I have figured it out in the meantime - simply when moving file on HDFS it
preserves its time stamp and on the other hand the spark filestream adapter
seems to care as much about filenames as timestamps - hence NEW files with
OLD time stamps will NOT be processed - yuk
The hack you can use is to
I have figured it out in the meantime - simply when moving file on HDFS it
preserves its time stamp and on the other hand the spark filestream adapter
seems to care as much about filenames as timestamps - hence NEW files with
OLD time stamps will NOT be processed - yuk
The hack you can use is to