Repository navigation
v0.9.1
Patch Changes
-
f94cae8: Large metadata ZIPs no longer time out on slow runners, and a failed metadata upload no longer prints success lines.
What failed before: the metadata ZIP and the coverage report are sent through the Scry upload service with a fixed 60 s timeout and no retry. With scry-sbcov 0.6+ (2× root captures) a 470-story Storybook's metadata ZIP is 28 MB, which a self-hosted runner uploading at 0.4-0.85 MB/s cannot send in 60 s: the deploy ended
NOTHING WILL BE INDEXED, and the log still printed✅ Archive uploaded.and✅ Upload complete.around the failure.What you see now:
- Both uploads get a timeout scaled to their size: max(60 s, size ÷ 100 KB/s), at most 15 minutes (28 MB → about 4.7 min). They are retried up to 3 attempts, with backoff, on a timeout, a 5xx/429 or a network error, and never on another 4xx. Each attempt logs its size and how long it took:
Metadata ZIP: sent 28.0 MB in 71.3 s (attempt 1/3). - When the metadata upload still fails, no success line follows it. The deploy ends red with
NOTHING WILL BE INDEXEDand names the build, which is left pending (the upload service cannot mark it failed yet). Re-run the job to index it. SCRY_UPLOAD_TIMEOUT_FLOOR_MSchanges the 60 s floor.
What you must do: nothing.
- Both uploads get a timeout scaled to their size: max(60 s, size ÷ 100 KB/s), at most 15 minutes (28 MB → about 4.7 min). They are retried up to 3 attempts, with backoff, on a timeout, a 5xx/429 or a network error, and never on another 4xx. Each attempt logs its size and how long it took: