May 19, 2026

NiFi 2.6.0 and NiFi 1.28.1 have been updated to the latest hotfix build.

You can now manage Flow Designer test sessions entirely from the CDP CLI, enabling automation and scripting of your development workflows.

Test session operations are automatically rejected with a clear error message if the deployment is in a transitional state (for example., upgrading, suspending), preventing accidental state conflicts.

For a list of available commands, see Managing test sessions using the CLI.

Full lifecycle management of shared parameter groups is now available through the CLI.

You can now reference shared parameter groups by CRN alone when creating deployments, deploying flows, or updating parameter group references - no need to specify full group details.

For a list of available commands, see Managing parameter groups using the CLI.

Data Provenance is now fully enabled for NiFi 1.x flow drafts. You can access "View Data Provenance" from the context menu on processors, funnels, input ports, and output ports to inspect data lineage and troubleshoot flow behavior.

The ConsumeMQTTIIoT processor and MQTTIIoTReader controller service are now recognized as built-in components. They no longer require a custom NAR upload to use in test sessions or deployments

Cloudera Data Flow now supports EKS/AKS 1.34.

Traefik has replaced NGINX as the cluster ingress controller for Cloudera Data Flow. This change improves routing capabilities and TLS handling. When you upgrade from version 3.0.0 or lower to version 3.1.0 or higher, the ingress controller is replaced without manual intervention.

Valkey and cert-manager base images have been replaced with hardened Chainguard equivalents, improving the security posture of the runtime environment.

Significant performance improvements for environments running multiple concurrent test sessions:

• Referencing components now load in approximately 0.1 seconds, down from over 1 minute in large flows

• Overall CPU usage for Flow Designer event processing reduced by 60–80% under load

• Optimized internal caching and indexing to reduce redundant computation

• Deployed flow update optimization

Updating a deployed flow no longer triggers unnecessary metadata computation when KPIs are not part of the update request, reducing processing overhead.

• PostgreSQL Driver Timeout

  The database driver timeout is now externally configurable via Helm chart values. The default remains 30 seconds, but can be increased for environments with large flows that require longer operations (such as flow draft deletion in high-volume environments).

Flow Designer event housekeeping

A new hourly scheduled task automatically purges historical flow event records, retaining only the first and last events per flow to prevent unbounded database growth.

When you terminate the last flow using an inbound connection endpoint in a deployment, an informational alert is raised at the deployment level to help you avoid unnecessary cloud costs:

"You have deleted a flow from a deployment associated with an inbound connection. If this inbound connection is no longer needed, unassign it from the deployment in the Manage Deployment view to save on cloud costs."

This alert resolves automatically when the inbound connection is unassigned.

• Kafka to Snowflake and Confluent Cloud to Snowflake ReadyFlows updated to use key-pair authentication (StandardPrivateKeyService) to comply with Snowflake's mandatory multi-factor authentication requirement.

• S3 to IBM watsonx ReadyFlow updated to no longer reference a default LLM model.

Resolved an issue where script-driven deployments would fail if a KPI payload was submitted without a specified component type.