GitHub Actions scheduled workflow runs are intermittently missing while workflow_dispatch works

[NoCheatDev1]

🏷️ Discussion Type

Bug

💬 Feature/Topic Area

Schedule & Cron Jobs

Discussion Details

Hi GitHub Community,

I am experiencing intermittent missing scheduled runs for a GitHub Actions workflow.

Repository:
https://github.com/NoCheatDev1/ShopGuard

Workflow:
ShopGuard monitor
.github/workflows/monitor.yml

Default branch:
main

The workflow has both workflow_dispatch and schedule triggers.

Manual workflow_dispatch runs work correctly and the workflow completes successfully when started.

The issue is specifically with scheduled executions.

What we have verified:

• Repository is active and not archived.
• Actions are enabled.
• Workflow exists on the default branch.
• Workflow is active.
• workflow_dispatch works consistently.
• The production workflow completes successfully when it starts.
• A separate minimal cron-test workflow in the same repository also received Scheduled events.
• We disabled and re-enabled the production workflow.
• We deleted and restored the workflow to verify workflow registration.
• There are no workflow execution errors explaining the missing scheduled runs.

We tested these schedules:

17,47 * * * *

5,35 * * * *

The current schedule is:

*/10 * * * *

The problem is that expected scheduled runs sometimes do not appear in the Actions history at all, or appear significantly later than expected.

GitHub documentation says scheduled workflows can be delayed during high Actions load and that queued scheduled jobs can sometimes be dropped. I understand that exact execution time is not guaranteed.

However, I would like to understand whether the behavior we are seeing is normal scheduler behavior or whether there may be a schedule-event registration/delivery issue.

Has anyone experienced the same behavior with a private repository where:

• workflow_dispatch works
• schedule is present on main
• the workflow is active
• Actions are enabled
• but schedule events intermittently do not appear?

Any guidance on how to distinguish a delayed scheduled run from a schedule event that was never delivered would be appreciated.

Thank you.

[4Raisan]

The distinction you are looking for is observable only once GitHub creates a run: list runs with event=schedule and compare their creation/start times with the expected UTC slots. If no run exists, the public run API cannot tell you whether an internal event was delayed, dropped or never generated. Manual success and the working cron-test prove useful parts of the setup, but not delivery guarantees for every production slot.

Keep a short table of expected UTC slots, observed run IDs, creation/start times, workflow SHA and the cron expression in effect. Include several missing slots in a Support ticket rather than repeatedly deleting/restoring the workflow, which changes the evidence.

For a monitoring service, store the last successful check and make the next run catch up missed intervals; use an external scheduler with authenticated workflow_dispatch if your required timing cannot tolerate best-effort schedule delivery. GitHub documents delays/dropped scheduled jobs here: https://docs.github.com/en/actions/reference/workflows-and-actions/events-that-trigger-workflows#schedule.

Disclosure: drafted with OpenAI GPT-6.