Skip to content

Scheduled e2e workflow failures are silent (no notification) #97

Description

@dboreham

Problem

The e2e-test.yml workflow runs on a weekly schedule (cron: "0 6 * * 1", Mondays 06:00 UTC) in addition to PR/merge triggers. When the scheduled run fails, no failure notification is delivered.

This was discovered while working on #95: the DigitalOcean e2e job had been failing for some time (an expired E2E_DO_TOKEN), but no failure emails were received, so the breakage went unnoticed until it surfaced on an unrelated PR's checks.

Why it happens

GitHub only emails scheduled-workflow failures to the account that last modified the workflow file, and it stops sending after repeated failures on the same schedule. So the missing emails are expected GitHub behavior, not a separate bug — but it means a broken weekly run can silently rot.

Suggested fix

Add an explicit failure-notification step to the scheduled path of e2e-test.yml so breakage is surfaced reliably. Options:

  • A notification step gated on if: failure() && github.event_name == 'schedule' (Slack webhook, email, or gh issue create to auto-open a tracking issue).
  • Alternatively/additionally, a lightweight monitor that alerts if the weekly run hasn't succeeded recently.

Related

Notes

Credentials rotate/expire periodically (DO token, Vultr API key, GCP creds), so this will recur without proactive alerting.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions