Re: [druid-user] What happens to the existing data in Druid segments during Hadoop Indexing?

Hi John,

  1. No. Segments are immutable. i.e till new segments are completely loaded, the old segments stay intact and respond to queries.
  2. Only when ingestion succeeds, then only the new segments are with old. If ingestion fails, it has no impact on existing segments & queries.
  3. I dont think there is a way to specify partition columns, iam not sure.

FYI: If you need sample examples, you can refer to my self-help repo

Regards, Chari.

+1 on 3) - I don’t believe it’s possible to use any elements of the file location in something like a regex transform to create a dimension value. This has been raised before in the context of using the filename as a dimension value, though - you may want to raise a PR on github?