Examples with ModifiableHadoopConfiguration com.thinkaurelius.titan.hadoop.config.ModifiableHadoopConfiguration used on opensource projects

Example 1 with ModifiableHadoopConfiguration

use of com.thinkaurelius.titan.hadoop.config.ModifiableHadoopConfiguration in project titan by thinkaurelius.

the class MapReduceIndexManagement method updateIndex.

/**
     * Updates the provided index according to the given {@link SchemaAction}.
     * Only {@link SchemaAction#REINDEX} and {@link SchemaAction#REMOVE_INDEX} are supported.
     *
     * @param index the index to process
     * @param updateAction either {@code REINDEX} or {@code REMOVE_INDEX}
     * @return a future that returns immediately;
     *         this method blocks until the Hadoop MapReduce job completes
     */
// TODO make this future actually async and update javadoc @return accordingly
public TitanManagement.IndexJobFuture updateIndex(TitanIndex index, SchemaAction updateAction) throws BackendException {
    Preconditions.checkNotNull(index, "Index parameter must not be null", index);
    Preconditions.checkNotNull(updateAction, "%s parameter must not be null", SchemaAction.class.getSimpleName());
    Preconditions.checkArgument(SUPPORTED_ACTIONS.contains(updateAction), "Only these %s parameters are supported: %s (was given %s)", SchemaAction.class.getSimpleName(), SUPPORTED_ACTIONS_STRING, updateAction);
    Preconditions.checkArgument(RelationTypeIndex.class.isAssignableFrom(index.getClass()) || TitanGraphIndex.class.isAssignableFrom(index.getClass()), "Index %s has class %s: must be a %s or %s (or subtype)", index.getClass(), RelationTypeIndex.class.getSimpleName(), TitanGraphIndex.class.getSimpleName());
    org.apache.hadoop.conf.Configuration hadoopConf = new org.apache.hadoop.conf.Configuration();
    ModifiableHadoopConfiguration titanmrConf = ModifiableHadoopConfiguration.of(TitanHadoopConfiguration.MAPRED_NS, hadoopConf);
    // The job we'll execute to either REINDEX or REMOVE_INDEX
    final Class<? extends IndexUpdateJob> indexJobClass;
    final Class<? extends Mapper> mapperClass;
    // The class of the IndexUpdateJob and the Mapper that will be used to run it (VertexScanJob vs ScanJob)
    if (updateAction.equals(SchemaAction.REINDEX)) {
        indexJobClass = IndexRepairJob.class;
        mapperClass = HadoopVertexScanMapper.class;
    } else if (updateAction.equals(SchemaAction.REMOVE_INDEX)) {
        indexJobClass = IndexRemoveJob.class;
        mapperClass = HadoopScanMapper.class;
    } else {
        // Shouldn't get here -- if this exception is ever thrown, update SUPPORTED_ACTIONS
        throw new IllegalStateException("Unrecognized " + SchemaAction.class.getSimpleName() + ": " + updateAction);
    }
    // The column family that serves as input to the IndexUpdateJob
    final String readCF;
    if (RelationTypeIndex.class.isAssignableFrom(index.getClass())) {
        readCF = Backend.EDGESTORE_NAME;
    } else {
        TitanGraphIndex gindex = (TitanGraphIndex) index;
        if (gindex.isMixedIndex() && !updateAction.equals(SchemaAction.REINDEX))
            throw new UnsupportedOperationException("External mixed indexes must be removed in the indexing system directly.");
        Preconditions.checkState(TitanGraphIndex.class.isAssignableFrom(index.getClass()));
        if (updateAction.equals(SchemaAction.REMOVE_INDEX))
            readCF = Backend.INDEXSTORE_NAME;
        else
            readCF = Backend.EDGESTORE_NAME;
    }
    titanmrConf.set(TitanHadoopConfiguration.COLUMN_FAMILY_NAME, readCF);
    // The MapReduce InputFormat class based on the open graph's store manager
    final Class<? extends InputFormat> inputFormat;
    final Class<? extends KeyColumnValueStoreManager> storeManagerClass = graph.getBackend().getStoreManagerClass();
    if (CASSANDRA_STORE_MANAGER_CLASSES.contains(storeManagerClass)) {
        inputFormat = CassandraBinaryInputFormat.class;
        // Set the partitioner
        IPartitioner part = ((AbstractCassandraStoreManager) graph.getBackend().getStoreManager()).getCassandraPartitioner();
        hadoopConf.set("cassandra.input.partitioner.class", part.getClass().getName());
    } else if (HBASE_STORE_MANAGER_CLASSES.contains(storeManagerClass)) {
        inputFormat = HBaseBinaryInputFormat.class;
    } else {
        throw new IllegalArgumentException("Store manager class " + storeManagerClass + "is not supported");
    }
    // The index name and relation type name (if the latter is applicable)
    final String indexName = index.name();
    final String relationTypeName = RelationTypeIndex.class.isAssignableFrom(index.getClass()) ? ((RelationTypeIndex) index).getType().name() : "";
    Preconditions.checkNotNull(indexName);
    // Set the class of the IndexUpdateJob
    titanmrConf.set(TitanHadoopConfiguration.SCAN_JOB_CLASS, indexJobClass.getName());
    // Set the configuration of the IndexUpdateJob
    copyIndexJobKeys(hadoopConf, indexName, relationTypeName);
    titanmrConf.set(TitanHadoopConfiguration.SCAN_JOB_CONFIG_ROOT, GraphDatabaseConfiguration.class.getName() + "#JOB_NS");
    // Copy the StandardTitanGraph configuration under TitanHadoopConfiguration.GRAPH_CONFIG_KEYS
    org.apache.commons.configuration.Configuration localbc = graph.getConfiguration().getLocalConfiguration();
    localbc.clearProperty(Graph.GRAPH);
    copyInputKeys(hadoopConf, localbc);
    String jobName = HadoopScanMapper.class.getSimpleName() + "[" + indexJobClass.getSimpleName() + "]";
    try {
        return new CompletedJobFuture(HadoopScanRunner.runJob(hadoopConf, inputFormat, jobName, mapperClass));
    } catch (Exception e) {
        return new FailedJobFuture(e);
    }
}

Example 2 with ModifiableHadoopConfiguration

use of com.thinkaurelius.titan.hadoop.config.ModifiableHadoopConfiguration in project titan by thinkaurelius.

the class HadoopScanRunner method runJob.

/**
     * Run a ScanJob on Hadoop MapReduce.
     * <p>
     * The {@code confRootField} parameter must be a string in the format
     * {@code package.package...class#fieldname}, where {@code fieldname} is the
     * name of a public static field on the class specified by the portion of the
     * string before the {@code #}.  The {@code #} itself is just a separator and
     * is discarded.
     * <p>
     * When a MapReduce task process prepares to execute the {@code ScanJob}, it will
     * read the public static field named by {@code confFieldRoot} and cast it to a
     * {@link ConfigNamespace}.  This namespace object becomes the root of a
     * {@link Configuration} instantiated, populated with the key-value pairs
     * from the {@code conf} parameter, and then passed into the {@code ScanJob}.
     * <p>
     * This method blocks until the ScanJob completes, then returns the metrics
     * generated by the job during its execution.  It does not timeout.
     *
     * @param conf configuration settings for the ScanJob
     * @param confRootField the root of the ScanJob's configuration
     * @param hadoopConf the Configuration passed to the MapReduce Job
     * @param inputFormat the InputFormat<StaticBuffer, Iterable<Entry>>
     *        that reads (row, columns) pairs out of a Titan edgestore
     * @return metrics generated by the ScanJob
     * @throws IOException if the job fails for any reason
     * @throws ClassNotFoundException if {@code scanJob.getClass()} or if Hadoop
     *         MapReduce's internal job-submission-related reflection fails
     * @throws InterruptedException if interrupted while waiting for the Hadoop
     *         MapReduce job to complete
     */
public static ScanMetrics runJob(Configuration conf, String confRootField, org.apache.hadoop.conf.Configuration hadoopConf, Class<? extends InputFormat> inputFormat, String jobName, Class<? extends Mapper> mapperClass) throws IOException, InterruptedException, ClassNotFoundException {
    Preconditions.checkArgument(null != hadoopConf);
    Preconditions.checkArgument(null != inputFormat);
    if (null != conf) {
        Preconditions.checkArgument(null != confRootField, "Configuration root field must be provided when configuration instance is provided");
    }
    ModifiableHadoopConfiguration scanConf = ModifiableHadoopConfiguration.of(TitanHadoopConfiguration.MAPRED_NS, hadoopConf);
    if (null != confRootField) {
        // Set the scanjob configuration root
        scanConf.set(TitanHadoopConfiguration.SCAN_JOB_CONFIG_ROOT, confRootField);
        // Instantiate scanjob configuration root
        ConfigNamespace confRoot = HadoopScanMapper.getJobRoot(confRootField);
        // Create writable view of scanjob configuration atop the Hadoop Configuration instance, where all keys are prefixed with SCAN_JOB_CONFIG_KEYS
        ModifiableConfiguration hadoopJobConf = ModifiableHadoopConfiguration.prefixView(confRoot, TitanHadoopConfiguration.SCAN_JOB_CONFIG_KEYS, scanConf);
        // Copy scanjob settings from the Titan Configuration instance to the Hadoop Configuration instance
        Map<String, Object> jobConfMap = conf.getSubset(confRoot);
        for (Map.Entry<String, Object> jobConfEntry : jobConfMap.entrySet()) {
            hadoopJobConf.set((ConfigOption) ConfigElement.parse(confRoot, jobConfEntry.getKey()).element, jobConfEntry.getValue());
        }
    }
    return runJob(scanConf.getHadoopConfiguration(), inputFormat, jobName, mapperClass);
}

Example 3 with ModifiableHadoopConfiguration

use of com.thinkaurelius.titan.hadoop.config.ModifiableHadoopConfiguration in project titan by thinkaurelius.

the class HadoopScanRunner method runScanJob.

public static ScanMetrics runScanJob(ScanJob scanJob, Configuration conf, String confRootField, org.apache.hadoop.conf.Configuration hadoopConf, Class<? extends InputFormat> inputFormat) throws IOException, InterruptedException, ClassNotFoundException {
    ModifiableHadoopConfiguration scanConf = ModifiableHadoopConfiguration.of(TitanHadoopConfiguration.MAPRED_NS, hadoopConf);
    tryToLoadClassByName(scanJob);
    // Set the ScanJob class
    scanConf.set(TitanHadoopConfiguration.SCAN_JOB_CLASS, scanJob.getClass().getName());
    String jobName = HadoopScanMapper.class.getSimpleName() + "[" + scanJob + "]";
    return runJob(conf, confRootField, hadoopConf, inputFormat, jobName, HadoopScanMapper.class);
}

Example 4 with ModifiableHadoopConfiguration

use of com.thinkaurelius.titan.hadoop.config.ModifiableHadoopConfiguration in project titan by thinkaurelius.

the class HadoopScanRunner method runVertexScanJob.

public static ScanMetrics runVertexScanJob(VertexScanJob vertexScanJob, Configuration conf, String confRootField, org.apache.hadoop.conf.Configuration hadoopConf, Class<? extends InputFormat> inputFormat) throws IOException, InterruptedException, ClassNotFoundException {
    ModifiableHadoopConfiguration scanConf = ModifiableHadoopConfiguration.of(TitanHadoopConfiguration.MAPRED_NS, hadoopConf);
    tryToLoadClassByName(vertexScanJob);
    // Set the VertexScanJob class
    scanConf.set(TitanHadoopConfiguration.SCAN_JOB_CLASS, vertexScanJob.getClass().getName());
    String jobName = HadoopScanMapper.class.getSimpleName() + "[" + vertexScanJob + "]";
    return runJob(conf, confRootField, hadoopConf, inputFormat, jobName, HadoopVertexScanMapper.class);
}

Example 5 with ModifiableHadoopConfiguration

use of com.thinkaurelius.titan.hadoop.config.ModifiableHadoopConfiguration in project titan by thinkaurelius.

the class HadoopScanMapper method setup.

@Override
protected void setup(Context context) throws IOException, InterruptedException {
    super.setup(context);
    org.apache.hadoop.conf.Configuration hadoopConf = DEFAULT_COMPAT.getContextConfiguration(context);
    ModifiableHadoopConfiguration scanConf = ModifiableHadoopConfiguration.of(TitanHadoopConfiguration.MAPRED_NS, hadoopConf);
    job = getJob(scanConf);
    metrics = new HadoopContextScanMetrics(context);
    Configuration graphConf = getTitanConfiguration(context);
    finishSetup(scanConf, graphConf);
}