Skip to content

Cron jobs: work that repeats

A cron job is an assignment that repeats on its own: a report every Monday, a review every afternoon, a summary on the 1st of every month. You ask once, it is approved once and from then on the installation’s clock takes care of it.

In Settings, Cron jobs tab. At the top, the form to request one; below, the list of what exists. The list refreshes on its own every 15 seconds and the header says at what time it last did: there is no refresh button.

Field What to write
Name A short identifier with no spaces, for example informe-lunes. It is unique.
Role Who is going to run it. Only active roles appear.
When The repetition rule in five-field cron format: minute, hour, day of the month, month and day of the week. Below it, in words: “Every Monday at 07:00”.
Instruction What it has to do, written as you would say it to a person. This is what the role receives every time.

On saving, the cron job is born pending. A pending cron job does not run: as far as the clock is concerned, it does not exist.

Expression Meaning
0 7 * * 1 Every Monday at 07:00
30 18 * * 1-5 Monday to Friday at 18:30
0 8 1 * * The 1st of every month at 08:00
*/15 * * * * Every 15 minutes

If the pattern is unusual, the list shows the exact expression instead of a half-baked translation.

  • The owner and the CEO can create cron jobs and they are approved straight away.
  • Any role can request its own, and another role’s as well, but in that case the cron job stays pending until management approves or rejects it.

Approval is what switches it on: until it is approved, the clock does not see it.

Button What it does
Approve Switches it on: the next run is calculated and the clock will take it into account from that moment.
Reject Voids it and marks it as rejected. It does not run and cannot be reactivated without requesting it again.
Stop / Activate Silences an approved cron job without deleting it, or starts it again.
  1. Every few seconds the clock checks whether any cron job is due.
  2. If the cron job already has a live run, this one is skipped: the same task never runs twice at once. It is recorded as “skipped”, not as an error.
  3. If nothing is running, the role receives the assignment through the same door as any office assignment: its model, its budget and its log.
  4. When it finishes, how long it took and what happened are stored.

A cron job that fails retries up to three times. If all three fail, the owner and the CEO are notified: they decide whether to retry or to hand it to whoever should own it. No silent retries all night long.

  • Times follow the installation’s time zone (by default, Europe/Madrid), with daylight saving applied: 07:00 is 07:00 in summer and in winter.
  • Each role can have a limited number of cron jobs at once (ten by default). At the limit, the application refuses more and says so.
  • It does not rush work: if the team is busy, it waits.
  • It does not run twice because it failed: retries go one after another.
  • It does not change on its own: if the instruction stops being useful, you stop it and request another one.