In Druid it's not possible to have a datasource without any segments. But is possible, in theory, to have an empty datasource: you would need a single segment that has no rows (the important part is that the segment exists, not that it actually has rows in it). But there are two problems with this:
1) Druid ingestion really tries to avoid creating segments with zero rows. So it may not be possible without hackery. 2) Druid SQL will probably not work right on a datasource with zero rows in it. (I don't think any of the columns, etc, will be detected properly.) I would consider this an unsupported use case currently. On Mon, Dec 23, 2019 at 11:05 AM David Glasser <glas...@apollographql.com> wrote: > Hmm, I may have answered too quickly. That will lead to no segments being > created, so the system will not consider the data source to exist, I think. > > On Mon, Dec 23, 2019 at 11:04 AM David Glasser <glas...@apollographql.com> > wrote: > > > You can do batch ingestion from an empty file, as documented (with > > multiple ways of invoking it) at > > https://druid.apache.org/docs/latest/tutorials/tutorial-batch.html > > > > On Mon, Dec 23, 2019 at 6:35 AM Andriod Guru <ashish...@gmail.com> > wrote: > > > >> Hi Team, > >> > >> I am trying to automate Druid pipeline using stream sets and one of the > >> use case demands an empty druid data source should be created for users. > >> > >> Do we have opportunity to create one? > >> > >> Also how can I start contributing to Druid releases. I have been using > >> Druid since 0.14 version and have fair idea about Java. > >> > >> Thanks, > >> Ashish > >> > >> --------------------------------------------------------------------- > >> To unsubscribe, e-mail: dev-unsubscr...@druid.apache.org > >> For additional commands, e-mail: dev-h...@druid.apache.org > >> > >> >
