Tag: sysadmin

du and df still work, but Rust-era tools like dust, duf, and fclones make disk triage faster and way less painful in 2026.

Backups and disaster recovery are not the same thing. Learn RTO and RPO, the 3-2-1 rule, Proxmox and Restic backups, and a runbook that survives a real outage.

zstd is as fast as gzip with near-xz compression ratios. Here's why you should drop bzip2 forever and how to use zstd in 2026.

Ollama holds models in VRAM after every request. Set keep_alive, force an unload through the API, and cap how many models stay resident at once.

Your container crashes and restarts. Your app is broken but says it's healthy. These are two different problems. Here's the distinction.

Linux suspend vs hibernate explained: sleep states, swap setup, initramfs resume hook, wake-on-LAN, lid close behavior, and fixing common hibernate failures on modern Linux systems.

The upload works locally and dies through NGINX. The four timeouts that matter, the config that breaks them, the buffering trap, and how to test first.

Set up Chrony for NTP time sync in your home lab. Covers chrony.conf, chronyc tracking, stratum levels, LAN NTP server setup, and why correct time matters more than you think.

Your backend logs every client as 127.0.0.1 because the proxy dropped the header. Set X-Forwarded-For, X-Real-IP, and X-Forwarded-Proto right in Nginx.

Kernel live patching applies CVE fixes to a running kernel with no reboot, using kpatch or Canonical Livepatch. Setup for Ubuntu and RHEL, plus what it can't fix.

You've got backups. Great. But do you know if they actually work? RTO and RPO mean nothing if you've never actually restored.

One app logging to one file becomes a 50GB disk emergency. logrotate config, size-based rotation, Docker log driver limits, and app-level log control.