read dataset from only one node in YARN cluster

[marc nicole]

Hi,

Spark 3.2, Hadoop 3.2, using YARN cluster mode, if one wants to read a
dataset that is found in one node of the cluster and not in the others, how
to tell Spark that?

I expect through DataframeReader and using path like
*IP:port/pathOnLocalNode*

PS: loading the dataset in HDFS is not an option.

Thanks