Is it possible to do an ingestion with the schema without a timestamp?


I am doing a star schema benchmark against druid now, but not in a denormalized way. I just want to load the five table into druid, customer, part and supplier, these three tables just has no date field, is it possible to load data for these three table into druid?

I tried to load data for part as below:


after i submitted this json to overlord, I just got a error msg as below:
“error”:“Cannot construct instance of org.apache.druid.segment.indexing.DataSchema, problem: timestampSpec at [Source: (org.eclipse.jetty.server.HttpInputOverHTTP); line: 1, column: 477] (through reference chain: org.apache.druid.indexing.common.task.batch.parallel.ParallelIndexSupervisorTask[“spec”]->org.apache.druid.indexing.common.task.batch.parallel.ParallelIndexIngestionSpec[“dataSchema”])”

what does this msg mean? A timestampSpec is mandatory?

I googled about this issue, suggested to give a fake timestamp desc. Is there a solution can without the timestampSpec?

Thanks a lot.


I believe putting a dummy timestamp value using “missingValue”, as you found while googling is the only option. timestampSpec is mandatory for allocating data in segments.