The context
I’m the only IT person at a mid-size manufacturer, and a year-end plant shutdown was coming with a long list of outage-window projects. The goal this week was to finish jobs early instead of just preparing them, while still handling day-to-day support. By Friday, more than half of the list was done.
What I did
Migrations and cutovers
- Moved the print server to Windows Server 2025. I built a new VM, migrated every queue and driver, and scripted the switch-over. When the unattended overnight run failed, I found the cause, rehearsed a fixed script, and cut over the next morning with a 5-minute outage.
- Replaced a single-purpose workstation VM with a scripted cutover and automatic rollback, a day early at the user’s request.
- Moved guest Wi-Fi onto the core network in an unattended after-hours cutover, and checked it the next morning. Nobody noticed.
- Brought a new company-owned firewall online next to the existing one, ready for a phased migration.
- Built a new PostgreSQL 16 server. I rehearsed the time-and-attendance database move and automated it for a planned weekend window.
Retirements
- Powered off, replaced or deleted nine legacy VMs. Before retiring one old web server, I moved over 100,000 archived documents to a secured share and verified them by hash.
- Cleaned up stale file-namespace targets left over from earlier migrations.
Security
- Investigated a password-spray campaign against cloud sign-ins (none got in) and tightened account lockout settings.
- Tightened permissions on a sensitive shared folder and removed stale access.
- Triaged phishing reports and confirmed no accounts were compromised.
New capability: security cameras at a remote site
- Rebuilt a retired PC as a Linux video recorder with hardware video decoding and object detection.
- Wrote a network standard that keeps cameras isolated from the rest of the network, and applied it.
- Set up secure remote viewing behind single sign-on.
- Added after-hours person and vehicle alerts with snapshots, plus alerts when a camera goes offline.
Backups and infrastructure health
- Traced data-overage alerts to a nightly offsite copy that kept re-sending a large disk image, and stopped it.
- Got a nightly offsite cloud file copy running clean for the first time (a long-path fix).
- Freed about 1 TB of backup storage by removing orphaned backup chains.
- Updated out-of-band management firmware on three hosts with no host reboots.
- Fixed hardware alert emails that had been failing silently.
- For the year-end cold start, wrote a restart-day checklist and an automated health-check script, and set a tiered VM start order.
Endpoints and users
- Turned on staged patch rings for the PC fleet.
- Recovered about 370 GB of disk space across a dozen machines, and stopped a runaway application archive at its source.
- Got a cloud call-in path on the phone system working over SIP/TCP, which avoided a carrier change request.
Outcome
Two production servers and the guest network moved with minimal or no user impact, nine legacy systems were retired, and the shutdown list moved well ahead of schedule. A new camera system also went from an empty PC to live alerts in two days.