Every 12 Hours Cron – Free, Online, Private

The expression

0 */12 * * * is a five-field cron expression: minute, hour, day of month, month, day of week. There is no seconds field and no year field. On the hour, every 12 hours, every day The next five run times on this page are computed in your browser with the same expression, in UTC, so a reload later in the hour shows a later list. Edit the box if you want to nudge a field and watch the list change, or open the line in the cron parser.

What each field is doing

The minute field is the single value 0. No other value in the 0–59 range matches. The hour field is */12, a step of 12 across the whole 0–23 range. The matches are aligned to the start of that range. It does not mean "the value 12" and it does not start from the moment you save the file. The day-of-month field is an asterisk, so every value in the 1–31 range is allowed. This field is not what limits the schedule. The month field is an asterisk, so every value in the 1–12 range is allowed. This field is not what limits the schedule. The day-of-week field is an asterisk, so every value in the 0–6, Sunday as 0 or 7 range is allowed. This field is not what limits the schedule.

When the day-of-month and day-of-week fields are both something other than *, many cron implementations run the job if either field matches. This preset leaves at least one of those two fields as *, so you do not have to resolve that rule to predict it. Names such as MON and JAN are readable and widely accepted. The snippets use numbers so a strict parser does not have to know the names.

Why this schedule

Twice a day, aligned to noon and midnight, is the twelve-hour step. It is different from a list of two business hours. Use this when the two runs should be opposite each other on the clock, for example a refresh before the UTC day starts and another before the UTC afternoon.

A full reindex, a large file mirror, and a report that management reads morning and evening are the jobs that fit. Split them if one of the two times is a high-traffic hour for the database.

Where it goes wrong

0 0,12 * * * is the same two hours written as a list. */12 is not 'every twelve hours from now'. If the process was down at 12:00, cron does not run it at 12:07 to catch up, unless the daemon you use has a catch-up feature. Most do not. The next run is 00:00.

A sixth field is the seconds field used by Quartz, some Spring apps, and some Kubernetes setups that are not the classic CronJob schedule. Classic crontab, GitHub Actions, and the schedule field of a batch/v1 CronJob are five fields. If you paste a six-field line into those systems, the extra token shifts every field one place to the left and the job runs on a different day. L, W, and the # 'nth weekday' modifier are real in some daemons and rejected or reinterpreted in others. They are not used on this page. The English description and the next runs both refuse to guess at a token the calculator does not share with every target.

Crontab, GitHub Actions, and Kubernetes

GitHub Actions accepts this cadence. The schedule is UTC, it can start late when the service is busy, and it is dropped if the repository has had no activity for sixty days. The workflow file only needs the expression. It does not need a seconds field.

A user crontab does not include the username. A system crontab, the kind in /etc/crontab, inserts the user between the expression and the command. The line below is the user form. Kubernetes CronJob has used spec.timeZone since 1.27. The manifest sets it to Etc/UTC so the schedule matches the next-run list. Remove the field on a cluster older than that, and know that the schedule then uses the controller's zone.

crontab0 */12 * * * /usr/local/bin/job

GitHub Actionson:
  schedule:
    - cron: "0 */12 * * *"

Kubernetes CronJobapiVersion: batch/v1
kind: CronJob
metadata:
  name: every-12-hours
spec:
  schedule: "0 */12 * * *"
  timeZone: "Etc/UTC"
  jobTemplate:
    spec:
      template:
        spec:
          containers:
            - name: job
              image: busybox:1.36
              args: ["echo", "every-12-hours"]
          restartPolicy: OnFailure

• Keep the command idempotent. Cron does not remember a failure and it does not skip a run that is already in progress.
• Write the timezone next to the expression. UTC in the file and local time in the ticket are how a job fires eight hours early.
• Offset the minute if several jobs would otherwise start at once.