By
default, the SAS Data in HDFS engine does not compute
the number of observations in a SASHDAT file. This improves performance
for SASHDAT files that are distributed among a large number of blocks,
or for HDFS directories that have a large number of SASHDAT files.
When you specify GETOBS=YES, the engine calculates the number of observations
in a SASHDAT file.
ods select attributes;
proc datasets library=hdfs;
contents data=sales2012(getobs=yes);
run;