Release Notes

This document provides a comprehensive overview of the latest updates and enhancements in HPE Ezmeral Unified Analytics Software (version 1.4.0), including new features, improvements, bug fixes, and known issues.

HPE Ezmeral Unified Analytics Software provides software foundations for enterprises to develop and deploy end-to-end data and advanced analytics solutions from data engineering to data science and machine learning across hybrid cloud infrastructures – delivered as a software-as-a-service model.

New Features

Integration with HPE GreenLake for File Storage
HPE Ezmeral Unified Analytics Software now integrates with HPE GreenLake for File Storage as an additional data source. This enhancement provides broader data accessibility and allows for more diverse and robust data analysis capabilities directly within the platform.

Additional GPU Model Support
HPE Ezmeral Unified Analytics Software extends GPU model support to A10G, A30, L40S, H100, V100 and A100 to leverage the enhanced capabilities of these models and ensure seamless performance and integration.

Storage Disk Selection
To accommodate unique storage resource allocation configurations and preferences, you can now specify which disksHPE Ezmeral Unified Analytics Software uses for storage on a per-node basis. If you do not want all the disks on a node or multiple nodes allocated for storage, configure the storage disks, as described in Designating Internal Data Fabric Storage Disks on Worker Nodes.

In-Place Upgrade
With this release, HPE Ezmeral Unified Analytics Software supports an in-place upgrade feature that allows you to update to the latest version of the platform without a full re-installation, minimizing downtime and ensuring that you can quickly benefit from new enhancements. You can upgrade from version 1.3 to 1.4. For details, see Upgrading HPE Ezmeral Unified Analytics Software.

Automation and Scaling
This release adds detailed CLI deployment guides for HPE Ezmeral Unified Analytics Software, enabling efficient automation. Enhanced availability for Apps APIs and new documentation for Bring Your Own Apps Backend APIs offer increased security and customization, streamlining user experience and extending functionality. For details, see https://github.com/HPEEzmeral/byoa-tutorials.

Enhancements

Optional Email Attribute Field
In an effort to support scenarios where AD/LDAP does not have an email attribute field, this field is now an optional field during installation. If the admin performing the installation selects the Allow Login By Email Address option, users can sign in using their email address or username; otherwise, users can only sign in with their usernames. For upgrades, the user sign in experience remains the same as it was prior to upgrading.

Username Flexibility
Username requirements (including the admin username) are now less restrictive, and syntax is like POSIX except that a username cannot begin or end with a dot or underscore. Usernames can have capital letters, alphanumeric beginnings and endings, dots, dashes, and underscores, which are all valid for use within a 63-character limit. Also, usernames are case-insensitive. If a username is bob, this user can sign in as BOB, Bob, or bob. Regardless of how bob signs in, the username always displays in lowercase (bob).
Retry and Reset Installation
A new Reset and Retry feature provides the ability to identify and correct any installation issues and then restart the installation anew, at the click of a button. You can also directly activate your term licenses without using activation codes, making start-up quicker.

Mandatory Installation Pre-Check
The pre-check script runs prior to starting the installation process to verify that nodes meet the installation prerequisites. If the pre-check script identifies any issues, the system provides detailed messages in the UI, as well as log files for troubleshooting. Installation cannot resume until the pre-check script runs successfully.

User Authentication Details
The User Authentication Details installation screen no longer displays the Internal LDAP Server fields unless you select the option. Instead, External LDAP Server options show by default.

Bulk Permissions
To streamline the permissions process and enhance collaborative security measures, administrators can efficiently assign and control access to data for groups of users by applying permissions in bulk. For details, see Granting Group Access to Data.

Build and Version Visibility
The Installer UI provides the build and version number in the upper right corner of the screen.

Other Application Upgrades
For a list of updated applications, including MLflow and Ray, see Support Matrix.

Resolved Issues

This release includes numerous fixes that enhance system stability and performance, including the following resolutions:
  • Long-running Spark applications exceed disk quotas for Spark History Server
  • Upgrade spark images for Livy and spark operator in HPE Ezmeral Unified Analytics Software UI
  • Notebook SQL magics
  • Schedule Presto pods on GPU
  • Audit logs for Data Access
  • Prometheus disk capacity
  • Alerts for GPU
  • Installer UI displaying the ingress gateway node IP addresses on OpenShift
  • Support token based kubeconfig for OpenShift
  • Installation of HPE Ezmeral Unified Analytics Software on Azure while avoiding concurrency challenges

Known Issues

The following sections describe known issues with workarounds where applicable:

The driver pod of the cloned Spark job remains in the container creating state.

When you use the Clone option to create a new Spark application with a similar configuration as an existing Spark application, the driver pod of the cloned Spark job remains in the container creating state.

EzPresto does not release memory when a query completes

EzPresto retains allocated memory after query completion for subsequent queries because of an open-source issue (https://github.com/prestodb/presto/issues/15637). For example, if a query uses 10GB of memory, EzPresto does not release the memory when the query completes and then uses it for the next query. If the next query requires additional memory, for instance, 12GB, EzPresto accumulates an extra 2GB and does not release it after query completion. For assistance, contact HPE support.

Configuration changes to long-running pods are not applied in Ray

Configuration changes or upgrades to long-running pods in Ray, such as adjusting resource capacities or expanding persistent volume (PV) storage are not applied in Ray.

Workaround

To ensure successful configuration changes or upgrades, manually delete relevant pods after the reconfiguration or upgrade process. For details, see https://github.com/ray-project/kuberay/issues/527.

Worker nodes do not automatically spawn with JobSubmissionClient in the Ray cluster

When submitting jobs to the Ray cluster using JobSubmissionClient, worker nodes do not spawn automatically.

Workaround

To ensure proper functionality when submitting Ray jobs using JobSubmissionClient, you must manually specify entry point resources as follows:
  • For CPU, set entrypoint_num_cpus to 1
  • For GPU, set entrypoint_num_gpus to 1
For details, see Using JobSubmissionClient to Submit Ray Jobs.

HPE is actively engaging with the community to address this open-source issue (https://github.com/ray-project/ray/issues/42436).

NVIDIA GPU cannot enforce SELinux
Due to a known NVIDIA GPU issue (https://github.com/NVIDIA/gpu-operator/issues/553), SELinux cannot be enforced for GPU deployments.
Workaround
Set GPU hosts to either disabled or permissive mode until this issue is resolved.

Ray dashboard UI
A known Ray issue prevents the Ray Dashboard UI from displaying the GPU worker group details correctly. To see updates regarding resolution and to learn more, see https://github.com/ray-project/ray/issues/14664.

Upgrade on OpenShift cluster
If you want to perform an in-place upgrade of HPE Ezmeral Unified Analytics Software on an Openshift cluster, contact HPE support for assistance to ensure a smooth transition and to address any potential complexities that can arise during the upgrade process.

Submitting an installation retry request using an existing Ezmeral Coordinator fails
If you install Unified Analytics using the existing Ezmeral Coordinator and the installation fails, clicking the Reset and Retry Installation option returns the following error:
Failed to start Ezmeral Unified Analytics prep work. Kubeconfig file not found.
This issue occurs because the reset deletes the kubeconfig file for the existing Ezmeral Coordinator.
Workaround
To resolve this issue, start a new installation using the existing Ezmeral Coordinator, resolve the issue(s) that caused the initial installation failure, and manually upload the kubeconfig file. For additional details, see Submitting an Installation Retry Request using an Existing Ezmeral Coordinator Fails.

Cannot add Iceberg as a data source when Catalog Type is Hadoop
Recent changes introduced by open source PrestoDB cause Iceberg data connections to fail in Unified Analytics when the Catalog Type is Hadoop.
Workaround
To workaround this issue, update the EzPresto images, sign in to Unified Analytics, and then connect the Iceberg data source with Hadoop set as the Catalog Type. For details, see the steps listed here.

Installing Unified Analytics in an AWS cluster on OpenShift fails during MLDE installation
This installation failure occurs because there is no DNS mapping.
Workaround
Change the istio-ingressgateway service object in the istio-system namespace from NodePort to LoadBalancer. Get the external IP of the LoadBalancer and then register the external IP as a CNAME certificate in the domain or an A certificate with the alias.
To successfully create the DNS mapping, you must complete the following steps after the istio installation completes and within 45 minutes of the start of the MLDE installation.
  1. To change the istio-ingressgateway service object in the istio-system namespace from NodePort to LoadBalancer, run:
    kubectl patch svc istio-ingressgateway  -p  '{"spec": {"type":"LoadBalancer"}}' -n istio-system
  2. To get the external IP of the LoadBalancer, run:
    kubectl get svc -n istio-system
  3. In your cloud domain service provider, use the external IP to create a DNS mapping using either of the following methods:
    • Register the external IP as a CNAME under the domain name.
    • Create an A certificate with the alias, as describe here for AWS and here for GCP.

    Note that the external IP is not an IPv4 address. An IPv4 address is typically used in A records.

A crashloopbackoff error occurs due to the activation code change
You no longer have to provide an activation code when you apply a capacity license to your cluster. If you applied a capacity license to your HPE Ezmeral Unified Analytics Software version 1.3 cluster, upgrading to version 1.4 can result in a crashloopbackoff error due to this activation code change.
Workaroud
To prevent the crashloopbackoff error, run the kubectl command(s) appropriate for your environment to remove the ezua-activationcode. Run these commands after upgrading to version 1.4.
  • In a connected environment, run:
    kubectl -n ezlicense-system delete ezactivationcode ezua-activationcode 

// Restart the license manager pod. 
kubectl -n ezlicense-system rollout restart deployment ezlicense-controller-manager
  • In an air-gapped environment, run:
    // Get the secret.
kubectl -n ezlicense-system get secret
 
// Delete the secret.
kubectl -n ezlicense-system delete secret <secret-name>

// Example: kubectl -n ezlicense-system delete secret ezua-activationcode-2023-09
 
// Restart the license manager pod.
kubectl -n ezlicense-system rollout restart deployment ezlicense-controller-manager

Installation

Before you install or upgrade, HPE recommends that you back up your data.  If you encounter any issues during or after the installation or upgrade process, please contact HPE support. We appreciate your feedback and strive to continually enhance your product experience. 

Additional Resources

Thank you for choosing HPE Ezmeral Unified Analytics Software. Enjoy the new features and improvements introduced in this release.