Cloudflare incident on March 21, 2025

Cloudflare reports a 1h07m outage (21:38–22:45 UTC, Mar 21, 2025) affecting R2 and dependent services after new storage credentials were mistakenly deployed to a non‑production Worker during key rotation, leaving production unable to authenticate; this caused 100% write failures and ~35% read failures globally (with varied downstream impact to Cache Reserve, Images, Stream, Vectorize, Billing, etc.), but no data loss. Root cause was omission of the --env parameter in wrangler commands; service recovered after deploying the correct credentials, and Cloudflare committed to improved logging/validation, enforced deployment tooling, stronger SOPs, and enhanced observability and health checks.