Since the aggregation is not done by influxdb a fill option is not available and instead needs to be performed manually. As such the order of writing must be from oldest to newest so that the list of product seen before can be kept and missing products for a given interval can be filled.