We have had multiple BFs where audit-log-enabled mongod's were started too soon after each other (within the same second) and one crashed due to its audit path already existing. We should investigate whether we can automatically give different instances different audit log paths to avoid this in the future.