Data Preserving Reprovisioning Process

What are the server archive files referred in the Data Preserving Reprovisioning process?

The Data Preserving Reprovisioning process uses a method where the system is reimaged with the ISO image of the latest Oracle Database Appliance release. The metadata about the resources that were created, such as databases, application KVMs, DB systems, Oracle Data Guard, Oracle ASR configuration, and others are first saved in an archive file. This metadata is used for reprovisioning the system after the reimage step. The saved metadata files are archived into a zip file with the name serverarchive_hostname.zip, where hostname is the name of the Oracle Database Appliance system being reprovisioned. This saved metadata file is called server archive file. You must copy this file outside the appliance before the reimage step in the Data Preserving Reprovisioning process.

What if I lose the server archive files?

The command odacli restore-node cannot run without the server archive files. The server archive files are generated when running the odacli detach-node command and contain metadata that is required to reprovision the system. The server archive file is generated and stored at /opt/oracle/oak/restore/out/. The /opt file system is deleted during the re-image step of the Data Preserving Reprovisioning process. The server archive file must be saved before starting the reimage step. Failing to do so will result in not being able to reprovision the system and Oracle Database Appliance will not be able to restart the resources such as database, DB systems, Oracle Data Guard, and others.

What are the files present in the server archive file?

Server archive files encapsulate metadata for Oracle Database Appliance resources, including databases, KVMs, DB systems, Oracle Data Guard, multi-user access, and Oracle ASR configuration. Additionally, these files encompass Oracle ASM and Oracle ACFS metadata, including volumes and file systems, as well as root file system files for both the database and Oracle Grid Infrastructure. It also includes other crucial files necessary for the seamless restoration of the system.

How many zip files are generated as server archive files?

One server archive file is generated in the /opt/oracle/oak/restore/out directory. A sha256 checksum file is also generated for the server archive file. Example:

# ls -l /opt/oracle/oak/restore/out/*.zip
-rw-r--r-- 1 root root 1454460 Dec 6 22:49 
/opt/oracle/oak/restore/out/serverarchive_oda412.zip

How do I verify if the server archive file was copied successfully out of the appliance?

After copying the server archive file outside the appliance using the scp command, verify the file sha256sum checksum with the generated checksum file. The SHA256 checksum of outputs from both command must match. Example:

$ sha256sum serverarchive_oda6m003.zip 
be7a517283e1aac89cb124fe68bc8e02b991a50ee336332af2e9fb06f753dd6f serverarchive_oda6m003.zip 
$ cat serverarchive_oda6m003.zip.sha256 
be7a517283e1aac89cb124fe68bc8e02b991a50ee336332af2e9fb06f753dd6f

Are server archive files generated for DB system upgrade also?

Yes. The server archive zip file is generated during the detach phase of the upgrade-dbsystem process and then copied to the bare metal system. After creation of Oracle Linux 8 DB system, this file is copied to the DB system, and the database is restored. For each DB system being upgraded, the storage location of its server archive zip on the bare metal system is based on its name and is, therefore, unique.

I encountered an error when running the odacli restore-node command. Can I modify the saved metadata?

No. Contact My Oracle Support for assistance in correcting any erroneous metadata.

What are the prerequisites for Oracle Grid Infrastructure restore?

Ensure system Oracle ILOM and firmware are updated. Also, update the repository with the saved server archive file and Oracle Grid Infrastructure clone file before starting the Oracle Grid Infrastructure restore process.

Can I use an earlier release of Oracle Grid Infrastructure clone file, for example 19.20 after reimaging with 19.21 ISO image, before running the odacli restore-node -g command?

No, the Oracle Grid Infrastructure clone file must be of the same evrsion as the DCS software.

Is a JSON input file required while running the odacli restore-node -g command, similar to the odacli create-appliance command?

No, a JSON input file is not required while running the odacli restore-node -g command. The Data Preserving Reprovisioning process collects the metadata during the detach of the system and uses the same metadata to reprovision the system. Ensure that you update the repository with the saved server archive file before running the odacli restore-node -g command.

Oracle Grid Infrastructure restore failed. How can I proceed?

To address the issue, review the error correlation report and the DCS agent logs. Address the reported problems and then run the cleanup.pl script on all nodes without using the -f, -erasedata, or -nodpr options. Subsequently, retry the Oracle Grid Infrastructure restore process.

Can I run the odacli restore-node -d command if there was a failure in an earlier run?

Yes. Before a rerun, the cause of the failure needs to be determined and fixed. Then rerun the command.

Are there any changes to be aware of after the system is reimaged with the Oracle Linux 8 ISO image?

If your appliance was reimaged with an Oracle Database Appliance release 19.12 or later ISO image, and you are now on a supported Oracle Database Appliance release, then you will not see any difference in the partition sizes. If your system was originally reimaged with an earlier release ISO image, then the partition sizes for /u01 and /opt are smaller after your system is reimaged with the Oracle Linux 8 ISO image.

/u01 40GB 100GB 
/opt 30GB 60GB