Linux administration, one known state at a time.
Begin by knowing what is running and what should be reachable. Make changes in small steps, verify the result, and leave enough context for the next person to understand the system.
Inspect before changing
Record the distribution, storage layout, running services, and listening sockets. Identify unexpected components before disabling them. A provider agent or storage service can be important even when its name is unfamiliar to the person inheriting the server.
Apply boundaries at the right layer
Host-based firewall rules and provider network controls may both affect access. Test the real administrative route before tightening either one. Give application processes only the file and service access they need, rather than resolving every problem through a broad permission grant.
Own the maintenance loop
Define who reviews software updates, when disruptive changes are scheduled, and how success is checked. Maintain a recovery inventory that includes configuration and state. Keep the runbook updated when the environment changes, not only after an incident exposes a gap.