MSK · UTC+3 · 18 Nov 2026one engagement open · Dec–Jan
Ilya Rutj
infrastructure engineer
Linux · networks · uptime
12 years, mostly small fleets
I keep small server fleets alive.
Contract work on Linux hosts, network edges and the boring parts of uptime: backups that restore, monitors people trust, and the incident that always happens at 3 a.m.
Services
01Host and network tuning.MTU and MSS, BBR, conntrack, nftables rules that survive a reboot. Measured before and after, not guessed.
02Backups that restore.restic to off-site storage, retention that fits the disk, and a restore drill on a schedule — because an untested backup is a rumour.
03Monitoring without noise.Checks that fire once and mean something: reachability, certificates, disk trend, and a short page that tells you what broke.
04Incident response.Packet captures, conntrack dumps, kernel logs read in order. Then a written postmortem with the one change that prevents the repeat.
Selected work
2026Renumbered a 14-node edgewithout downtime: moved addresses, certificates and DNS in one planned window; the only visible change was the routing table.
2025Cut backup restore time from 6 hours to 40 minutesfor a 1.2 TB dataset by switching to restic with a local cache and testing the drill monthly.
2025Found a 9-month-old packet loss bugon a hosting network: an MTU mismatch on one transit link, invisible to ping, visible in tcpdump after two days of captures.
2024Wrote the runbooka four-person team still uses for certificate renewals across 20 hosts, after a renewal missed silently.
Notes
Short write-ups of things that broke and what actually explained them.