@Override
public int run(String[] args) throws Exception {
String instance = args[0];
String zookeepers = args[1];
String user = args[2];
String tokenFile = args[3];
String input = args[4];
String tableName = args[5];
Job job = Job.getInstance(getConf());
job.setJobName(TokenFileWordCount.class.getName());
job.setJarByClass(this.getClass());
job.setInputFormatClass(TextInputFormat.class);
TextInputFormat.setInputPaths(job, input);
job.setMapperClass(MapClass.class);
job.setNumReduceTasks(0);
job.setOutputFormatClass(AccumuloOutputFormat.class);
job.setOutputKeyClass(Text.class);
job.setOutputValueClass(Mutation.class);
// AccumuloInputFormat not used here, but it uses the same functions.
AccumuloOutputFormat.setZooKeeperInstance(
job, ClientConfiguration.loadDefault().withInstance(instance).withZkHosts(zookeepers));
AccumuloOutputFormat.setConnectorInfo(job, user, tokenFile);
AccumuloOutputFormat.setCreateTables(job, true);
AccumuloOutputFormat.setDefaultTableName(job, tableName);
job.waitForCompletion(true);
return 0;
}
@Override
public int run(String[] args) throws Exception {
Job job = new Job(getConf(), this.getClass().getSimpleName());
job.setJarByClass(this.getClass());
if (job.getJar() == null) {
log.error("M/R requires a jar file! Run mvn package.");
return 1;
}
job.setInputFormatClass(AccumuloInputFormat.class);
AccumuloInputFormat.setConnectorInfo(job, args[0], new PasswordToken(args[1]));
AccumuloInputFormat.setInputTableName(job, args[2]);
AccumuloInputFormat.setScanAuthorizations(job, Authorizations.EMPTY);
AccumuloInputFormat.setZooKeeperInstance(job, args[3], args[4]);
job.setMapperClass(SeqMapClass.class);
job.setMapOutputKeyClass(Text.class);
job.setMapOutputValueClass(Mutation.class);
job.setNumReduceTasks(0);
job.setOutputFormatClass(AccumuloOutputFormat.class);
AccumuloOutputFormat.setConnectorInfo(job, args[0], new PasswordToken(args[1]));
AccumuloOutputFormat.setCreateTables(job, true);
AccumuloOutputFormat.setDefaultTableName(job, args[5]);
AccumuloOutputFormat.setZooKeeperInstance(job, args[3], args[4]);
job.waitForCompletion(true);
return job.isSuccessful() ? 0 : 1;
}
/** The run method which sets the configuration and starts the MapReduce job */
public int run(String[] args) throws Exception {
if (USE_MINI_ACCUMULO) {
Connector connector = LocalEnvUtil.getConnector(userPass);
userName = "******";
instanceName = connector.getInstance().getInstanceName();
zookeepers = connector.getInstance().getZooKeepers();
}
// Create and initialize a MapReduce Job
Job job = Job.getInstance(getConf(), "tweetIndexer");
job.setJarByClass(IndexedDocIndexer.class);
// Set the AccumuloInputFormat so the mapper can read from Accumulo
AccumuloInputFormat.setConnectorInfo(job, userName, new PasswordToken(userPass));
AccumuloInputFormat.setInputTableName(job, twitterDataTable);
AccumuloInputFormat.setScanAuthorizations(job, new Authorizations());
ClientConfiguration clientConfig = new ClientConfiguration();
clientConfig.withInstance(instanceName);
clientConfig.withZkHosts(zookeepers);
AccumuloInputFormat.setZooKeeperInstance(job, clientConfig);
AccumuloOutputFormat.setConnectorInfo(job, userName, new PasswordToken(userPass));
AccumuloOutputFormat.setCreateTables(job, createTables);
AccumuloOutputFormat.setDefaultTableName(job, tweetDocIndex);
AccumuloOutputFormat.setZooKeeperInstance(job, clientConfig);
// Set the map and reduce classes
job.setMapperClass(TweetMapper.class);
job.setReducerClass(TweetReducer.class);
// Set the output key and value class for the mapper
job.setMapOutputKeyClass(Text.class);
job.setMapOutputValueClass(Text.class);
// Set the output key and value class for the reducer
job.setOutputKeyClass(Text.class);
job.setOutputValueClass(Mutation.class);
// Set the InputFormat and OutputFormat for the job
job.setInputFormatClass(AccumuloInputFormat.class);
job.setOutputFormatClass(AccumuloOutputFormat.class);
// Run the MapReduce job and return 0 for success, 1 otherwise
return job.waitForCompletion(true) ? 0 : 1;
}
