On 2026-05-13 at 06:09 UTC, a scheduled maintenance routine inadvertently deleted the Horizon production database, causing a full system outage. Our team identified the issue and restored the database from the most recent available backup. Horizon was brought back online at 07:04 UTC, resulting in approximately 55 minutes of downtime.
The restored backup was from 05:15 UTC, which means all data created or modified between 05:15 UTC and 06:09 UTC (approximately 54 minutes) has been permanently lost. No data was lost during the downtime window itself, as the system was inaccessible during that period.
We made a mistake. This incident occurred in the context of our ongoing infrastructure migration from AWS to Cycle, which we are undertaking to move our hosting from the US to the EU for improved data locality and compliance. During this migration, a scheduled maintenance routine was not properly adapted to the new environment, causing it to delete the production database instead of performing its intended operation.
We take full responsibility for this error.
The faulty routine has been immediately disabled. We are now auditing all scheduled routines in the new environment to ensure nothing similar can occur. Beyond that, we are implementing safeguards to prevent automated deletion of production databases.
We sincerely apologize for the disruption and data loss. If you believe you are affected, please reach out to our customer care team and we will do our best to assist you.