Live status across ingestion, storage, and webhook delivery.
We publish these regardless of severity.
A queue backlog in one region delayed webhook delivery by up to 6 minutes for roughly 40 minutes. Uploads themselves were unaffected. Root cause was a misconfigured autoscaling threshold, since corrected.
A deploy introduced a regression in request validation, causing roughly 2% of requests to fail for 11 minutes before automatic rollback.
Prevents duplicate storage costs from client-side retry logic on large files.
Projects can now pin new uploads to eu-west in addition to eu-central.
Existing shared-secret verification continues to work until the next major version.