<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Posts on 5glinux</title><link>https://blog.5glinux.com/posts/</link><description>Recent content in Posts on 5glinux</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Mon, 19 Oct 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://blog.5glinux.com/posts/index.xml" rel="self" type="application/rss+xml"/><item><title>Reachy Mini on a local AI stack: vision, face recognition and a menu-bar remote</title><link>https://blog.5glinux.com/posts/reachy-mini-local-ai-faces-swiftbar/</link><pubDate>Mon, 19 Oct 2026 00:00:00 +0000</pubDate><guid>https://blog.5glinux.com/posts/reachy-mini-local-ai-faces-swiftbar/</guid><description>I pointed a Reachy Mini robot at my own Kubernetes-hosted speech, language and vision models. Here&amp;rsquo;s how it sees with a VLM, why face recognition uses embeddings instead, and the SwiftBar menu that controls it all from my Mac.</description></item><item><title>Running DeepSeek V4 Flash on two DGX Sparks with Kubernetes and Codex CLI</title><link>https://blog.5glinux.com/posts/deepseek-v4-dgx-spark-codex/</link><pubDate>Mon, 19 Oct 2026 00:00:00 +0000</pubDate><guid>https://blog.5glinux.com/posts/deepseek-v4-dgx-spark-codex/</guid><description>DeepSeek V4 Flash needs more memory than one DGX Spark has. This guide splits it across two Sparks with vLLM tensor parallelism over RoCE, reaches 51 tokens per second, and connects OpenAI&amp;rsquo;s Codex CLI as a fully local coding agent.</description></item><item><title>Windows VMs on SUSE Virtualization: Hyper-V enlightenments make the difference</title><link>https://blog.5glinux.com/posts/windows-vms-hyperv-enlightenments/</link><pubDate>Sat, 17 Oct 2026 00:00:00 +0000</pubDate><guid>https://blog.5glinux.com/posts/windows-vms-hyperv-enlightenments/</guid><description>Windows runs best when it knows it&amp;rsquo;s virtualized. Since SUSE Virtualization 1.9, the Windows templates turn on Hyper-V enlightenments and virtio-blk, which removed periodic I/O stalls, cut a write-heavy phase by 24%, and kept older Windows from losing two-thirds of its random IOPS.</description></item><item><title>Managed PostgreSQL on SUSE Virtualization with WSO2's DBaaS operator</title><link>https://blog.5glinux.com/posts/wso2-dbaas-suse-virtualization/</link><pubDate>Wed, 14 Oct 2026 00:00:00 +0000</pubDate><guid>https://blog.5glinux.com/posts/wso2-dbaas-suse-virtualization/</guid><description>WSO2&amp;rsquo;s open-source DBaaS operator turns a Kubernetes object into a PostgreSQL VM on SUSE Virtualization, with TLS, credentials and monitoring wired up, in about 100 seconds. Here&amp;rsquo;s a quick install, including how to build its image without Packer.</description></item><item><title>NetApp Trident on SUSE Virtualization: just the YAMLs</title><link>https://blog.5glinux.com/posts/netapp-trident-harvester-yamls/</link><pubDate>Tue, 13 Oct 2026 00:00:00 +0000</pubDate><guid>https://blog.5glinux.com/posts/netapp-trident-harvester-yamls/</guid><description>A copy-paste version of my Harvester knowledge base article on NetApp Trident: every manifest and command for NFS, iSCSI and NVMe/TCP backends, with as little prose as possible.</description></item><item><title>Longhorn v2 on SPDK: TCP catches up, RDMA still wins on latency (part 1)</title><link>https://blog.5glinux.com/posts/longhorn-v2-spdk-rdma-vs-tcp/</link><pubDate>Mon, 12 Oct 2026 00:00:00 +0000</pubDate><guid>https://blog.5glinux.com/posts/longhorn-v2-spdk-rdma-vs-tcp/</guid><description>Two snapshots of Longhorn&amp;rsquo;s SPDK-based v2 engine on the same three Ampere nodes: 1.12 with my RDMA patches, then 1.14 development builds with a newer SPDK. NVMe/TCP got about 7× lower write latency and now fills a 25 GbE link. RDMA still halves write latency.</description></item><item><title>LVM thin storage on SUSE Virtualization: chunk sizes, tuning knobs and encryption</title><link>https://blog.5glinux.com/posts/lvm-thin-suse-virtualization/</link><pubDate>Fri, 09 Oct 2026 00:00:00 +0000</pubDate><guid>https://blog.5glinux.com/posts/lvm-thin-suse-virtualization/</guid><description>What a summer of work on SUSE Virtualization&amp;rsquo;s LVM CSI driver taught me: why a thin pool&amp;rsquo;s chunk size can cost you 98% of your IOPS, what changed upstream, the new VM storage settings in 1.9.1, and what encryption at rest costs.</description></item><item><title>VDI on SUSE Virtualization: Kasm Workspaces with autoscaling Harvester VMs</title><link>https://blog.5glinux.com/posts/kasm-vdi-suse-virtualization/</link><pubDate>Wed, 07 Oct 2026 00:00:00 +0000</pubDate><guid>https://blog.5glinux.com/posts/kasm-vdi-suse-virtualization/</guid><description>Kasm Workspaces 1.19 has a built-in Harvester provider: it creates agent VMs on demand, streams desktops to the browser, and deletes the VMs when they sit idle. Here&amp;rsquo;s a working setup on SUSE Virtualization v1.9, including Windows desktops, and five upstream bugs to work around.</description></item></channel></rss>