 to Oracle Linux 8 and involves a reimage of the system with an Oracle Linux 8 ISO image. During the reimage, the LVM is reinitialized and the snapshots on the boot drive are lost. The database homes and database files on the shared storage are not lost due to this reimage.

How much time does upgrading Oracle Database Appliance bare metal systems using Data Preserving Reprovisioning take?

The time taken for upgrading Oracle Database Appliance X7-2 is as follows:

Table 1-1 Time Taken for Patching Oracle Database Appliance Bare Metal Systems

Component being updated Time taken for the update (h:mm:ss)
Infrastructure update of bare metal systems
  • Update DCS admin: 0:00:12
  • Update DCS components: 0:03:00
  • Update DCS Agent: 0:06:18
Infrastructure update of DB systems
  • Update DCS admin: 0:00:20
  • Update DCS components: 0:04:40
  • Update DCS Agent: 0:08:00
Pre-upgrade Report 0:02:09
Detach node operation 0:14:26
Re-image with Oracle Linux 8 ISO image, configure first network, update repository with server zip, and update repository with Oracle Grid Infrastructure clone 1:30:00
Restore of Oracle Grid Infrastructure, databases, Oracle KVM, and DB systems
  • Oracle Grid Infrastructure restore: 0:43:33
  • Databases restore: 0:04:57
  • KVM restore: 0:00:24
  • DB system restore: 0:04:31
Total time taken approximately 2:40:00

The databases in my deployment are not on a supported release. How do I proceed to upgrade my bare metal system to Oracle Linux 8?

If the databases are at one of the following Release Update (RU) or later, it is recommended that you upgrade the databases to any of the following database versions before you run the odacli detach-node command.
  • 18.14.0.0.210420
  • 12.2.0.1.220118
  • 12.1.0.2.220719
  • 11.2.0.4.210119

What customizations are not restored when upgrading using the Data Preserving Reprovisioning process?

Customizations made to user profiles, kernel parameters, additional RPMs installed, or any additional services created on the databases are not restored during an upgrade using Data Preserving Reprovisioning.

I have additional operating system RPMs installed on my system. Are these restored as part of Data Preserving Reprovisioning?

No. Since the Data Preserving Reprovisioning process involves reimaging using the latest Oracle Database Appliance ISO image, any previous customizations in operating system RPMs, or any customizations in the settings or configurations of services in the operating system are lost. You must uninstall these RPMs and then continue with the next step in the upgrade process. You can reinstall these custom RPMs as required, after the upgrade.

I have configured VLANs on my system. Will these VLANs be configured on the upgraded system?

Yes.

Do I need to take backups of my databases before running the odacli detach-node command?

The database files are left intact on Oracle ASM disk groups and are used for bringing up the databases after successful reprovisoning. You must, however, take database backups as a safety measure.

I have a 2-JBOD configuration. Is Data Preserving Reprovisioning supported with this configuration?

Yes.

I have additional options (for example, ASO) enabled on my source databases. Are these options preserved?

If the source database home is on the root file system, then the options that are part of the Oracle software are not preserved. As part of the odacli restore-node -d command process, a new database home is created with the default option. If the source database homes are on Oracle ACFS file systems, then the options are preserved because the database home are not deleted.

Which are the supported Oracle Database Appliance releases as the source versions for Data Preserving Reprovisioning?

The source versions must be at the minimum on Oracle Database Appliance release 19.17. The DCS software and Oracle Grid Infrastructure must be on Oracle Database Appliance release 19.17 on the bare metal system. The same requirement exists for DB systems running on the bare metal systems. Note that Oracle Database Appliance release 19.20.0.1 is not a valid source version, since it already runs on Oracle Linux 8.

What happens to my database audit data after the reimage process in Data Preserving Reprovisioning?

The audit data of databases can be either in database tables or in operating system files or in SYSLOG. Only the data in database tables is preserved. You must save the audit data that is present on the root file system before initiating reimage of the system.

Can I run cleanup.pl after detaching the node on source ODA version?

No. Do not run cleanup.pl at any time during the upgrade process. This utility erases the Oracle ASM disk groups and then you cannot restore the databases. If there is a failure in running the command odacli restore-node -g, then you can use cleanup.pl to clean up the system while preserving Oracle ASM disk groups. Run cleanup.pl in DPR mode and do not use the -nodpr option.

Are all files present in database homes preserved and restored?

Local database homes are deleted during the reimage. Any files on these database homes (such as TDE wallets) are not preserved.

Can database homes be created on the root file system after reprovisioning to Oracle Database Appliance release 19.21?

No, database homes can only be created on Oracle ACFS.

How much space is required for database homes and database clones on Oracle ACFS?

For database home on Oracle ACFS volumes, the database home is preserved and all files in these homes are preserved For database clones on Oracle ACFS, a total of 150 GB is required. The space required for database homes is 15 GB multiplied by the number of database homes to be restored. The space availability is also validated at the source by the odacli create-preupgradereport prechecks.

Should I unpack Oracle Grid Infrastructure and Oracle Database clones together?

No, it is recommended that only the Oracle Grid Infrastructure clone is unpacked before running the command odacli restore-node -g. After the command odacli restore-node -g runs successfully, the Oracle Grid Infrastructure clone is moved to Oracle ACFS which has 150 GB of space.

Do the nodes restart during any of the operations?

The command odacli restore-node -g does not restart the nodes. When you run the odacli update-server command after the reimage, the node restarts.

Why does cleanup with -erasedata or -f fail after reprovisioning using Data Preserving Reprovisioning?

On Data Preserving Reprovisioning environments, the default cleanup mode is the Data Preserving Reprovisioning mode. This mode cleans up the appliance without erasing Oracle ASM disks to allow the command odacli restore-node -g to be reattempted. This mode is enabled after the command odacli update-repository completes the step to generate the server archives. This mode does not support usage of -erasedata or -f options.

Override this behavior and force regular cleanup with the -nodpr option as follows:
cleanup.pl -nodpr -erasedata / -f

When is the Data Preserving Reprovisioning tab displayed in the BUI ?

The Data Preserving Reprovisioning tab is displayed on the left-hand side navigation list in the BUI only after the DCS infrastructure is upgraded to Oracle Database Appliance release 19.21.

Why are some tabs disabled in the Data Preserving Reprovisioning screens in the BUI?

The Data Preserving Reprovisioning process comprises of the following five tabs:
  1. Re-provision
  2. Restore Databases
  3. Restore VM Instances
  4. Restore DB Systems
  5. Upgrade DB Systems

The Re-provision tab is always enabled, but tabs 2 to 5 are enabled only after the appliance is reimaged to Oracle Database Appliance release 19.21.

Can I use BUI to reimage the appliance?

No, you must reimage the appliance using the Oracle ILOM console. After you detach the nodes, the BUI prompts you to save a copy of the server archive file to a location outside of the appliance, and then proceed with the reimage.

Are there any steps to be completed, if I have LACP enabled on the netowrk interface?

You must run the odacli configure-firstnet command to enable LACP after reimaging the appliance.