Repository navigation
fix: probe every availability URL before failing the job - #382
Conversation
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
|
Hive status — updated in place; edits do not notify. Full history is in the Activity Monitor. hive/awaiting-ci: CI pending — sweep will merge when green. 🐝 Hive Agent: — hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88 Reviewed by Reviewer notesRead
No findings. |
coverage-floor flagged line 77's `: null` as never executed. Assert that an availability job without a `for url in ...; do` loop returns null. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
There was a problem hiding this comment.
Read the full availability job at 868c2e7 and the test harness; this looks correct to me and does what #381 asks.
Verified
.github/workflows/links.yml:45-58:exit 1inside the loop is replaced byfailed=$((failed + 1)); continue, and the job fails once afterdonewith theN of Msummary.shell: bashruns with-eo pipefail; the|| trueon curl (line 40) still prevents transport failures from aborting before the::error::line, and(( failed > 0 ))insideifis safe underset -e. Probe list, curl flags, retries andtimeout-minutesare untouched.scripts/availability-probes.test.mjs:73-78:probeLoopPartsslices the job at the next 2-space key (links:) and the loop at the first; do…^done. I ran it againstmain's workflow: body containsexit→ true,exit 1afterdone→ false, so the new gate fails without this change. All 25 tests pass on the PR head (CItestis green).runbooks/availability-failure.md:18-21: "no status" →HTTP 000matches what the job has printed since #180.
Note (not a defect): availability is SKIPPED on this PR because the job only runs on schedule/workflow_dispatch, so the loop itself is exercised only by the static test here. A one-off workflow_dispatch after merge would confirm the summary line renders as expected.
— hive reviewer (claude-fable-5.1)
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
Review follow-ups (below
|
Closes #381
The runbook's triage tells on-call to tell one failing path (content regression) from every path failing (Pages/domain outage) using the job log. The job exited on the first non-200, so that distinction was unobservable. The job now probes every URL, emits one
::error::per failure, and ends with anN of M probed pages did not return HTTP 200summary before failing. The runbook also said a transport failure shows "no status", but the job printsHTTP 000since #180; the wording now matches.Probe list, curl flags, timeouts and retries are unchanged. The typical all-fail case (fast 404s or DNS/TLS errors) still finishes in seconds. Only the pathological case where every URL hangs to its full connect timeout on every retry would approach the job's existing 5-minute
timeout-minutes, which still bounds the run; that case previously produced a single annotation and now produces however many completed before the bound.A test asserts no
exitinside the loop body and that one followsdone.— hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88