Hadoop 0.20.2 Eclipse plugin not fully functioning - can't 'Run on Hadoop'

Question

I've just finished installing Hadoop 0.20.2 under Cygwin on Windows 7 with Eclipse Helios (3.6). Hadoop is now fully started, and I'm trying to run a test application within a newly created MapReduce test project in Eclipse. I'm using the Hadoop 0.20.2 plugin from the Hadoop download.

The Map/Reduce Location perspective operates correctly, as does DFS Locations tree in the Package Explorer. However, when I right-click the driver, select 'Run As' > 'Run on Hadoop', nothing happens and no errors spawn on the Console (silent fail :(). I believe a dialog window should appear asking for config before it runs, but this is not happening.

There seems to be a few others with the same problem, but I've yet to find an answer that works. I've tried the 0.20.1 plugin (total fail). The following bug report seems to describe my issue, though I'm a bit of a newbie to all this, so could do with a hand / voice of experience to help out: https://issues.apache.org/jira/browse/MAPREDUCE-1280

Fluoroscope

Answer 1

The hadoop eclipse plugin bundled with the hadoop distribution is compatible with eclipse up to version 3.3. The JIRA-ticket MAPREDUCE-1280 contains a patch for running the plugin in eclipse 3.4 and upwards.

I just compiled the patched plugin with the fixes from the JIRA-ticket MAPREDUCE-1280. The file is attached to the ticket. You can find it here.

Simply remove the old plugin from your eclipse-installation and put the new version of the plugin into the dropins-folder of your eclipse-installation.

After upgrading from an older version of the plugin you will have to start eclipse with the "-clean" command line switch. Help on eclipse command line switches can be found here.

Traumatize

It worked for me (the -clean command-line switch was crucial). Thanks! – Enjoyable

It worked for me too, though I didn't need the -clean switch. Thanks for salvaging a weekend of frustration! – Carol

Answer 2

I don't know whether the plugin has been updated or not, but as far as I know, that one is out of maintenance for several previous releases.

One of the solution is that you should download the source code and try to re-compile the jar file for that plugin (for the latest version of eclipse), however I didn't try it so don't know whether it is working.

Maybe you can try to use Karmasphere.

Gunsmith

Answer 3

askswOrder is correct that the Eclipse plugin has not seen much attention for quite a long time. The JIRA you reference does provide a fix, but it's only been applied to Hadoop 20.3 and above. One option would be to try to apply the patch to 20.2 and recompile, but that's asking quite a lot from a newbie. I'd second the suggestion to use Karmasphere; it's a great product for working with MapReduce and those gents have taken on the work of staying current with Hadoop releases.

Recreant