ci: ping a liveness heartbeat after successful fetch runs - #104
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
Merge as is — external heartbeat closes the fetch workflow's liveness gapThe liveness gap is real and not covered by the existing Axiom telemetry: that signal is emitted only after the fetch job starts, while this change adds a final Merge state: The PR is open and marked mergeable. The recorded CI state contains completed checks plus checks still in progress or queued; review-thread resolution and any gating human review state are unknown because no thread-state record is available. Evidence · 2 verified · 2 not verifiedVerified
Not verified
Read the full one-file diff, changed workflow, context pack, repository memories, sibling workflows, CONTRIBUTING.md, git history, PR status, and conversation context; searched the checkout for heartbeat and liveness primitives. No tests or PR code were executed. Prior-review and review-thread records were unavailable; the recorded CI state remained partially unsettled.
Comment Review by Rupic. |
Deploying with
|
| Status | Name | Latest Commit | Preview URL | Updated (UTC) |
|---|---|---|---|---|
| ✅ Deployment successful! View logs |
modelpedia-api | 4bd89ae | Commit Preview URL Branch Preview URL |
Aug 31 2026, 05:19 AM |
problem
the daily fetch workflow has no liveness signal independent of the repo itself: if the cron is disabled, the workflow file breaks, or GitHub stops scheduling it, nothing is left alive to report a failure (the Axiom telemetry step only reports runs that happen).
change
ping a Better Stack heartbeat (
BETTERSTACK_HEARTBEAT_URLrepo secret, already set) as the final step on successful runs only; a missing secret skips silently. the heartbeat expects one ping per day with a 6 hour grace window.verification
yaml lints clean; the step degrades to a workflow warning when the ping fails.