The discussion suggests using a FLUSH command to ensure all records inserted before the command are viewed/sinked. However, this may not be suitable for high-frequency updates. Another approach is to use a 'neutral' record with a timestamp to simulate progress tracking. Additionally, exposing Kafka consumer offsets or using a custom timestamp forwarded to all sinks could be considered. The idea of progress messages in outputs and a configurable interval for these messages was also discussed. Ultimately, the suggestion is to use a timestamp column in the upstream table to report progress, which RisingWave can use to refresh metadata and indicate the processing status of the data.