145 Days of Uptime, 56 TiB of Traffic, One Debian Upgrade
2 days ago
- After 145 days of uptime, a home server was upgraded from Debian 12 to Debian 13, handling 51 TiB inbound and 5 TiB outbound traffic across diverse tasks.
- The server runs services like PHP apps, MariaDB, Samba, CoreDNS, MQTT, openHAB, TVHeadend, Frigate, Prometheus, Grafana, and more directly on bare-metal Linux.
- The author argues against overcomplicating homelabs with hypervisors or containers, favoring native systemd services where appropriate.
- The upgrade involved simplifying package sources, fixing ZFS DKMS issues, updating NodeSource keys, rebuilding pipx for Python 3.13, and moving openHAB to OpenJDK 21.
- Post-upgrade, system load dropped from ~3.6 to ~1.2, RAM usage from ~14 GB to ~6.6 GB, and many old packages were removed.
- Infrastructure as code (Ansible) and backups made the upgrade reproducible and low-risk, emphasizing engineering quality over uptime alone.
- AI tools accelerated troubleshooting, but operational judgment remained crucial for evaluating solutions.