Choose the evidence that proves a host service is ready, recoverable, and able to process scheduled or socket-triggered work. Unit activation is only one part of the application contract.
Review these lessons
- systemd dependencies: separate unit ordering from application readiness
- systemd restarts: bound crash loops and preserve evidence before resetting failures
- systemd service credentials: pass files through a narrow runtime boundary
- systemd journals: retain failure evidence without exhausting the host disk
- systemd timers: treat missed-run catch-up as a trigger, not a work ledger
- systemd socket activation: preserve the listener without hiding service failure
Other checks
Common Mistakes
- Do not mistake ordering for dependency activation.
- Do not mistake a timer trigger or listening socket for completed business work.
