D-Net

package eu.dnetlib.data.hadoop.blackboard;

import java.io.IOException;

import java.io.StringWriter;

import java.util.Map;

import java.util.Map.Entry;

import eu.dnetlib.miscutils.functional.xml.XMLIndenter;

import eu.dnetlib.rmi.data.hadoop.ClusterName;

import org.apache.commons.logging.Log;

import org.apache.commons.logging.LogFactory;

import org.apache.hadoop.conf.Configuration;

import org.apache.hadoop.fs.Path;

import org.apache.hadoop.mapred.JobConf;

import org.apache.hadoop.mapred.RunningJob;

import eu.dnetlib.data.hadoop.HadoopJob;

import eu.dnetlib.data.hadoop.mapreduce.MapreduceJobMonitor;

import eu.dnetlib.rmi.data.hadoop.HadoopServiceException;

import eu.dnetlib.data.hadoop.utils.JobProfile;

import eu.dnetlib.data.hadoop.utils.ScanFactory;

import eu.dnetlib.data.hadoop.utils.ScanProperties;

import eu.dnetlib.enabling.tools.blackboard.BlackboardJob;

import org.springframework.beans.factory.annotation.Autowired;

public class SubmitMapreduceJobAction extends AbstractSubmitAction {

	/**

	 * logger.

	 */

	private static final Log log = LogFactory.getLog(SubmitMapreduceJobAction.class); // NOPMD by marko on 11/24/08 5:02 PM

	@Autowired

	private ScanFactory scanFactory;

	@Override

	public void submit(final JobCompletion callback, final BlackboardJob bbJob, final String jobName, final JobProfile jobProfile)

			throws HadoopServiceException {

		final ClusterName clusterName = ClusterName.valueOf(bbJob.getParameters().get("cluster"));

		try {

			final JobConf jobConf = prepareJob(getConf(clusterName), jobName, jobProfile, bbJob.getParameters());

			if (!hadoopClientMap.isMapreduceAvailable(clusterName))

				throw new HadoopServiceException("mapreduce not available for cluster: " + clusterName.toString());

			logJobDetails(jobConf);

			final RunningJob runningJob = hadoopClientMap.getJtClient(clusterName).submitJob(jobConf);

			final String jobId = newJobId(clusterName, runningJob.getID().getId());

			jobRegistry.registerJob(HadoopJob.newInstance(jobId, clusterName, jobProfile,

					new MapreduceJobMonitor(runningJob, callback)));

		} catch (final IOException e) {

			throw new HadoopServiceException("error executing hadoop job: " + jobName, e);

		}

	}

	protected JobConf prepareJob(final Configuration configuration, final String jobName, final JobProfile jobProfile, final Map<String, String> parameters)

			throws IOException, HadoopServiceException {

		log.info("creating job: " + jobName);

		final JobConf jobConf = new JobConf(configuration);

		jobConf.setJobName(jobName);

		jobConf.set("dnet.mapred.job.description", jobProfile.getDescription());

		final String jobLib = getJobLib(configuration, jobProfile);

		jobConf.setJar(new Path(jobLib).toString());

		set(jobConf, jobProfile.getJobDefinition());

		set(jobConf, parameters);

		// if we're reading from hbase table then we need also to set a scanner.

		final ScanProperties scanProperties = jobProfile.getScanProperties();

		if (jobProfile.getRequiredParams().contains("hbase.mapreduce.inputtable") && (scanProperties != null)) {

			jobConf.set("hbase.mapreduce.scan", scanFactory.getScan(scanProperties));

		}

		return jobConf;

	}

	protected String getJobLib(final Configuration configuration, final JobProfile jobProfile) throws HadoopServiceException {

		String jobLib = getDefaultLibPath(configuration.get("fs.defaultFS"));

		if (jobProfile.getJobDefinition().containsKey("job.lib")) {

			jobLib = jobProfile.getJobDefinition().get("job.lib");

		}

		if ((jobLib == null) || jobLib.isEmpty()) throw new HadoopServiceException("job.lib must refer to an absolute or relative HDFS path");

		if (!jobLib.startsWith("hdfs://")) {

			jobLib = configuration.get("fs.defaultFS") + jobLib;

		}

		log.info("using job.lib: " + jobLib);

		return jobLib;

	}

	protected void set(final JobConf jobConf, final Map<String, String> properties) {

		for (final Entry<String, String> e : properties.entrySet()) {

			if (checkHdfsProperty(e)) {

				final String v = jobConf.get("fs.defaultFS") + e.getValue();

				e.setValue(v);

			}

			jobConf.set(e.getKey(), e.getValue());

		}

	}

	protected void logJobDetails(final JobConf jobConf) {

		final StringWriter sw = new StringWriter();

		try {

			jobConf.writeXml(sw);

			if (log.isDebugEnabled()) {

				log.debug("\n" + XMLIndenter.indent(sw.toString()));

			}

		} catch (final IOException e) {

			log.warn("unable to log job details: " + jobConf.getJobName());

		}

	}

}