Posts on Aleksei Riumin | Senior Infrastructure Architecthttps://lxryumin-best.github.io/posts/Recent content in Posts on Aleksei Riumin | Senior Infrastructure ArchitectHugoen-usWed, 17 Jun 2026 12:00:00 +0300Automating a Senior Architect's Portfolio: Hugo, GitHub Pages, and AIhttps://lxryumin-best.github.io/posts/17062026_hugo_portfolio_automation/Wed, 17 Jun 2026 12:00:00 +0300https://lxryumin-best.github.io/posts/17062026_hugo_portfolio_automation/How to build a zero-maintenance, blazingly fast portfolio using Hugo, GitHub Actions, VS Code, and Claude Opus 4.6 as an automated translator and technical writer.Mitigating 99% Disk Exhaustion Caused by AI & SEO Scrapershttps://lxryumin-best.github.io/posts/17062026_highloadbottrafficmitigation/Wed, 17 Jun 2026 12:00:00 +0300https://lxryumin-best.github.io/posts/17062026_highloadbottrafficmitigation/How we resolved a critical server disk space issue on a high-traffic sports portal by identifying and blocking aggressive AI bots and SEO crawlers.Seamless Bare-Metal Migration: NVMe Lift-and-Shift Between Identical Nodeshttps://lxryumin-best.github.io/posts/16062026_baremetal_nvme_migration/Tue, 16 Jun 2026 11:31:39 +0300https://lxryumin-best.github.io/posts/16062026_baremetal_nvme_migration/How to execute a fast bare-metal hardware swap using NVMe migration on identical chipsets, mitigating TPM, Secure Boot, and network configuration conflicts.Physics vs. Marketing: Real AV1300 PowerLine Speeds on Old Aluminum Wirehttps://lxryumin-best.github.io/posts/15062026_powerline_aluminum_mesh_bridge/Mon, 15 Jun 2026 12:00:00 +0300https://lxryumin-best.github.io/posts/15062026_powerline_aluminum_mesh_bridge/How to salvage a remote network link when underground conduits fail. Testing TP-Link AV1300 over 2-core aluminum and optimizing IP cameras for constrained links.Cost-Effective Kubernetes: Hardware Isolating Dev & Prod Environments in a Single Clusterhttps://lxryumin-best.github.io/posts/15062026_kubernetes_node_affinity_dev_prod/Mon, 15 Jun 2026 10:00:00 +0300https://lxryumin-best.github.io/posts/15062026_kubernetes_node_affinity_dev_prod/Learn how to use Kubernetes Node Affinity to safely run Development and Production environments within a single cluster, saving B2B infrastructure costs.Proxmox Server Crash: Recovering XFS Data After a Catastrophic NVMe Failurehttps://lxryumin-best.github.io/posts/14062026_proxmox_nvme_crash_recovery/Sun, 14 Jun 2026 22:41:25 +0300https://lxryumin-best.github.io/posts/14062026_proxmox_nvme_crash_recovery/How to handle a sudden NVMe controller failure in Proxmox VE. Step-by-step XFS troubleshooting, read-only data salvage, and hardware migration.Welcome to My Digital Garden: Infrastructure & DevOpshttps://lxryumin-best.github.io/posts/welcome/Sun, 14 Jun 2026 12:00:00 +0300https://lxryumin-best.github.io/posts/welcome/About my expertise as an Infrastructure Architect, System Administrator, and Independent IT Consultant — from network engineering to cloud-native solutions.<link>https://lxryumin-best.github.io/posts/21062026_workstation_upgrade_rtx3080_throttling/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://lxryumin-best.github.io/posts/21062026_workstation_upgrade_rtx3080_throttling/</guid><description><p>title: &ldquo;Architect&rsquo;s Workstation Upgrade &amp; RTX 3080 Thermal Throttling Analysis&rdquo; date: 2026-06-21T12:00:00+03:00 draft: false description: &ldquo;Upgrading a local infrastructure architect&rsquo;s lab. Diagnosing severe RTX 3080 thermal throttling and planning VRM/VRAM maintenance to restore 320W TDP.&rdquo; tags: [Hardware, Workstation, RTX3080, Troubleshooting] The Context (Business Challenge / Problem): Maintaining an optimal local environment is critical for an Infrastructure Architect working with Proxmox, Docker, and preparing for the CKA (Kubernetes) certification. I recently migrated my primary workstation to an Intel Core i5-12400 with 32GB of DDR4—currently the most cost-effective platform for running virtualized local labs. However, during stress testing, the Palit RTX 3080 GPU exhibited severe thermal throttling. Instead of operating at its design power envelope of 320W-330W under heavy load, the card drastically downclocked, dropping power consumption to just 140W to prevent critical silicon damage. The Architecture &amp; Work (Solution): The core platform rebuild was a success, providing a stable and highly efficient foundation for complex routing and containerized workloads. To address the GPU bottleneck, I analyzed the hardware telemetry. The issue stems from degraded thermal pads on the GDDR6X memory modules and VRM. GDDR6X runs notoriously hot, and once it hits the 105°C+ junction temperature limit, the firmware aggressively throttles performance. The engineering solution requires a complete teardown of the cooling system, precise PCB cleaning, and the application of high-performance thermal pads with exact thickness tolerances to bridge the gap between the components and the heatsink.</p></description></item></channel></rss>