Icon browsing and distribution service based on selfh.st/icons, providing fast and stable CDN access for users in mainland China

Complete guide to deploying Pocket ID passwordless authentication service on Dokploy, implementing Passkey-based OIDC single sign-on, including version fixes, environment configuration, and application integration.

Comprehensive bandwidth performance testing from Tsinghua University campus network to 15 global nodes using iperf3, covering Asia, Europe, Americas, and domestic ISP nodes

Deploy Tailscale Derper relay server via Docker or manual compilation to optimize network latency with client verification and automatic certificate management

A comprehensive review of all the self-hosted services I deployed and used throughout the year, from infrastructure to applications

Complete guide to deploying Misaka Danmu Server with Docker Compose. Add danmaku (live comments) support to your private media library. Covers MySQL setup, web management, player integration, and webhook automation.

BeroHost Christmas promotion featuring AMD EPYC 9275F processor - 4 cores, 20GB RAM for just €6.9/month. Stackable packages with bonus options. Includes comprehensive Geekbench, UnixBench, PassMark, and network benchmarks.

Build a complete security and monitoring system in Dokploy, including CrowdSec threat detection, Traefik Cloudflare IP allowlist, and Grafana/Loki/VictoriaMetrics log monitoring

2025年年终总结

A modern self-hosted image hosting solution built on Cloudflare ecosystem with automatic format conversion and random image API

How I fixed Out of Memory (OOM) errors when building a 500+ post Astro blog on Vercel, reducing build time to 2 minutes 19 seconds

This guide walks through building a full NVIDIA GPU compute node on Ubuntu 20.04/22.04, targeting HGX A800 and Tesla data center GPUs. It covers disabling Nouveau, installing the NVIDIA driver and CUDA, configuring Infiniband with MLNX OFED, enabling key kernel modules, building OpenMPI, setting up NCCL and FabricManager, and validating with single-node and multi-node NCCL+MPI benchmarks.