After a maintenance operation last night (server reboot), a configuration was not taken into account, causing the server to reach a threshold during the increase in traffic this morning. The issue has been resolved. We continue to investigate the exact reason for the problem.