Repository navigation
Count the time the polls take toward the release wait, not just the sleeps - #218
Conversation
Each poll's request and jq add to the 30s sleep, so a counter of sleeps alone ran past 5.5 hours and could reach the job's 340-minute limit before the timeout error reported. SECONDS measures the wall clock. Ports hey-sdk f26a09b.
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
There was a problem hiding this comment.
🟢 Approved
The focused timer change correctly includes polling overhead without changing release gating, and no blocking issues remain.
No open issues 🎉
What changed in this PR
Corrects the release wait introduced in #201 so polling overhead counts toward the 5.5-hour budget.
Changes:
- Replaces the sleep-only counter with Bash’s
SECONDS. - Updates elapsed-time logging while preserving polling frequency and release gating.
[!TIP]
If you aren't ready for review, convert to a draft PR.
Click "Convert to draft" or rungh pr ready --undo.
Click "Ready for review" or rungh pr readyto reengage.
| File | Description |
|---|---|
.github/workflows/release-github.yml |
Counts requests, processing, and sleeps toward the release wait budget. |
Review effort: Balanced
#201's wait loop added 30 to
ELAPSEDper sleep, so each poll's request and jq ran on top of the 5.5-hour budget. Over ~660 polls that overhead can push the loop past the job'stimeout-minutes: 340, and the job gets killed before it prints the "Timed out waiting" error and final status.This uses bash's
SECONDS, which measures wall-clock time, so the wait ends at 19,800 s and leaves about 10 minutes of the 340 for the report. It's the same fix as hey-sdk f26a09b, which came from Copilot's finding on hey-sdk#264.The request budget is unchanged: one
actions/runsrequest per 30 s poll, about 120 an hour against GITHUB_TOKEN's 1,000.Summary by cubic
Fixes the release wait loop so wall-clock time, not just the sleep intervals, counts toward the 5.5-hour budget.
ELAPSEDcounter added 30 s per sleep, ignoring each poll's API request andjqprocessing, so ~660 polls could push the loop past the job's 340-minute timeout before the "Timed out waiting" error and final status printed.SECONDSmeasures wall-clock time, ending the wait at 19,800 s and leaving about 10 minutes of the 340 for the report.actions/runsrequest per 30 s poll.Written for commit f700e55. Summary will update on new commits.