Deploy Skew Outage: Stale Tabs Caused Chunk Errors and Data Corruption After Release

On a routine deployment at 10:00 UTC, a CI pipeline shipped a workspace schema migration along with new hashed JavaScript bundles and pruned old chunks from the CDN. Smoke tests in fresh incognito sessions passed, but within about 12 minutes, users with long-lived tabs reported errors. The first mitigation, forcing Service Worker activation and reloading on chunk errors, made the outage worse by interrupting active transactions. The incident analysis attributes the failure to state divergence across Service Worker registration, HTTP caching, module bindings, bfcache history restoration, and IndexedDB storage, where an older runtime wrote outdated records into the upgraded database. The report concludes that the browser environment held multiple app versions at once, a condition the rollout had not accounted for.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.
Discussion (0)
Log in to join the discussion and vote.
Log in