Search in sources :

Example 21 with TableId

use of co.cask.cdap.data2.util.TableId in project cdap by caskdata.

the class HBaseTableFactory method createMetadataTable.

@Override
public MetadataTable createMetadataTable(String tableName) throws IOException {
    TableId tableId = tableUtil.createHTableId(NamespaceId.SYSTEM, tableName);
    HTable hTable = null;
    // If the table descriptor is in the cache, we assume the table exists.
    if (!tableDescriptors.containsKey(tableId)) {
        synchronized (this) {
            if (!tableDescriptors.containsKey(tableId)) {
                try (HBaseDDLExecutor ddlExecutor = ddlExecutorFactory.get()) {
                    ColumnFamilyDescriptorBuilder cfdBuilder = HBaseTableUtil.getColumnFamilyDescriptorBuilder(Bytes.toString(COLUMN_FAMILY), hConf);
                    TableDescriptorBuilder tdBuilder = HBaseTableUtil.getTableDescriptorBuilder(tableId, cConf).addColumnFamily(cfdBuilder.build());
                    ddlExecutor.createTableIfNotExists(tdBuilder.build(), null);
                    hTable = tableUtil.createHTable(hConf, tableId);
                    tableDescriptors.put(tableId, hTable.getTableDescriptor());
                }
            }
        }
    }
    if (hTable == null) {
        hTable = tableUtil.createHTable(hConf, tableId);
    }
    return new HBaseMetadataTable(tableUtil, hTable, COLUMN_FAMILY, cConf.getInt(Constants.MessagingSystem.HBASE_SCAN_CACHE_ROWS), createExceptionHandler(tableId));
}
Also used : TableId(co.cask.cdap.data2.util.TableId) HBaseDDLExecutor(co.cask.cdap.spi.hbase.HBaseDDLExecutor) ColumnFamilyDescriptorBuilder(co.cask.cdap.data2.util.hbase.ColumnFamilyDescriptorBuilder) HTableDescriptorBuilder(co.cask.cdap.data2.util.hbase.HTableDescriptorBuilder) TableDescriptorBuilder(co.cask.cdap.data2.util.hbase.TableDescriptorBuilder) HTable(org.apache.hadoop.hbase.client.HTable)

Example 22 with TableId

use of co.cask.cdap.data2.util.TableId in project cdap by caskdata.

the class HBaseTableFactory method upgradeCoProcessor.

private void upgradeCoProcessor(TableId tableId, Class<? extends Coprocessor> coprocessor) throws IOException {
    try (HBaseDDLExecutor ddlExecutor = ddlExecutorFactory.get()) {
        HTableDescriptor tableDescriptor;
        try (HBaseAdmin admin = new HBaseAdmin(hConf)) {
            // If table doesn't exist, then skip upgrading coprocessor
            if (!tableUtil.tableExists(admin, tableId)) {
                LOG.debug("TMS Table {} was not found. Skip upgrading coprocessor.", tableId);
                return;
            }
            tableDescriptor = tableUtil.getHTableDescriptor(admin, tableId);
        }
        // Get cdap version from the table
        ProjectInfo.Version version = HBaseTableUtil.getVersion(tableDescriptor);
        String hbaseVersion = HBaseTableUtil.getHBaseVersion(tableDescriptor);
        if (hbaseVersion != null && hbaseVersion.equals(HBaseVersion.getVersionString()) && version.compareTo(ProjectInfo.getVersion()) >= 0) {
            // If cdap has version has not changed or is greater, no need to update. Just enable it, in case
            // it has been disabled by the upgrade tool, and return
            LOG.info("Table '{}' has not changed and its version '{}' is same or greater than current CDAP version '{}'." + " The underlying HBase version {} has also not changed.", tableId, version, ProjectInfo.getVersion(), hbaseVersion);
            enableTable(ddlExecutor, tableId);
            return;
        }
        // create a new descriptor for the table update
        HTableDescriptorBuilder newDescriptor = tableUtil.buildHTableDescriptor(tableDescriptor);
        // Remove old coprocessor
        Map<String, HBaseTableUtil.CoprocessorInfo> coprocessorInfo = HBaseTableUtil.getCoprocessorInfo(tableDescriptor);
        for (Map.Entry<String, HBaseTableUtil.CoprocessorInfo> coprocessorEntry : coprocessorInfo.entrySet()) {
            newDescriptor.removeCoprocessor(coprocessorEntry.getValue().getClassName());
        }
        // Add new coprocessor
        CoprocessorDescriptor coprocessorDescriptor = coprocessorManager.getCoprocessorDescriptor(coprocessor, Coprocessor.PRIORITY_USER);
        Path path = coprocessorDescriptor.getPath() == null ? null : new Path(coprocessorDescriptor.getPath());
        newDescriptor.addCoprocessor(coprocessorDescriptor.getClassName(), path, coprocessorDescriptor.getPriority(), coprocessorDescriptor.getProperties());
        // Update CDAP version, table prefix
        HBaseTableUtil.setVersion(newDescriptor);
        HBaseTableUtil.setHBaseVersion(newDescriptor);
        HBaseTableUtil.setTablePrefix(newDescriptor, cConf);
        // Disable auto-splitting
        newDescriptor.setValue(HTableDescriptor.SPLIT_POLICY, cConf.get(Constants.MessagingSystem.TABLE_HBASE_SPLIT_POLICY));
        // Disable Table
        disableTable(ddlExecutor, tableId);
        tableUtil.modifyTable(ddlExecutor, newDescriptor.build());
        LOG.debug("Enabling table '{}'...", tableId);
        enableTable(ddlExecutor, tableId);
    }
    LOG.info("Table '{}' update completed.", tableId);
}
Also used : HBaseDDLExecutor(co.cask.cdap.spi.hbase.HBaseDDLExecutor) Path(org.apache.hadoop.fs.Path) HTableDescriptorBuilder(co.cask.cdap.data2.util.hbase.HTableDescriptorBuilder) HTableDescriptor(org.apache.hadoop.hbase.HTableDescriptor) HBaseAdmin(org.apache.hadoop.hbase.client.HBaseAdmin) ProjectInfo(co.cask.cdap.common.utils.ProjectInfo) CoprocessorDescriptor(co.cask.cdap.spi.hbase.CoprocessorDescriptor) Map(java.util.Map) ConcurrentHashMap(java.util.concurrent.ConcurrentHashMap)

Example 23 with TableId

use of co.cask.cdap.data2.util.TableId in project cdap by caskdata.

the class HBaseTableFactory method createTable.

/**
 * Creates a new instance of {@link HTable} for the given {@link TableId}. If the hbase table doesn't
 * exist, a new one will be created with the given number of splits.
 */
private HTableWithRowKeyDistributor createTable(TableId tableId, int splits, Class<? extends Coprocessor> coprocessor) throws IOException {
    // Lookup the table descriptor from the cache first. If it is there, we assume the HBase table exists
    // Otherwise, attempt to create it.
    HTable hTable = null;
    HTableDescriptor htd = tableDescriptors.get(tableId);
    if (htd == null) {
        synchronized (this) {
            htd = tableDescriptors.get(tableId);
            if (htd == null) {
                boolean tableExists;
                try (HBaseAdmin admin = new HBaseAdmin(hConf)) {
                    tableExists = tableUtil.tableExists(admin, tableId);
                }
                // Create the table if the table doesn't exist
                try (HBaseDDLExecutor ddlExecutor = ddlExecutorFactory.get()) {
                    // If table exists, then skip creating coprocessor etc
                    if (!tableExists) {
                        TableId metadataTableId = tableUtil.createHTableId(NamespaceId.SYSTEM, cConf.get(Constants.MessagingSystem.METADATA_TABLE_NAME));
                        ColumnFamilyDescriptorBuilder cfdBuilder = HBaseTableUtil.getColumnFamilyDescriptorBuilder(Bytes.toString(COLUMN_FAMILY), hConf);
                        TableDescriptorBuilder tdBuilder = HBaseTableUtil.getTableDescriptorBuilder(tableId, cConf).addColumnFamily(cfdBuilder.build()).addProperty(Constants.MessagingSystem.HBASE_MESSAGING_TABLE_PREFIX_NUM_BYTES, Integer.toString(1)).addProperty(Constants.MessagingSystem.KEY_DISTRIBUTOR_BUCKETS_ATTR, Integer.toString(splits)).addProperty(Constants.MessagingSystem.HBASE_METADATA_TABLE_NAMESPACE, metadataTableId.getNamespace()).addProperty(HTableDescriptor.SPLIT_POLICY, cConf.get(Constants.MessagingSystem.TABLE_HBASE_SPLIT_POLICY)).addCoprocessor(coprocessorManager.getCoprocessorDescriptor(coprocessor, Coprocessor.PRIORITY_USER));
                        // Set the key distributor size the same as the initial number of splits,
                        // essentially one bucket per split.
                        byte[][] splitKeys = HBaseTableUtil.getSplitKeys(splits, splits, new RowKeyDistributorByHashPrefix(new OneByteSimpleHash(splits)));
                        ddlExecutor.createTableIfNotExists(tdBuilder.build(), splitKeys);
                        hTable = tableUtil.createHTable(hConf, tableId);
                        htd = hTable.getTableDescriptor();
                        tableDescriptors.put(tableId, htd);
                    } else {
                        hTable = tableUtil.createHTable(hConf, tableId);
                        htd = hTable.getTableDescriptor();
                        tableDescriptors.put(tableId, htd);
                    }
                }
            }
        }
    }
    if (hTable == null) {
        hTable = tableUtil.createHTable(hConf, tableId);
    }
    hTable.setAutoFlushTo(false);
    return new HTableWithRowKeyDistributor(hTable, new RowKeyDistributorByHashPrefix(new OneByteSimpleHash(getKeyDistributorBuckets(tableId, htd))));
}
Also used : HBaseDDLExecutor(co.cask.cdap.spi.hbase.HBaseDDLExecutor) TableId(co.cask.cdap.data2.util.TableId) HBaseAdmin(org.apache.hadoop.hbase.client.HBaseAdmin) RowKeyDistributorByHashPrefix(co.cask.cdap.hbase.wd.RowKeyDistributorByHashPrefix) ColumnFamilyDescriptorBuilder(co.cask.cdap.data2.util.hbase.ColumnFamilyDescriptorBuilder) OneByteSimpleHash(co.cask.cdap.hbase.wd.RowKeyDistributorByHashPrefix.OneByteSimpleHash) HTableDescriptorBuilder(co.cask.cdap.data2.util.hbase.HTableDescriptorBuilder) TableDescriptorBuilder(co.cask.cdap.data2.util.hbase.TableDescriptorBuilder) HTable(org.apache.hadoop.hbase.client.HTable) HTableDescriptor(org.apache.hadoop.hbase.HTableDescriptor)

Example 24 with TableId

use of co.cask.cdap.data2.util.TableId in project cdap by caskdata.

the class LevelDBDatasetMetricsReporter method report.

private void report(Map<TableId, LevelDBTableService.TableStats> datasetStat) throws DatasetManagementException {
    for (Map.Entry<TableId, LevelDBTableService.TableStats> statEntry : datasetStat.entrySet()) {
        String namespace = statEntry.getKey().getNamespace();
        // emit metrics for only user datasets, tables in system namespace are ignored
        if (NamespaceId.SYSTEM.getNamespace().equals(namespace)) {
            continue;
        }
        String tableName = statEntry.getKey().getTableName();
        Collection<DatasetSpecificationSummary> instances = dsFramework.getInstances(new NamespaceId(namespace));
        for (DatasetSpecificationSummary spec : instances) {
            DatasetSpecification specification = dsFramework.getDatasetSpec(new DatasetId(namespace, spec.getName()));
            if (specification.isParent(tableName)) {
                MetricsContext collector = metricsService.getContext(ImmutableMap.of(Constants.Metrics.Tag.NAMESPACE, namespace, Constants.Metrics.Tag.DATASET, spec.getName()));
                int sizeInMb = (int) (statEntry.getValue().getDiskSizeBytes() / BYTES_IN_MB);
                collector.gauge("dataset.size.mb", sizeInMb);
                break;
            }
        }
    }
}
Also used : TableId(co.cask.cdap.data2.util.TableId) MetricsContext(co.cask.cdap.api.metrics.MetricsContext) DatasetSpecification(co.cask.cdap.api.dataset.DatasetSpecification) NamespaceId(co.cask.cdap.proto.id.NamespaceId) DatasetSpecificationSummary(co.cask.cdap.proto.DatasetSpecificationSummary) ImmutableMap(com.google.common.collect.ImmutableMap) Map(java.util.Map) DatasetId(co.cask.cdap.proto.id.DatasetId)

Example 25 with TableId

use of co.cask.cdap.data2.util.TableId in project cdap by caskdata.

the class HBaseQueueTest method testQueueTableNameFormat.

// TODO: CDAP-1177 Should move to QueueTest after making getApplicationName() etc instance methods in a base class
@Test
public void testQueueTableNameFormat() throws Exception {
    QueueName queueName = QueueName.fromFlowlet(NamespaceId.DEFAULT.getEntityName(), "application1", "flow1", "flowlet1", "output1");
    HBaseQueueAdmin hbaseQueueAdmin = (HBaseQueueAdmin) queueAdmin;
    TableId tableId = hbaseQueueAdmin.getDataTableId(queueName);
    Assert.assertEquals(NamespaceId.DEFAULT.getEntityName(), tableId.getNamespace());
    Assert.assertEquals("system." + hbaseQueueAdmin.getType() + ".application1.flow1", tableId.getTableName());
    String tableName = tableUtil.buildHTableDescriptor(tableId).build().getNameAsString();
    Assert.assertEquals("application1", HBaseQueueAdmin.getApplicationName(tableName));
    Assert.assertEquals("flow1", HBaseQueueAdmin.getFlowName(tableName));
    queueName = QueueName.fromFlowlet("testNamespace", "application1", "flow1", "flowlet1", "output1");
    tableId = hbaseQueueAdmin.getDataTableId(queueName);
    Assert.assertEquals(String.format("%s_testNamespace", TABLE_PREFIX), tableId.getNamespace());
    Assert.assertEquals("system." + hbaseQueueAdmin.getType() + ".application1.flow1", tableId.getTableName());
    tableName = tableUtil.buildHTableDescriptor(tableId).build().getNameAsString();
    Assert.assertEquals("application1", HBaseQueueAdmin.getApplicationName(tableName));
    Assert.assertEquals("flow1", HBaseQueueAdmin.getFlowName(tableName));
}
Also used : TableId(co.cask.cdap.data2.util.TableId) QueueName(co.cask.cdap.common.queue.QueueName) Test(org.junit.Test) QueueTest(co.cask.cdap.data2.transaction.queue.QueueTest)

Aggregations

TableId (co.cask.cdap.data2.util.TableId)102 Test (org.junit.Test)49 HTableDescriptor (org.apache.hadoop.hbase.HTableDescriptor)41 HRegion (org.apache.hadoop.hbase.regionserver.HRegion)39 Put (org.apache.hadoop.hbase.client.Put)34 NamespaceId (co.cask.cdap.proto.id.NamespaceId)26 Cell (org.apache.hadoop.hbase.Cell)24 Scan (org.apache.hadoop.hbase.client.Scan)23 HTableDescriptorBuilder (co.cask.cdap.data2.util.hbase.HTableDescriptorBuilder)20 RegionScanner (org.apache.hadoop.hbase.regionserver.RegionScanner)18 HBaseTableUtil (co.cask.cdap.data2.util.hbase.HBaseTableUtil)16 HBaseTableUtilFactory (co.cask.cdap.data2.util.hbase.HBaseTableUtilFactory)15 IOException (java.io.IOException)14 HTable (org.apache.hadoop.hbase.client.HTable)12 HBaseDDLExecutor (co.cask.cdap.spi.hbase.HBaseDDLExecutor)11 HBaseAdmin (org.apache.hadoop.hbase.client.HBaseAdmin)10 Delete (org.apache.hadoop.hbase.client.Delete)9 Result (org.apache.hadoop.hbase.client.Result)9 Path (org.apache.hadoop.fs.Path)8 HColumnDescriptor (org.apache.hadoop.hbase.HColumnDescriptor)8