Red Hat OpenShift AI Self-Managed Life Cycle

Overview

Red Hat provides a product life cycle for Red Hat OpenShift AI (RHOAI) Self-Managed allowing for customers and partners to effectively plan, deploy, and support their applications running on the platform. Red Hat publishes this life cycle to provide transparency but might make exceptions from these policies if conflicts might arise.

Notice of change:

Starting from the 2.8 release, the RHOAI life cycle will transition to a new release type approach. Starting from 2.11 the release cadence will change from a three week time period to monthly. The following definition illustrates its full implementation after the transition period.

RHOAI release types, and their respective life cycles, generally fall under three main categories:

  • Fast releases: These releases include Full Support for one month, or until the next fast release is available. In some cases, the next release might be later than a month.
  • Stable releases: These releases include Full Support for seven months. Red Hat issues a stable release every three minor releases.
  • Extended Update Support (EUS) releases: These releases include Full Support for seven months followed by Extended Update Support for eleven months. Red Hat issues a EUS release every nine minor releases.

In order to understand how to effectively deploy and maintain a specific release type, refer to the Upgrade Strategy section.

During the Full Support phase, qualified Critical and Important Security Advisories (RHSAs) will be released as they become available. Urgent and Selected High Priority Bug Fix Advisories (RHBAs) will be released as they become available; all other available fix and qualified patches may be released via periodic updates. In order to receive security and bug fixes, customers are expected to upgrade their OpenShift AI environment to the most current supported micro (x.y.z) version.

Red Hat OpenShift AI Self-Managed is available as an Operator to Red Hat OpenShift and maintains a release schedule that is independent from other Red Hat products and services.

The Red Hat OpenShift Life Cycle provides information on supported versions for Red Hat OpenShift. Red Hat OpenShift AI Self-Managed currently supports the following OpenShift versions:

  • 4.16
  • 4.15
  • 4.14
  • 4.13
  • 4.12
For information on RHOAI supported configurations, see the Supported Configurations article.

Life-cycle Dates

Red Hat OpenShift AI Self-Managed Life Cycle Dates

Upgrade Policy

The RHOAI operator and installed components, are automatically updated to the latest version, unless the manual upgrade strategy is opted for. For more information about how to install the operator and configure the update strategy, see the RHOAI Documentation. Customers are advised to deploy the latest available minor version at their earliest convenience.

This upgrade policy includes feature releases, as well as bug and security fix releases.

Upgrade Strategy and Paths

Customers are advised to choose their upgrade strategy according to their needs, which might vary in terms of release longevity or number of features available. When defining this strategy, It is important to consider that choosing the automatic approach ensures that customers will receive all the latest security and bug fixes for the currently supported version. Red Hat OpenShift AI uses major (x.), minor (x.y), and micro (x.y.z) release versions and maintains a release schedule that is independent from other Red Hat products and services. Red Hat tests and supports upgrade paths that are allowed according to the OLM rules enforced by the operator. The customer is free to change the streaming channels accordingly.
Note: Red Hat recommends that you plan your upgrades so that you are on a supported channel at all times. You must be on the latest available version in your selected channel to receive support for Red Hat OpenShift AI.

fast

Customers that want access to the latest product features are recommended to choose the fast streaming channel. Be advised that Red Hat supports only one fast release at a given time. These types of deployments are recommended for where having the lasted features is desirable. Red Hat recommends choosing this streaming channel with selecting the automatic update strategy in order to avoid the overhead of upgrading manually monthly. For the fast update channel, Red Hat supports direct, single-step upgrades from the previous minor version only (y-1). Red Hat does not test updates to noncontiguous versions and cannot guarantee compatibility with earlier versions. For example, upgrading directly from OpenShift AI 2.8 (fast) to the latest version of 2.10 (fast) is not supported. Instead, users could upgrade as follows:

  • 2.8 (fast) -> 2.9 (fast)
  • 2.9 (fast) > 2.10 (fast)

stable and stable-x.y

Customers who prioritize stability over new feature availability are recommended to choose the stable or stable-x.y streaming channels. Selecting the automatic updates strategy with the stable, unnumbered channel, will result in the deployments being upgraded to the latest stable minor version as soon as it is released. This choice will reduce the overhead of updating manually as soon as a new stable release is available and will grant access to the latest stable features. Alternatively, the selection of the numbered stable channels will allow customers to plan and execute the upgrade to the next stable release while keeping their deployment under full support within a four months time window. Be advised that Red Hat supports from two to three stable releases at a given time. These types of deployments are recommended for most stage and production environments. In the stable and stable-x.y update channels, Red Hat supports single-step upgrades from the most recent previous minor stable version to the latest minor stable version, or from the most recent previous minor stable version to the latest fast version. For example, users could upgrade from OpenShift AI 2.8.0 (stable) as follows:

  • 2.8.0 (stable) -> 2.8.3 (stable)
  • 2.8.3 (stable) -> 2.10.0 (stable)
or:
  • 2.8.0 (stable) -> 2.8.3 (stable)
  • 2.8.3 (stable) -> 2.9 (fast)

eus-x.y

Customers who have a strong stability preference over time are recommended to choose the eus-x.y streaming channels. On these channels, customers will have up to nine months to plan and execute the upgrade to the next EUS release while keeping its deployment under extended update support. These types of deployments are recommended for enterprise-grade environments that cannot update in a 7 month window. In the eus-x.y update channel, Red Hat supports single-step upgrades from the most recent previous minor EUS version to the latest minor EUS version, or from the most recent previous minor EUS version to the latest fast version.

Red Hat tests and supports upgrade paths that are allowed according to the OLM rules enforced by the operator. The customer is free to change the streaming channels accordingly.

Components Life Cycle

Data science pipelines

Since the RHOAI 2.9 release, the data science pipelines component has been upgraded to version 2. This upgrade follows the trajectory of its upstream equivalent, KubeFlow Pipelines. Data science pipelines v2 will support pipeline versions and pipeline logs, along with support for running data science pipelines in a disconnected RHOAI environment. With the introduction of data science pipelines v2, it is no longer possible to deploy, view, and edit the details of a data science pipelines v1 pipeline from the RHOAI dashboard. Further information on using data science pipelines v2 are available in the RHOAI 2.9 documentation.