A storage volume suffered errors, causing file read or write operations to be very slow. This in turn caused the API worker pool to be exhausted, causing a high percentage of API requests to be dropped. From the users perspective, the user interface would show a lot of errors or appear to be stuck, or both.
No data has been lost, and documents sent and received via integrations have all been processed normally.