Updating Oracle Exadata Infrastructure

About Oracle Exadata Infrastructure Patching

Oracle Fleet Patching and Provisioning supports rolling patches for Exadata storage cells, network switches, and database nodes. Patching can be driven from the FPP Server with passwordless root SSH access, or from an FPP Client after deploying the required image. Both workflows include evaluation prechecks and use rhpctl update exadata for execution; cell server and database node patches can be rolled back, while switch patches can be downgraded if necessary.

Updating Oracle Exadata Storage Server

Updating an Exadata storage server requires importing an EXAPATCHSOFTWARE image with the exact patchmgr-compatible version, deploying it to the client cluster when patching from an FPP Client, and running rhpctl update exadata first with the eval option for prechecks and then without it to apply the update. The operation targets specified storage cells and may include patch manager options and a driving system. The resulting image can be verified with rhpctl query image.

Rolling Back Oracle Exadata Storage Server Patch

To roll back a successfully updated Exadata storage server, first verify its current patch version, run rhpctl update exadata with the target cells, image, client, and patch manager arguments, and then verify that the storage server has returned to its previous software version.

Updating Oracle Exadata Database Node

Updating Exadata database nodes requires the latest Patch Manager, an imported EXAPATCHSOFTWARE image containing the required update files, and image deployment to the client when patching from an FPP Client. Perform evaluation checks, ensuring the image path includes both the database node ISO ZIP and dbnodeupdate.zip or specify patchmgrloc, then back up the current configuration and apply the update with rhpctl update exadata. The operation supports node batches, automatic discovery of active nodes, and full live updates for Exadata 24.1 and later; results can be reviewed with rhpctl query client.

Rolling Back Oracle Exadata Database Node Patch

To roll back an Exadata database node patch, check the current patch version, execute rhpctl update exadata with the affected nodes, image, client, Patch Manager location, and rollback option, and then verify that the previous database node software version has been restored.

Updating Oracle Exadata RoCE Switches

Updating Exadata RDMA over Converged Ethernet switches involves importing an exact patchmgr-compatible EXAPATCHSOFTWARE image, deploying it to the client cluster when required, and running rhpctl update exadata with the RoCE switch list and eval option for prechecks. Run the command again without eval to apply the firmware, then use rhpctl query image to confirm that the image is identified as an Exadata network switch image.

Downgrading Oracle Exadata RoCE Switches

To downgrade successfully updated RoCE switches, verify the current firmware version, perform pre-downgrade checks with rhpctl update exadata and the downgrade plus eval options, and then repeat the command without eval to apply the downgrade. Confirm the final firmware version after completion.

Updating Oracle Exadata InfiniBand Switches

Updating Exadata InfiniBand switches requires importing an exact patchmgr-compatible EXAPATCHSOFTWARE image and deploying it to the client cluster when patching from a client. Run rhpctl update exadata with the InfiniBand switch list and eval for prechecks, then repeat without eval to apply the firmware. Specify patchmgrdrivingsystem when the FPP Server lacks direct switch access and an intermediary dom0 node is required; image type can be verified with rhpctl query image.

Downgrading Oracle Exadata InfiniBand Switches

To downgrade InfiniBand switches, verify the current firmware, run pre-downgrade evaluation with rhpctl update exadata using the downgrade and eval options, and then execute the same command without eval to perform the downgrade. Check the firmware version afterward to confirm success.

Combined Oracle Exadata Database Server and Grid Infrastructure Update

The combined update integrates Exadata database node patching with Grid Infrastructure patching, reducing downtime by requiring only one database shutdown and startup sequence per node. Create the Exadata update image once on the FPP Server, deploy it to each target server without downtime, and use rhpctl move gihome to process database node patching and Grid Infrastructure home movement in batches. FPP patches each database node first with patchmgr and then patches Grid Infrastructure, supports parallel processing within batches, and allows the operation to resume after recovery from patchmgr failures or timeouts.
Read More: Combined Oracle Exadata Database Server and Grid Infrastructure Update