linux-sysadmin

name	linux-sysadmin
description	Linux sysadmin for Ubuntu 24.04 production — Angie (Nginx fork), PM2, PostgreSQL 18, Redis 8, Docker 29, PHP 8.5, UFW. Use when: server health, nginx, angie, reverse proxy, SSL, certbot, ACME, firewall, ufw, fail2ban, backup, disk space, OOM, load average, pm2, ecosystem.config.js, docker, journalctl, logrotate, systemd, systemctl, cron, sysctl, SIGTERM, fd exhaustion, conntrack. Also: site down, 502, 504, disk full, service crash, OOM-killed, SSL expired. SKIP: app code (→ app skill), Kubernetes, cloud-managed (ECS/GKE), non-Ubuntu.
stacks	["sysadmin"]
tags	["sysadmin","linux","devops"]
category	devops
color	green
displayName	Linux SysAdmin
risk	high-stakes

🎯 Version Requirements (May 2026)

Primary pins:

• Ubuntu: 24.04 LTS
• Angie: 1.11.x (Nginx fork, production default)
• nginx: 1.28.x (mainline) / 1.26.x (stable)
• Docker Engine: 29.x
• PostgreSQL: 18.x
• Redis: 8.x
• PHP: 8.5.x (Active)
• Node.js: 24.x (Active LTS)

Source of truth: STACK_VERSIONS.md — verified 2026-05-16

Usage

Loaded automatically when its description matches the active task. Read only the section you need.

Use this skill when

• Managing Ubuntu 24.04 production servers (this stack)
• Configuring or troubleshooting Angie (Nginx fork) as a reverse proxy
• Managing SSL/TLS via Certbot OR Angie's built-in ACME
• Configuring UFW firewall, fail2ban, SSH hardening
• Managing services via PM2 (Node.js) or systemd (everything else)
• Diagnosing: 502/504, disk full, OOM, slow load, connection refused, service crash, cron silent fail, fd exhaustion
• Tuning PostgreSQL/Redis at the OS host level (memory split, persistence, sysctl)
• Setting up backups, logrotate, cron/timer jobs

Do not use this skill when

• Target OS is not Ubuntu/Debian Linux (other distros, macOS, Windows)
• Task is purely application code — use the matching app skill (fastify, nodejs, nextjs, etc.)
• Infrastructure is Kubernetes or cloud-managed (ECS, GKE, App Runner)
• Task is DB query tuning (use postgresql), Redis command semantics (use redis), or queue logic (use bullmq)

Purpose

This skill owns the OS layer of a Vechkasov production server: Ubuntu 24.04 with Angie 1.11.3 reverse-proxying PM2-managed Node 24 apps and PHP 8.5-FPM, backed by PostgreSQL 18 and Redis 8, fronted by UFW and fail2ban, with Certbot or Angie built-in ACME for TLS. It covers safe edits to system config, service lifecycle, observability (journalctl + PM2 logs), backup baselines, and incident response when the box (not the app) is misbehaving.

This skill does NOT cover: application logic (use the app skill), Kubernetes (different paradigm), cloud-managed services. For Redis or Postgres command-level questions, defer to / . This skill handles the OS host they run on.