On August 25, 2025, a significant increase in timeouts occurred on the Connector API's restrictions/getAll endpoint. This was triggered by a sudden, unexpected increase in external traffic following a customer’s marketing campaign.
We worked with the customer to reduce the impact on the system by temporarily reducing their rate limits. Additionally we took steps to limit the impact for other Connector API users by switching the reads in the operation to database replica.
The underlying cause was too strict usage quota on restrictions/getAll operation, causing repeated timeouts and subsequent cascade failure in external system.
We took steps to prevent the likelihood of similar issues in restrictions/getAll operation by reducing load on the primary database and increased usage quota.
For further planned improvements, we will be reviewing usage patterns and performance of restrictions/getAll operation. We also plan on providing Connector API users with better guidance in case of timeout errors and best practices for fault tolerance.