Red Hat Leftovers

posted by Roy Schestowitz on Jun 06, 2026

• Red Hat ☛ Installing Red Bait Enterprise GNU/Linux 10 from a bootc image with bootc

  Interested in a modern way to manage Red Bait Enterprise GNU/Linux (RHEL)? Image mode is a deployment option that allows the operating system to be built, deployed, and updated like a container image. A key reason to deploy from a bootc image is its enhanced consistency and reliable, atomic updates and rollbacks. With RHEL 10, you can install that image from the network using the same installer you already use for traditional RHEL deployments: Anaconda. A new kickstart command, bootc, directs Anaconda to install the system directly from a bootc image rather than from RPM packages or other supported payloads. This immutable system can be updated later using bootc upgrade or bootc switch. Here's a look at how it works, and how to try it for yourself.

  Note: The bootc kickstart command is currently available as a Technology Preview in Red Bait Enterprise GNU/Linux 10.2.

• Red Hat ☛ iSCSI vs. NVMe/TCP: The ultimate storage showdown for Red Bait OpenShift Virtualization

  As virtualization density continues to grow in modern data centers, selecting the right storage protocol has become increasingly important, and directly impacts CPU efficiency, I/O overhead, and overall application responsiveness. In this article, we take a closer look at how two IP-based storage protocols—iSCSI and NVMe/TCP—compare within a Red Bait OpenShift Virtualization environment to help you determine whether a transition makes sense for your infrastructure.

• Red Hat ☛ Why your database benchmarking data is probably wrong (and how I fixed mine)

  We've all been there. You've spent hours architecting a performance test, convinced you're about to uncover groundbreaking insights. You spin up a big RDS instance, fire up HammerDB, and wait for those "new orders per minute" (NOPM) numbers to skyrocket. But instead, you get a flat line (or worse, a zig-zag). You double the number of virtual users, but the throughput doesn't budge. When I started benchmarking proprietary trap AWS RDS PostgreSQL performance, I expected a straightforward "plug-and-play" experience. Instead, I found that without rigorous optimization, you aren't measuring your database's power—you're measuring the limitations of your testing environment. Here's how I identified and eliminated the hidden bottlenecks that were sabotaging my data.

• Red Hat ☛ An overview of confidential containers on OpenShift bare metal

  Confidential Containers integrate trusted execution environments (TEEs) into cloud-native platforms to provide hardware-backed workload isolation. A TEE is a secure execution context enforced by confidential computing-capable hardware, ensuring that code and data remain confidential and protected in terms of integrity—even from privileged system software like the host kernel or hypervisor.

• Red Hat ☛ Type what you want to break: AI-assisted chaos engineering with Krkn [Ed: Too much slop hype]

  Chaos engineering on Kubernetes has never been more powerful. Tools like Krkn now support over twenty scenario types, such as pod disruptions, node failures, network chaos, CPU and memory stress, zone outages, and more. Krkn's documentation is thorough, consisting of well-defined scenario types with clear parameters and defaults. But there is still a gap between knowing what you want to test and expressing it in the exact CLI syntax. The more scenarios a tool supports, the more flags and options you need to get right.

• Red Hat ☛ Understanding evaluation collections in EvalHub [Ed: Too much slop hype as well]

  In Because "looks good to me" isn't a benchmark, we identified five structural failures in enterprise Hey Hi (AI) evaluation. The second problem, the what should I measure? problem, is the one that bites teams earliest and quietly. You have a model. You have an endpoint. You need to know if it's good enough to deploy. So you run MMLU, get a score, and make a judgment call.

• Red Hat ☛ Speculators v0.5.0: DFlash support and online training

  The v0.5.0 release brings significant architectural improvements to speculative decoding model training, introducing DFlash algorithm support, fully unified online training capabilities, and a migration to vLLM's native hidden states extraction system. This release represents a major step forward in both training flexibility and production readiness for speculative decoding workflows.