The incident has been resolved.

The issue has been resolved for all paid customers.

Backups are being installed successfully. Most customers should have their clean data back by now, we're working on some stragglers.

We've begun replaying backups for affected customers.

The affected time range is March 16 through March 24, and we've identified affected namespaces. We'll start reloading those customers from backup after some additional checks.

We're investigating reports of duplicated events for some or all customers. The affected time range seems to be the last 7 days. We're exploring restoring that data from backups.