Hi all,
to increase productivity in our current project (which makes a heavy use
of Hadoop), I wrote a small Eclipse-based GUI application which
basically consists in 2 views:
* a HDFS explorer adapted from Eclipse filesystem explorer example.
For now, it includes the following features:
o classical tree-based browsing interface, with directory
content being detailed in a 3 columns table (file name, file
size, file type)
o refresh button
o delete file or directory (with confirm dialog): select files
in the tree or table and click the "Delete" button
o rename file or directory: simple click on the file in the
table, type the new name and validate
o open file with system editor: select the file in the table
and click "Open" button (works on Windows, not on Linux)
o internal drag & drop
o external drag & drop from the local filesystem to the HDFS
(the opposite doesn't work)
* a MapReduce *very* simple job launcher:
o select the job XML configuration file
o run the job
o kill the job
o visualize map and reduce progress with progress bars
o open a browser on the Hadoop job tracker web interface
Do you see any interest for such a tool in Hadoop (or in a connate
project)? If yes, I see no problem contributing it. However I don't know
what is your current thinking about providing Hadoop with some GUI
functions? Indeed it may overlap with the current Web-based interface.
WDYT?
Fred
