Skip to content
@deckyard/routeros

Scheduler: scripts on a timer

File: src/topics/scheduler.j. Path: /system/scheduler (SCHEDULER_PATH). Reference: cheatsheet.

The scheduler runs RouterOS script on the router itself, so a nightly backup or a watchdog keeps working while your management host is down. A task's on-event is either script source or the name of a stored /system/script (see script.md). routeros has one call per shape: every interval (scheduleScript), daily at a time (scheduleDaily), once per boot (scheduleAtStartup).

Intervals are validated locally: "30s", "10m", "2h", "1d", "1w", combinations such as "1d12h", bare seconds ("90"), or "HH:MM:SS". Daily times must be "HH:MM:SS" (hours 0-23, minutes and seconds 0-59).

Examples

Nightly configuration backup at 03:00:

jennifer
mt.scheduleDaily($c, "nightly-backup", "03:00:00", "/system backup save name=nightly");

Reboot when the uplink has been gone for good:

jennifer
mt.scheduleScript($c, "uplink-watchdog", "10m",
    ":if ([/ping 1.1.1.1 count=5] = 0) do={/system reboot}");

Log a marker at every boot:

jennifer
mt.scheduleAtStartup($c, "boot-marker", ":log info \"booted\"");

Check that tasks run:

jennifer
def tasks as list of mt.ScheduledTask init mt.scheduledTasks($c);
for (def t in $tasks) {
    io.printf("%s: every %s, next %s, ran %d times\n",
        $t.name, $t.interval, $t.nextRun, $t.runCount);
}

Pitfalls

  • routeros validates the timing, not the script. A typo shows up in the router log when the task fires. Try one-liners in the RouterOS terminal first.
  • A task runs with the policies of the user who created it (the user this client logged in as). A user without the reboot policy creates a reboot task that fails silently every night.
  • runCount == 0 long after nextRun has passed means the task errored; check the router log (log.md).
  • Times are router-local. Set the clock, timezone or NTP (clock.md) before relying on a 03:00 backup.
  • RouterOS quotes inside the source need \" in a Jennifer string.