mirror of
https://github.com/koala73/worldmonitor.git
synced 2026-04-25 17:14:57 +02:00
9feefecd1e
* fix(consumer-prices): prevent Playwright hang from blocking aggregate + publish jobs browser.close() can silently hang after scraping completes, keeping the node process alive indefinitely. This blocked the && chain in the Railway startCommand, so aggregate.js and publish.js never ran. Two-layer fix: - teardown(): race browser.close() against a 5s timeout so Chromium unresponsiveness never blocks teardownAll() - scrape.ts: add a 12-minute hard-kill timer as an ultimate safety net in case any other async handle prevents natural exit * fix(consumer-prices): replace startup watchdog with bounded closePool timeout The 12-minute hard-kill timer started at process startup, meaning a legitimate long scrape would be killed mid-job with exit code 0 — letting aggregate and publish run against partial data. Replace with a 5s race on closePool() in the finally block, mirroring the teardown() fix in playwright.ts. With both hang points bounded, main() always resolves promptly after scraping completes and process.exit() fires immediately with the correct exit code.