Problem
This issue stemmed from a faulty code push approximately 12 hours ago. The code added a historical versioning feature for booth caching. Due to insufficient testing during local development, several obvious errors in the code went unnoticed: the code incorrectly assumed that all booth metadatas are difference, leading to the repeated insertion of identical data into the database. Because the metadata was large, each access resulted in an insertion, quickly consuming server disk space. To make matters worse, the server's disk space was already nearly full at the time. Since the forum software didn't report any errors, and server monitoring showed no anomalies, everything "appeared" normal.
Solution
We have reverted the code changes and upgraded the server specifications.
Summary
We sincerely apologize for the prolonged service outage and the lack of response. This incident highlights the need for improved server health monitoring and tracking.