Examples with CassandraHadoopScanRunner com.thinkaurelius.titan.hadoop.scan.CassandraHadoopScanRunner used on opensource projects

Example 1 with CassandraHadoopScanRunner

use of com.thinkaurelius.titan.hadoop.scan.CassandraHadoopScanRunner in project titan by thinkaurelius.

the class CassandraScanJobIT method testSimpleScan.

@Test
public void testSimpleScan() throws InterruptedException, ExecutionException, IOException, BackendException {
    int keys = 1000;
    int cols = 40;
    String[][] values = KeyValueStoreUtil.generateData(keys, cols);
    //Make it only half the number of columns for every 2nd key
    for (int i = 0; i < values.length; i++) {
        if (i % 2 == 0)
            values[i] = Arrays.copyOf(values[i], cols / 2);
    }
    log.debug("Loading values: " + keys + "x" + cols);
    KeyColumnValueStoreManager mgr = new CassandraThriftStoreManager(GraphDatabaseConfiguration.buildGraphConfiguration());
    KeyColumnValueStore store = mgr.openDatabase("edgestore");
    StoreTransaction tx = mgr.beginTransaction(StandardBaseTransactionConfig.of(TimestampProviders.MICRO));
    KeyColumnValueStoreUtil.loadValues(store, tx, values);
    // noop on Cassandra, but harmless
    tx.commit();
    SimpleScanJobRunner runner = (ScanJob job, Configuration jobConf, String rootNSName) -> {
        try {
            return new CassandraHadoopScanRunner(job).scanJobConf(jobConf).scanJobConfRoot(rootNSName).partitionerOverride("org.apache.cassandra.dht.Murmur3Partitioner").run();
        } catch (ClassNotFoundException e) {
            throw new RuntimeException(e);
        }
    };
    SimpleScanJob.runBasicTests(keys, cols, runner);
}

Example 2 with CassandraHadoopScanRunner

use of com.thinkaurelius.titan.hadoop.scan.CassandraHadoopScanRunner in project titan by thinkaurelius.

the class MapReduceIndexJobs method cassandraRepair.

public static ScanMetrics cassandraRepair(Properties titanProperties, String indexName, String relationType, String partitionerName, Configuration hadoopBaseConf) throws InterruptedException, IOException, ClassNotFoundException {
    IndexRepairJob job = new IndexRepairJob();
    CassandraHadoopScanRunner cr = new CassandraHadoopScanRunner(job);
    ModifiableConfiguration mc = getIndexJobConf(indexName, relationType);
    copyPropertiesToInputAndOutputConf(hadoopBaseConf, titanProperties);
    cr.partitionerOverride(partitionerName);
    cr.scanJobConf(mc);
    cr.scanJobConfRoot(GraphDatabaseConfiguration.class.getName() + "#JOB_NS");
    cr.baseHadoopConf(hadoopBaseConf);
    return cr.run();
}

Example 3 with CassandraHadoopScanRunner

use of com.thinkaurelius.titan.hadoop.scan.CassandraHadoopScanRunner in project titan by thinkaurelius.

the class MapReduceIndexJobs method cassandraRemove.

public static ScanMetrics cassandraRemove(Properties titanProperties, String indexName, String relationType, String partitionerName, Configuration hadoopBaseConf) throws InterruptedException, IOException, ClassNotFoundException {
    IndexRemoveJob job = new IndexRemoveJob();
    CassandraHadoopScanRunner cr = new CassandraHadoopScanRunner(job);
    ModifiableConfiguration mc = getIndexJobConf(indexName, relationType);
    copyPropertiesToInputAndOutputConf(hadoopBaseConf, titanProperties);
    cr.partitionerOverride(partitionerName);
    cr.scanJobConf(mc);
    cr.scanJobConfRoot(GraphDatabaseConfiguration.class.getName() + "#JOB_NS");
    cr.baseHadoopConf(hadoopBaseConf);
    return cr.run();
}