问题
I have the following folders in HDFS :
hdfs://x.x.x.x:8020/Air/BOOK/AE/DOM/20171001/2017100101
hdfs://x.x.x.x:8020/Air/BOOK/AE/INT/20171001/2017100101
hdfs://x.x.x.x:8020/Air/BOOK/BH/INT/20171001/2017100101
hdfs://x.x.x.x:8020/Air/BOOK/IN/DOM/20171001/2017100101
hdfs://x.x.x.x:8020/Air/BOOK/IN/INT/20171001/2017100101
hdfs://x.x.x.x:8020/Air/BOOK/KW/DOM/20171001/2017100101
hdfs://x.x.x.x:8020/Air/BOOK/KW/INT/20171001/2017100101
hdfs://x.x.x.x:8020/Air/BOOK/ME/INT/20171001/2017100101
hdfs://x.x.x.x:8020/Air/BOOK/OM/INT/20171001/2017100101
hdfs://x.x.x.x:8020/Air/BOOK/Others/DOM/20171001/2017100101
hdfs://x.x.x.x:8020/Air/BOOK/QA/DOM/20171001/2017100101
hdfs://x.x.x.x:8020/Air/BOOK/QA/INT/20171001/2017100101
hdfs://x.x.x.x:8020/Air/BOOK/SA/DOM/20171001/2017100101
hdfs://x.x.x.x:8020/Air/BOOK/SA/INT/20171001/2017100101
hdfs://x.x.x.x:8020/Air/SEARCH/AE/DOM/20171001/2017100101
hdfs://x.x.x.x:8020/Air/SEARCH/AE/INT/20171001/2017100101
hdfs://x.x.x.x:8020/Air/SEARCH/BH/DOM/20171001/2017100101
hdfs://x.x.x.x:8020/Air/SEARCH/BH/INT/20171001/2017100101
hdfs://x.x.x.x:8020/Air/SEARCH/IN/DOM/20171001/2017100101
hdfs://x.x.x.x:8020/Air/SEARCH/IN/INT/20171001/2017100101
Each folder has close to 50 files in it.My intention is to merge all the files within a folder to get a single file while copying it on S3 from HDFS. The issue I am having is with the regex with the groupBy option.I tried this and this does not seem to work :
s3-dist-cp --src hdfs:///Air/ --dest s3a://HadoopSplit/Air-merged/ --groupBy '.*/(\w+)/(\w+)/(\w+)/.*' --outputCodec lzo
The command works per se but i don't get the files within each folder merged into a single file which makes me believe that the issue is with my regex.
回答1:
i figured this out myself only..the correct regex is
.*/Air/(\w+)/(\w+)/(\w+)/.*/.*/.*
and the command to merge and copy is :
s3-dist-cp --src hdfs:///Air/ --dest s3a://HadoopSplit/Air-merged/ --groupBy '.*/Air/(\w+)/(\w+)/(\w+)/.*/.*/.*' --outputCodec lzo
来源：https://stackoverflow.com/questions/46833387/using-groupby-while-copying-from-hdfs-to-s3-to-merge-files-within-a-folder