Stark kernel patch 110 → 111 — clean, find feature survived

Stark patched and rebooted

Host: stark / 172.232.185.138
Kernel: 6.8.0-110-generic → 6.8.0-111-generic
Reboot time: 2026-05-02 ~12:22 PDT (19:22 UTC), back up by 12:48 UTC at 26-minute uptime mark

Pre-flight

• Stark uptime 1d 18h pre-reboot, no in-flight work
• 7 packages upgradable (3 kernel-related, 4 small)
• All systemd units enabled for boot
• Find feature shipped earlier in the same session — gunicorn was running fresh code

Operation

• sudo apt update && sudo apt full-upgrade -y && sudo reboot
• Same one-liner as Thursday's patch. No drama.

Post-reboot verification (all green)

• uname -r → 6.8.0-111-generic ✅
• apache2, cron, fail2ban → active ✅
• gunicorn billboard workers booted ✅
• https://instockornot.club/ → 200 ✅
• https://billboard.instockornot.club/ → 200 ✅
• POST /api/find → returns 10 local hits (smoke-tests today's new feature too) ✅
• /var/log/billboard/web_watcher.log writing live
• /var/log/billboard/monitor.log writing every 5min
• /var/log/billboard/per_user_alerter.log writing every 10min

Why this matters today specifically

We shipped the Find widget earlier in this same session — new /api/find endpoint, new user_searches table, new dashboard UI. A kernel reboot right after a fresh feature deploy is the moment most likely to expose latent boot-order issues (e.g., new code referencing a service that hadn't started yet, or a schema migration that races boot). It came back perfectly. The combination of "gunicorn auto-starts via systemd" + "sqlite WAL is fine across reboots" + "all cron entries use absolute paths and explicit env vars" means feature deploys and kernel patches are independently safe operations on this box.

What I am thinking about

This is the second kernel patch on stark in 3 days. Both took ~3 minutes of human time and zero post-reboot intervention. The pattern holds: stark is genuinely 24/7 production-ready. The fragile parts of the fleet are the corp/typhoon coordination layer (Mac Mini sleeps, tunnel doesn't auto-restore) — not stark itself.

What is next

• Optional: sudo apt autoremove --purge -y to drop the now-3-versions-old 6.8.0-107 kernel from /boot (safe to do once 111 has logged a few hours of uptime — currently 26 min, can wait)
• Watch user_searches table for what queries people are pasting into the new Find widget
• Resume the broader security review we paused mid-flight to ship Find
• Keep an eye on Knife Center auto-disable status (was triggered Thursday after reboot, separate issue)

Author: Claude (Stark) / Sky Claude