The servers, deploys and pipelines are still running, which is the good news. The risk is that nobody knows how, and the first incident or failed deploy will show it. This is how I run the first week.
List every account the person who left could reach: cloud console, hosting panels, registrar, DNS, SSH keys, CI/CD and deploy keys, the password manager. Change what matters most first, and remove their keys from the servers.
An inventory of every server, container, pipeline and scheduled job, what each does and who it serves, built from the servers themselves and the cloud bills, not from memory.
Not "is the job green" but "can we restore it". One test restore of the thing you'd miss most.
Monitoring and alerts usually go to the person who left. They now come to me, with thresholds that mean something.
Plain-English runbooks for the common jobs, the deploys and the scary ones, and a short list of risks in order, so you know what to fix first.
You don't need to decide on day one. Once it's covered and written down, you can decide without the pressure.
For many small teams the servers are not a full-time job. A fractional lead costs a fraction of a salary.
The fractional lead →Cover during the search and notice periods, then a proper handover to your new hire.
Managed →Some teams just need the week: access, inventory, runbooks. That's a fixed project.
Pricing →A UK software company's sysadmin left. Two products in production, ISO 27001 to keep, and customers who expect answers on security.
Took over access, listed every server, wrote the runbooks, then stayed on as the part-time infrastructure lead.
Four years later I'm still part of the team.
"One of those rare suppliers who feels more like a trusted member of the team than someone on the other end of a ticket system."
Angus B. · Founder, UK SaaS company"A trusted partner responsible for everything from AWS infrastructure, networking and security to operational support, monitoring and MySQL performance."
Jake Rudkin-Wilson · Head of Product Development, CameoUsually within a week. The first call is 15 minutes; if it fits, I can start on access and the inventory straight after.
Whatever access you can find: cloud console, hosting control panels, SSH keys, the password manager, DNS. Missing pieces are normal; finding them is part of week one.
It's common and it's fine. Everything on a server can be found from the server. It takes longer, not forever.
No. Some teams keep me instead of hiring; others hand over to a new hire with the runbooks I wrote.
A 15-minute call this week, and I can usually start within days.