CronDrive
Guides

GitHub Actions scheduled workflow not running: what to check

Updated 15 November 2026

A scheduled workflow can fail to start without a failed run appearing in the Actions tab. Go through these checks in order: first check whether GitHub can start the workflow, then whether the schedule is configured as you expect. The commands use the GitHub CLI (gh); run them from your repository and replace backup.yml with your workflow filename.

1. Is the workflow on the default branch?

A scheduled workflow runs on the latest commit of the default branch, and only if its file is in that branch. A schedule added in another branch does nothing until you merge it. See which branch is the default, and which workflows GitHub finds in it:

gh repo view --json defaultBranchRef --jq .defaultBranchRef.name
gh workflow list --all

If the workflow isn’t in the list, check that its file is in .github/workflows/ on the default branch.

2. Did GitHub disable it?

gh workflow list --all also shows the state of each workflow. If the state starts with disabled, the workflow is disabled and cannot run. Common causes include:

  • In a public repository, GitHub disables the scheduled workflows after 60 days without activity in the repository. GitHub’s docs don’t say what counts as activity; in practice, new commits do, and the scheduled runs themselves don’t. It’s the usual cause in a repository that only exists to run something on a schedule.
  • In a fork of a public repository, scheduled workflows start disabled.
  • Someone disabled it in the Actions tab.

Enable it again in the Actions tab, with the “Enable workflow” button of the workflow, or with:

gh workflow enable backup.yml

Also check whether Actions is disabled for the whole repository under Settings → Actions → General.

If your enterprise uses managed user accounts, check the account of the last person who changed the schedule. If that account was deprovisioned, the scheduled workflow stops running. Someone with write access can reactivate it by committing a change to the cron line.

3. Is the cron expression valid for GitHub?

on:
  schedule:
    - cron: '17 3 * * *'
  • It has five fields, as in a crontab: minute, hour, day of the month, month and day of the week, with *, ,, - and /. Expressions with seconds or ?, such as those used by Quartz or AWS, aren’t valid here.
  • It goes in quotes. In YAML, a value that starts with * means something else.
  • @daily, @hourly and the other shortcuts aren’t supported. Month and weekday abbreviations are supported, as in JAN-DEC and SUN-SAT.
  • The shortest interval is 5 minutes.

Paste the expression in the cron expression tester to see what it means and when it runs next.

4. Is it in the time zone you think?

Schedules use UTC unless you configure another time zone: 0 9 * * 1-5 is scheduled for 9:00 UTC, which is 4:00 or 5:00 in New York, depending on the time of year. To use another time zone, add its name next to the expression:

on:
  schedule:
    - cron: '30 5 * * 1-5'
      timezone: "America/New_York"

When daylight saving time starts, a schedule in the hour that doesn’t exist that day moves to the next valid time: one at 2:30 runs at 3:00. More in the time zone of cron.

5. It runs late, or not every day

GitHub doesn’t guarantee that a run will start at the specified minute. Its docs say that scheduled runs can be delayed when many workflows start at once, especially at the start of each hour. Under heavy load, some queued runs may be dropped.

  • Avoid minute 0. Using 17 3 * * * instead of 0 3 * * * helps avoid the busiest period.
  • Don’t use it for something that must start at an exact time, or that must not skip a day without anyone knowing.

To see when the scheduled runs really started:

gh run list --workflow=backup.yml --event=schedule --limit 10

6. Does the run start, but skip its jobs?

If the run appears in the Actions tab but its jobs (groups of steps) are skipped, check their if: conditions. A condition written for another event, like github.event_name == 'push', is false in a scheduled run. Also, workflow_dispatch inputs are empty. If the workflow has several schedules, github.event.schedule contains the expression that triggered this run:

on:
  schedule:
    - cron: '17 3 * * *'
    - cron: '17 15 * * 1'

jobs:
  weekly-report:
    if: github.event.schedule == '17 15 * * 1'

Run it now

A workflow with only schedule can’t be started by hand. Add workflow_dispatch next to it:

on:
  schedule:
    - cron: '17 3 * * *'
  workflow_dispatch:

Once that is on the default branch, the Actions tab shows a “Run workflow” button, and you can start it and follow it from the terminal:

gh workflow run backup.yml
gh run watch

A manual run lets you check whether the jobs work. To check the schedule too, use gh run list after the next scheduled time and look for a new run.

Know when it stops running

When a scheduled run fails, GitHub notifies one person: the one who last changed the cron line, or who enabled the workflow again. When it doesn’t start, because GitHub disabled the workflow or dropped the run, it notifies nobody.

To find out in both cases, the workflow has to report each run to a monitor that expects it: when it starts, and how it ends. With CronDrive, add a step at the start and another at the end of the job:

jobs:
  backup:
    runs-on: ubuntu-latest
    env:
      CRONDRIVE: https://job.crondrive.com/<token>
    steps:
      - run: curl -fsS -m 10 "$CRONDRIVE/start/$GITHUB_RUN_ID.$GITHUB_RUN_ATTEMPT" || true
      - uses: actions/checkout@v4
      - run: ./backup.sh
      - if: always()
        run: |
          if [ "${{ job.status }}" = "success" ]; then code=0; else code=1; fi
          curl -fsS -m 10 "$CRONDRIVE/end/$code/$GITHUB_RUN_ID.$GITHUB_RUN_ATTEMPT" || true

if: always() makes the last step run when an earlier one fails, and || true keeps a network error in the call from failing the job. In a public repository, keep the URL in a secret: whoever has the token can report runs.

Create the job in CronDrive with the same expression and time zone: UTC, or the value of timezone. Since GitHub can start a run late, allow a grace period for the start; for example, 30 minutes. CronDrive alerts you if a run fails, doesn’t start within that grace period, or exceeds the time limit you set for completion.

CronDrive emails you, or posts to Slack, when a job with this schedule fails, or doesn't start or finish on time. Free for up to 15 jobs.

What you'd see with this schedule

An example with the last runs of this schedule and made-up results: one failed and one didn't run.

Times in UTC.

Due
Sat 14 Nov, 03:17 On time Started 03:17, finished 03:30, took 13 min.
Fri 13 Nov, 03:17 Didn't run Alerts sent at 03:22 and 03:47.
Thu 12 Nov, 03:17 On time Started 03:17, finished 03:31, took 14 min.
Wed 11 Nov, 03:17 Failed with code 1 Alert sent at 03:21. Started 03:17, finished 03:21, took 4 min.
Tue 10 Nov, 03:17 On time Started 03:17, finished 03:30, took 13 min.
Mon 9 Nov, 03:17 On time Started 03:17, finished 03:31, took 14 min.

The alert of the failed run

Email from CronDrive, Wed 11 Nov, 03:21

Failed (exit code 1): "Your job"

The job "Your job" failed with exit code 1.

If you connect Slack, the same message goes to your channel.

The change in your crontab

17 3 * * * curl -fsS -m 10 https://job.crondrive.com/uG2Exw/start/$$; /path/to/job.sh; curl -fsS -m 10 https://job.crondrive.com/uG2Exw/end/$?/$$

Two calls around your command, in the same line. Your job's line comes with its own token.

Monitor this schedule

Or let your AI agent set up all the cron jobs on your server:

Paste this in Claude Code, Cursor, Codex or your AI agent:

Set up CronDrive to monitor the cron jobs on this server. Follow the instructions at https://crondrive.com/docs/agents