Pre-upgrade information

Learn about tasks you need to perform and information you need to consider before starting a Cloudera Data Flow service upgrade.

Warnings and notes

Depending on your upgrade scenario, you may have to perform one or more of the actions listed here before launching a service upgrade.

important

Revert the patch that reduces resources for Cloudera Data Flow workload application while increasing them for Flow Designer, using the following command:

kubectl --kubeconfig kubeconfig.yaml -n dfx-local patch hr dfx-local --type=merge -p '{"spec":{"postRenderers":null}}'

The command is idempotent, it may be executed it more than once, but Cloudera Data Flow workload application will restart automatically every time the command is executed even if it is effectively a no-op, which is a designed behavior.

important

Remove the temporary trimmer cron job and garbage collect all scheduled jobs before an upgrade, using the following command:

kubectl --kubeconfig kubeconfig.yaml -n valkey delete cronjob valkey-flow-changes-stream-trimmer

The command is idempotent. You may receive the following output:

Error from server (NotFound): cronjobs.batch "valkey-flow-changes-stream-trimmer" not found

This can be ignored because it simply means there is no patch to revert for Valkey.

important

Revert the Valkey patch described in TSB 2025-871 before performing an upgrade, using the following command:

kubectl -n valkey patch hr valkey --type=merge -p '{"spec":{"values":null}}'

The command is idempotent, executing it more than once, or without applying the patch beforehand causes no issue.

important

TSB 2025-839 Mitigation steps for Cloudera DataFlow on cloud prescribed preventive steps to mitigate the effects of ‘IngressNightmare’ vulnerabilities.

The mitigation steps apply a patch to the Cloudera Data Flow environments. Before upgrading the patched environment of Cloudera Data Flow to another version, the patch must be reverted temporarily. To revert the patch, execute the following command:
```
kubectl --kubeconfig kubeconfig.yaml -n nginx-ingress patch helmreleases ingress-nginx -p '{"spec":{"values":null}}' --type=merge
```
When creating a new deployment of Cloudera Data Flow version 2.9.0 and below, customers need to repeat the mitigation steps outlined above once the Cloudera Data Flow environment is successfully created.

About service upgrades

You trigger an upgrade for a selected Cloudera Data Flow environment. Upgrades move from an older version to the latest version available.
Once you trigger the upgrade, the service enters UPGRADING state and is upgraded to the latest supported Cloudera Data Flow and Kubernetes versions. Kubernetes upgrade can take up to an hour to complete. During the upgrade service actions are restricted, that is, you are not able to create or manage deployments in the given Cloudera Data Flow service.
As part of a Cloudera Data Flow Service (Environment) upgrade, all existing deployments are also upgraded.
To keep existing Cloudera Data Flow deployments working, you may choose not to upgrade the NiFi version of existing deployments during the upgrade. This selection applies to all existing deployments.
note
The option to skip NiFi version upgrade is not always available: NiFi version upgrade is mandatory if the upgrade wizard determines that the new workload version is not compatible with the current NiFi version running in any deployment. The same automatism prevents you from accidentally downgrading NiFi below the minimum supported version later.
You can manually upgrade NiFi for individual deployments later.
If upgrading a deployment fails, all components are rolled back to their original state and it causes the overall Cloudera Data Flow service upgrade to be rolled back as well.