As deployments of Red Hat Enterprise Linux AI (RHEL AI) workloads grow, Red Hat has been working to create a new set of recommendations to help RHEL AI administrators streamline operations. Connecting a RHEL AI system to Insights remains simple, as with any other RHEL system. To register, run this command:
$ rhc connectOnce registered, there's no further action required. You can immediately see your findings for this host at console.redhat.com/insights/advisor and benefit from the new set of RHEL AI recommendations. Read more in Configuring accounts for RHEL AI or use the Red Hat Insights registration assistant to help you get started.
New Red Hat Insights recommendations for RHEL AI
RHEL AI is based on the same underlying image as traditional RHEL, but it's purposefully built and tested as an appliance. Modifications to the image itself aren't expected, nor recommended. This has led us to adjust existing recommendations in Red Hat Insights to accommodate the unique characteristics of RHEL AI, mostly due to bootc upgrades only preserving contents of the /etc and /var directories. We've modified the resolution steps and detection logic of existing recommendations to avoid incompatibilities with RHEL AI.
New recommendations designed especially for RHEL AI workloads cover a wide range. You can see them in the Red Hat Enterprise Linux AI topic within Red Hat Insights Advisor.
Supported hardware
RHEL AI requires high-performance capacity to support all phases of the LAB model alignment methodology, like model fine-tuning, synthetic data generation and AI model serving. It's important to make sure a RHEL AI system has the resources it needs and runs on supported hardware. Red Hat Insights Advisor is now capable of detecting unsupported hardware and alerting customers about this suboptimal setup.
To learn more about hardware requirements, read Red Hat Enterprise Linux AI hardware requirements.
Supported RHEL AI version
Because RHEL AI is rapidly developing, bug fixes and new feature development are aimed at the latest versions of the product. Keeping up with the latest supported version can be a challenge, so Red Hat Insights keeps an eye on the currently supported versions of RHEL AI and alerts you whenever it detects an unsupported version in your environment.
RHEL AI image updates
RHEL AI is built on image mode RHEL and bootable container technology (bootc), which relies on pulling images from Red Hat's container registry. To avoid facing bootc authentication problems, it's recommended to log in to redhat.registry.io and make a copy of /run/containers/0/auth.json to /etc/ostree . Red Hat Insights helps with precisely this task, alerting you when the file is missing or empty.
Power supply of GPUs
When a GPU is not receiving adequate power, it results in a downward throttling of its clock speed. As a result, Instructlab’s training and synthetic data generation performance decreases. Insights can detect the hardware power brake slowdown of certain GPU brands and alert about it.
Deploy AI with confidence
These new recommendations demonstrate our commitment to supporting the growth of RHEL AI workloads and ensuring that you have a smooth day-2 operations experience when managing RHEL AI with Red Hat Insights.
Try deploying RHEL AI through our product trial and connect it to Red Hat Insights.
关于作者
With more than 10 years of experience in the software industry, Stefan Bunciak is currently the Product Manager for Red Hat Insights. He completed his master's degree in Informatics at Masaryk University in Brno and is skilled in project and people management, quality engineering, and software development. In his spare time, he plays violin in a folklore band.
