An important consideration when using fail-fast mode is the size of the queue for the logger. Each synchronous logger has a queue that will hold audit records received while the logger's JDBC function is in fail-fast mode. Loggers in fail-fast mode continue to accept, and queue up, audit logging records. The queues for each of these loggers must be sized appropriately. The Maximum Queue Size field on the Settings > Logging > Logger Details page must be large enough to hold all the audit logging records it may receive while the audit logging database is unavailable. If the queue reaches its maximum size, the logger writes the log entries to a file called FailedAudit_yyyymmdd_hhmmss.log in the Integration Server_directory \instances\instance_name\logs directory. You can scan the file to find records that were not logged.