Replies: 2 comments
|
Your measurement (delay happens before the run is created, jobs start promptly once it exists) points at the scheduler itself, not your cron or runners. Nothing in your config looks wrong. What the docs say about You're not alone. #210888 shows the same pattern: a weekday JST schedule running about 5h late since early October, with no config cause found. Can you see why? As far as I know, not from your side. No run object exists until the event fires, so there's nothing to log. What you can do is collect clean evidence for GitHub Support: GitHub-native mitigation. This doesn't make on:
schedule:
- cron: '20 10 * * 1-5'
- cron: '50 10 * * 1-5'
- cron: '20 11 * * 1-5'
workflow_dispatch:
permissions:
actions: read
contents: read
concurrency:
group: daily-job
jobs:
check:
runs-on: ubuntu-latest
outputs:
skip: ${{ steps.c.outputs.skip }}
steps:
- id: c
env:
GH_TOKEN: ${{ github.token }}
run: |
n=$(gh run list -R "$GITHUB_REPOSITORY" --workflow <your-file>.yml \
--status success --created "$(date -u +%F)" --json databaseId --jq length)
echo "skip=$([ "$n" -gt 0 ] && echo true || echo false)" >> "$GITHUB_OUTPUT"
main:
needs: check
if: needs.check.outputs.skip == 'false'
# ...your existing jobI'm not sure this helps much. If the backlog delays every scheduled event equally, the extra entries will be late too. It mainly protects you when individual events are dropped or delayed. If you need it at a fixed time, an external trigger calling |
|
Follow-up: Additional evidence from two repositories Hello, I would like to provide an update to my original report regarding GitHub Actions scheduled workflow delays. Our latest investigation reviewed 48 scheduled runs in one private repository and 7 runs in another. All 55 runs were reported as delayed by more than five hours. The investigation indicates that the main delay occurs between the scheduled cron time and workflow run creation. Once a run is created, the job generally starts within a few seconds. We also observed that workflows scheduled for the same time in two separate repositories were registered within approximately 1–5 minutes of each other on four different dates. The internal root cause is still unknown. Could the GitHub team investigate whether these delays are related to scheduled event delivery, internal scheduling queues, or workflow run registration? We would appreciate any information about known issues and possible mitigations. Thank you for your assistance. |
Uh oh!
There was an error while loading. Please reload this page.
🏷️ Discussion Type
Question
💬 Feature/Topic Area
Schedule & Cron Jobs
Discussion Details
Hello,
I’m experiencing consistent multi-hour delays with GitHub Actions scheduled workflows.
Configuration
Observed behavior
Over 10 scheduled sessions between October 13 and October 26, 2026, the median delay between the scheduled time and workflow run creation was approximately 4 hours and 58 minutes.
For example:
The workflow jobs generally begin shortly after the runs are created. This suggests the primary delay occurs before run creation, rather than during runner assignment or workflow execution.
Questions
The repository is private, so I have omitted identifying details.
Any technical guidance would be appreciated. Thank you!
All reactions