Manage Kubernetes Node Fleets with NodeWright

Kubernetes Multi-Tenancy, Layer by Layer: Isolation Proved by Being Refused

“” is published by Filipe Motta in ITNEXT.

PyTorch 2.15 tackles the 'one bad GPU' problem

PyTorch 2.15 ships a rewritten NCCL backend that stops a single failing GPU from forcing a full training restart. The new ProcessGroupNCCL allows only one communicator per process group during initialization, which by default eliminates the crosstalk that the old version's many communicators caused. A third-party orchestrator can now exclude a crashed or slow GPU and assign a replacement worker. A new 'flight recorder' captures operation snapshots for debugging stalls. The legacy backend remains as a rollback option but will be removed in 2.16, planned for early next year. The release also totals 3,117 commits from 583 contributors and includes parallel processing improvements for Apple gear and support for Nvidia's next-generation Blackwell GPUs.

What Is OpenTelemetry? A Guide to the OTel Framework

What is OpenTelemetry? See how the OTel framework standardizes traces, metrics, and logs so you can switch backends freely.

100% GPU Utilization, 39% of Rated FP16

nvidia-smi GPU-Util says a kernel ran, not how much work the GPU did. How DCGM metrics on Kubernetes show real GPU throughput, power caps and node health.

Faster LLM inference startup: What you need to know

Adding an inference replica creates capacity only after the new instance can serve requests. Between starting a process and generating a response…

Nvidia Kubernetes Engine (NVKE)

NVIDIA Kubernetes Engine (NVKE). A management plane for GPU Kubernetes on bare metal. - dsx-ai-factory/nvke

OWASP Top 10 Kubernetes Security Risks Explained

OWASP Top 10 Kubernetes Security Risks phân tích 44 cụm Kubernetes bị xâm phạm do rò rỉ secret. Nguyên nhân kỹ thuật chính là việc sử dụng credentials hardcode và RBAC cấu hình lỏng lẻo. Hệ quả bao gồm access không ủy quyền, data theft, và lateral movement trong cụm. Các attack vector phổ biến là Pod escape và namespace compromise. Bài viết cung cấp chiến lhardening với Pod Security Policies (PSPs) và admission controllers.

Lập trình viên phát triển ứng dụng trên Kubernetes nên đọc bài này để hiểu rõ những rủi ro an ninh cơ bản nhất, từ việc tiết lộ mật khẩu đến khai thác lỗ hổng trong cấu trúc cluster, giúp họ xây dựng các giải pháp bảo mật hiệu quả từ giai đoạn thiết kế.

5 GPUs that are finally too old for modern games in 2026

Every good thing must come to an end