You should provide JobTracker address and port through configuration.
On 7/11/08, Larry Compton <[EMAIL PROTECTED]> wrote:
>
> I'm coming up to speed on the Hadoop APIs. I need to be able to invoke a
> job
> from within a Java application (as opposed to running from the command-line
> "hadoop" executable). The JobConf and JobClient appear to support this and
> I've written a test program to configure and run a job. However, the job
> doesn't appear to be submitted to the JobTracker. Here's a code excerpt
> from
> my client...
>
> String rdfInputPath = args[0];
> String outputPath = args[1];
> String uriInputPath = args[2];
> String jarPath = args[3];
>
> JobConf conf = new JobConf(MaterializeMap.class);
> conf.setJobName("materialize");
>
> conf.setOutputKeyClass(Text.class);
> conf.setOutputValueClass(Text.class);
>
> conf.setMapperClass(MaterializeMapper.class);
> conf.setCombinerClass(MaterializeReducer.class);
> conf.setReducerClass(MaterializeReducer.class);
> conf.setJar(jarPath);
>
> DistributedCache.addCacheFile(new Path(uriInputPath).toUri(), conf);
>
> FileInputFormat.setInputPaths(conf, new Path(rdfInputPath));
> FileOutputFormat.setOutputPath(conf, new Path(outputPath));
>
> conf.setInputFormat(TextInputFormat.class);
> conf.setOutputFormat(TextOutputFormat.class);
>
> JobClient.runJob(conf);
>
> It seems like I should be providing a URL to the JobTracker somewhere, but
> I
> can't figure out where to provide the information.
>
> --
> Larry Compton
>
--
朱盛凯
Jash Zhu
复旦大学软件学院
Software School, Fudan University
