This message was deleted.

I think the console may have been redesigned since then

@Gian Merlino yeah i actually did that, we are doing batch data so i already have a sample data wanted to add another column, so i open in data loader and added the column in dimensions but its failing i don't why

for batch you should be able to do the same thing you did originally (data loader, task submit, API, etc) with the addition of the new columns

no can't i add new column to the same datasource?

Druid stores all data in segments that are immutable. You can not just add a column to an existing segment. Any data modification is made by creating new segments that overshadow the old segments. So if you previously ingested a CSV file and it had, say, 10 columns, and now you have a new CSV file with 11 columns and you want the data in Druid to reflect that you can just ingest the new CSV file again with the same datasource name (make sure "Append to existing" is set to False):

an incremental, new ingest can absolutely add new columns though

Oh yes. Absolutely ☝️ . I was just imagining reloading the same data with more columns and was 100% focused on that in my answer.

ok Thanks got it

because Druid assumes that segments are immutable and, once generated, can not be changed. This allows Druid to offer best-in-class scaling abilities. Imagine a single segment being loaded on 3 servers (because maybe you want to support HUGE query load). You want each server to be able to answer queries on top of that segment. These 3 segments have to be identical and not change. If you want to change anything you get Druid to create new segments (that are also immutable) and have these new segments overshadow the previous versions.

there is a video here about how the storage format works: https://imply.io/videos/inside-apache-druids-storage-and-query-engine/. long story short: the immutability of segments plays a huge part in performance and scalability