Could you please let me know, if
- there exist any repos for batch ingestion, via a custom mapper reduce java code, directly from hdfs to druid.
Have you looked at Druid HadoopIndexTask, It internally uses MR to ingest data, you do not need to write up a custom MR job to ingest data.
- how do I supply the hdfs paths otherwise in a flat file “paths” for batch data ingestion in druid ?
To ingest data from HDFS, you can set your pathSpec for the batch ingestion to the location where your input files are present in hdfs.