What's new in 1.5.5 SP3

Cloudera AI Registry

Cloudera AI Registry in-place upgrade

Cloudera AI Registry can be upgraded in place on the existing release within the same namespace in Cloudera AI 1.5.5 SP3 and higher releases.

The in-place upgrade preserves all Kubernetes resources, allowing the process to complete in a short time, typically within 15 minutes. For more details, see Cloudera AI Registry upgrade.

Multiple Cloudera AI Registries working with a single Cloudera AI Inference service

Cloudera supports deploying Cloudera AI Inference service connecting to multiple Cloudera AI Registries in Cloudera AI. For more details, see Multiple Cloudera AI Registries connected to a single Cloudera AI Inference service.

API endpoint for MLFlow model download

This new API endpoint allows you to download a model artifact for a specific model version from the Cloudera AI Registry. It supports models that use the MLFLOW repository type; other repository types are not compatible. For more details, see Downloading MLFlow models using API endpoint.

Cloudera AI Inference service

Fine-grained access control

Fine-grained access control allows Administrators to define specific access levels for Model Endpoints for individual users or groups. For more details, see Configuring fine-grained access control for Model Endpoints.

Knox API key support

Cloudera AI Workbench

Quota management for General Availability

Quota Management is generally available in Cloudera AI on premises 1.5.5 SP3 and higher releases for fresh workbench installations. Quota Management enables you to control how resources are allocated within your Cloudera AI Workbench on user and on team level. For more details, see Quota Management overview.

New alert for insufficient quota

When creating a session, job, or application, an alert notifies the user if the configuration does not have sufficient quota.

User ID decoupled from user namespace

With the introduction of dynamic pool assignment, UserIDs no longer map predictably to user namespace sequence numbers, and the ds-username label is no longer included in new user namespaces.

Diagnostic bundles for a Cloudera AI Workbench

New types of logs, the User logs, can be generated and downloaded from the Diagnostic page in the Site Administration settings of the Cloudera AI. For more details, see Downloading diagnostic bundles for a workbench.

Automated application restart functionality

Failed applications are now automatically restarted up to three times (with a five-minute delay between attempts) to minimize downtime. This feature is enabled by default, but Site Administrators can disable it globally under Site Administration settings. For more information, see Disabling global Application restarts.

Refined Project list filters and scoping

The Projects list page now features a refined My Projects filter that displays only projects explicitly owned by the logged-in user, moving shared work to a brand-new My Team Projects filter for individual collaborators and team members. This behavioral change makes it easier to isolate personal projects without scrolling through shared repositories.

Self-service Run As service account assignment for contributors

Project Contributors can now independently assign service accounts (machine users) to workloads using the "Run as" feature, eliminating the need for Site Administrator intervention. To maintain platform security, Contributors are strictly limited to selecting service accounts that hold an Operator role within the project or team. Service accounts with Administrator privileges remain restricted to Site Administrators, Project Administrators, and Project Owners. This self-service capability applies to creating and updating jobs, applications, and models across both v1 and v2 APIs. For more information, see, Creating a Workload as a Contributor.

ML Runtimes

Hardened Chainguard ML Runtimes

This release introduces new Hardened Edition ML Runtimes based on Chainguard images. These runtimes are designed to meet strict security standards and provide enhanced protection for your workloads and are released behind a paywall.

Hardened Runtime workloads do not use the Java version provided by the Hadoop Runtime add‑on. Instead, they rely on the Java 17 installation included in the Runtime image.

Hadoop Runtime add-on images

Hadoop Runtime add-on images are upgraded to Ubuntu 24.04 LTS.