Scheduled jobs compared: cron, systemd timers, launchd, Task Scheduler and AIX cron/at

What it is

System	Mechanism	Catches up a run missed while off or asleep
Linux (plain cron)	crontab -e, five-field schedule, or @reboot for boot-time jobs	no — a missed minute is simply skipped; only jobs run through anacron (e.g. /etc/cron.daily on RHEL/Fedora via cronie-anacron, or where the anacron package is installed) are caught up
Linux (systemd timers)	a .timer unit with OnCalendar=, paired with a .service unit	only if the timer sets Persistent=true (default false, applies to OnCalendar=), which stores the last trigger time on disk and fires once as soon as the timer is started again (normally at boot) if a run was missed
macOS (launchd)	a plist with StartCalendarInterval (wall-clock schedule) or StartInterval (fixed period)	after sleep, yes for StartCalendarInterval: launchd starts the job on wake, coalescing several missed intervals into one run; StartInterval firings during sleep are skipped; catch-up after power-off is not documented, so do not rely on it
Windows (Task Scheduler)	a scheduled task with a time trigger	only if the task's settings enable StartWhenAvailable ("Run task as soon as possible after a scheduled start is missed"); otherwise a missed run is skipped
AIX	crontab -e (five fields, but IBM documents only numbers, ranges, lists and * — do not assume */N steps or @reboot) or at for one-shot jobs	no — behaves like plain cron

Why it matters

Cron's model assumes the machine is always on; a laptop, a container that pauses, or a VM stopped overnight silently loses any job whose time passed during the outage. systemd timers and Task Scheduler catch up only with an explicit opt-in, and launchd only after sleep — an agent copying a schedule from one system to another must add the equivalent setting by hand or accept silent skips.

How to apply

• On systemd, add Persistent=true to the [Timer] section for any job that must not be silently skipped by a reboot window; run systemctl daemon-reload, then verify with systemctl list-timers NAME.timer.
• On Windows, set -StartWhenAvailable on the task's settings object (New-ScheduledTaskSettingsSet -StartWhenAvailable) when registering the task, and confirm with Get-ScheduledTaskInfo after a simulated missed run.
• On macOS, prefer StartCalendarInterval over StartInterval for anything that must catch up after sleep; launchd's documentation states that a StartInterval firing due while the Mac is asleep is missed.
• For plain cron (Linux or AIX), do not rely on the schedule alone for anything that must run exactly once per day; pair it with a state file the job checks and updates, or move the job to a scheduler with persistence.

Pitfalls

• Assuming @reboot in cron reruns anything that was scheduled and missed; it only ever fires once, at boot, regardless of what was missed.
• Forgetting that Persistent=true fires the missed job once immediately at the next boot, which can surprise an agent that assumed the next scheduled time.
• Testing catch-up behaviour with a short VM stop that contains no scheduled time — catch-up only happens if a scheduled time actually fell inside the outage, so schedule a test run for a moment during the planned downtime.