This message was deleted.

Druid ingests timestamps in millisecond precision. If you need to keep the nano, you need to keep a copy column and use __time 'just' for the index

I use scan query in following format: select __time, dimension1, dimension2 from DATASOURCE ORDER BY __time. I think order by is limited to __time column. It does not work on other columns.

the precision of nanoseconds is lost after the ingestion you gonna need to reingest saving the data in another column either as string, or just the nanoseconds as long, and then order by both by __time then the nanosecond