Impact: All customer databases in the Dallas region experienced unavailability during this window. No data loss occurred.
On July 28th, 2025, the Latitude.sh Databases cluster in our Dallas region experienced a critical failure due to a broader site-level outage. The incident led to the loss of the cluster’s internal state, rendering it unsalvageable. As a result, all customer databases in this region became unavailable.
Our engineering team immediately initiated recovery efforts. We provisioned a new environment, redeployed the database control plane, and restored each customer database from off-site backups.
After 35 hours of continuous work, services were fully restored. All customer data is safe and accessible. However, configuration-level metadata (such as trusted sources) could not be recovered and must be manually recreated.
The failure originated from the control plane node pool of the Dallas cluster. The resulting corruption of the cluster’s internal state made it impossible to safely recover or rejoin the remaining nodes. Compounding the issue was the absence of recent control plane snapshots, which limited restoration options.
We are conducting a full forensic analysis to determine the root cause of the corruption and evaluate the failover mechanisms that were expected to mitigate such an event.
Impact
Immediate Actions Taken
Customer Actions Required
Since all databases were recreated in a new environment, customers must take the following steps: