Jacob Tolar created OOZIE-3548:
----------------------------------
Summary: FS EL action to compute size of glob
Key: OOZIE-3548
URL: https://issues.apache.org/jira/browse/OOZIE-3548
Project: Oozie
Issue Type: Improvement
Reporter: Jacob Tolar
There is an FS el function (dirSize) that computes the size of a single
directory.
It would be helpful if oozie could give the size of a list or glob of
directories (for example, to get the total size of an input dependency with
multiple paths).
I'd like to use this to compute a parallelism value in Pig. Pig sort of
supports this with the bytesPerReducer setting but that settings is applied to
the entire job; I'd like more flexibility.
Unfortunately, correctly supporting both a list of filese (as provided by
oozie) and globs requires copying in this method:
[https://github.com/apache/hadoop/blob/trunk/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapred/FileInputFormat.java#L498-L538]
[~rohini] [~puru] I am happy to submit a patch. Any suggestions before I do so?
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
