Examples with IMetaStoreClient org.apache.hadoop.hive.metastore.IMetaStoreClient used on opensource projects

Example 21 with IMetaStoreClient

use of org.apache.hadoop.hive.metastore.IMetaStoreClient in project hive by apache.

the class DefaultOutputCommitterContainer method cleanupJob.

@Override
public void cleanupJob(JobContext context) throws IOException {
    getBaseOutputCommitter().cleanupJob(HCatMapRedUtil.createJobContext(context));
    // Cancel HCat and JobTracker tokens
    IMetaStoreClient client = null;
    try {
        HiveConf hiveConf = HCatUtil.getHiveConf(context.getConfiguration());
        client = HCatUtil.getHiveMetastoreClient(hiveConf);
        String tokenStrForm = client.getTokenStrForm();
        if (tokenStrForm != null && context.getConfiguration().get(HCatConstants.HCAT_KEY_TOKEN_SIGNATURE) != null) {
            client.cancelDelegationToken(tokenStrForm);
        }
    } catch (Exception e) {
        LOG.warn("Failed to cancel delegation token", e);
    } finally {
        HCatUtil.closeHiveClientQuietly(client);
    }
}

Example 22 with IMetaStoreClient

use of org.apache.hadoop.hive.metastore.IMetaStoreClient in project hive by apache.

the class FileOutputFormatContainer method checkOutputSpecs.

@Override
public void checkOutputSpecs(JobContext context) throws IOException, InterruptedException {
    OutputJobInfo jobInfo = HCatOutputFormat.getJobInfo(context.getConfiguration());
    IMetaStoreClient client = null;
    try {
        HiveConf hiveConf = HCatUtil.getHiveConf(context.getConfiguration());
        client = HCatUtil.getHiveMetastoreClient(hiveConf);
        handleDuplicatePublish(context, jobInfo, client, new Table(jobInfo.getTableInfo().getTable()));
    } catch (MetaException e) {
        throw new IOException(e);
    } catch (TException e) {
        throw new IOException(e);
    } finally {
        HCatUtil.closeHiveClientQuietly(client);
    }
    if (!jobInfo.isDynamicPartitioningUsed()) {
        JobConf jobConf = new JobConf(context.getConfiguration());
        getBaseOutputFormat().checkOutputSpecs(null, jobConf);
        // checkoutputspecs might've set some properties we need to have context reflect that
        HCatUtil.copyConf(jobConf, context.getConfiguration());
    }
}

Example 23 with IMetaStoreClient

use of org.apache.hadoop.hive.metastore.IMetaStoreClient in project hive by apache.

the class HCatOutputFormat method setOutput.

/**
 * Set the information about the output to write for the job. This queries the metadata server
 * to find the StorageHandler to use for the table.  It throws an error if the
 * partition is already published.
 * @param conf the Configuration object
 * @param credentials the Credentials object
 * @param outputJobInfo the table output information for the job
 * @throws IOException the exception in communicating with the metadata server
 */
@SuppressWarnings("unchecked")
public static void setOutput(Configuration conf, Credentials credentials, OutputJobInfo outputJobInfo) throws IOException {
    IMetaStoreClient client = null;
    try {
        HiveConf hiveConf = HCatUtil.getHiveConf(conf);
        client = HCatUtil.getHiveMetastoreClient(hiveConf);
        Table table = HCatUtil.getTable(client, outputJobInfo.getDatabaseName(), outputJobInfo.getTableName());
        StorageDescriptor sd = table.getTTable().getSd();
        if (sd.isCompressed()) {
            throw new HCatException(ErrorType.ERROR_NOT_SUPPORTED, "Store into a compressed partition from Pig/Mapreduce is not supported");
        }
        if (sd.getBucketCols() != null && !sd.getBucketCols().isEmpty()) {
            throw new HCatException(ErrorType.ERROR_NOT_SUPPORTED, "Store into a partition with bucket definition from Pig/Mapreduce is not supported");
        }
        if (sd.getSortCols() != null && !sd.getSortCols().isEmpty()) {
            throw new HCatException(ErrorType.ERROR_NOT_SUPPORTED, "Store into a partition with sorted column definition from Pig/Mapreduce is not supported");
        }
        if (AcidUtils.isTransactionalTable(table)) {
            throw new HCatException(ErrorType.ERROR_NOT_SUPPORTED, "Store into a transactional table " + table.getFullyQualifiedName() + " from Pig/Mapreduce is not supported");
        }
        // Set up a common id hash for this job, so that when we create any temporary directory
        // later on, it is guaranteed to be unique.
        String idHash;
        if ((idHash = conf.get(HCatConstants.HCAT_OUTPUT_ID_HASH)) == null) {
            idHash = String.valueOf(Math.random());
        }
        conf.set(HCatConstants.HCAT_OUTPUT_ID_HASH, idHash);
        if (table.getTTable().getPartitionKeysSize() == 0) {
            if ((outputJobInfo.getPartitionValues() != null) && (!outputJobInfo.getPartitionValues().isEmpty())) {
                // attempt made to save partition values in non-partitioned table - throw error.
                throw new HCatException(ErrorType.ERROR_INVALID_PARTITION_VALUES, "Partition values specified for non-partitioned table");
            }
            // non-partitioned table
            outputJobInfo.setPartitionValues(new HashMap<String, String>());
        } else {
            // partitioned table, we expect partition values
            // convert user specified map to have lower case key names
            Map<String, String> valueMap = new HashMap<String, String>();
            if (outputJobInfo.getPartitionValues() != null) {
                for (Map.Entry<String, String> entry : outputJobInfo.getPartitionValues().entrySet()) {
                    valueMap.put(entry.getKey().toLowerCase(), entry.getValue());
                }
            }
            if ((outputJobInfo.getPartitionValues() == null) || (outputJobInfo.getPartitionValues().size() < table.getTTable().getPartitionKeysSize())) {
                // dynamic partition usecase - partition values were null, or not all were specified
                // need to figure out which keys are not specified.
                List<String> dynamicPartitioningKeys = new ArrayList<String>();
                boolean firstItem = true;
                for (FieldSchema fs : table.getPartitionKeys()) {
                    if (!valueMap.containsKey(fs.getName().toLowerCase())) {
                        dynamicPartitioningKeys.add(fs.getName().toLowerCase());
                    }
                }
                if (valueMap.size() + dynamicPartitioningKeys.size() != table.getTTable().getPartitionKeysSize()) {
                    // If this isn't equal, then bogus key values have been inserted, error out.
                    throw new HCatException(ErrorType.ERROR_INVALID_PARTITION_VALUES, "Invalid partition keys specified");
                }
                outputJobInfo.setDynamicPartitioningKeys(dynamicPartitioningKeys);
                String dynHash;
                if ((dynHash = conf.get(HCatConstants.HCAT_DYNAMIC_PTN_JOBID)) == null) {
                    dynHash = String.valueOf(Math.random());
                }
                conf.set(HCatConstants.HCAT_DYNAMIC_PTN_JOBID, dynHash);
                // if custom pattern is set in case of dynamic partitioning, configure custom path
                String customPattern = conf.get(HCatConstants.HCAT_DYNAMIC_CUSTOM_PATTERN);
                if (customPattern != null) {
                    HCatFileUtil.setCustomPath(customPattern, outputJobInfo);
                }
            }
            outputJobInfo.setPartitionValues(valueMap);
        }
        // To get around hbase failure on single node, see BUG-4383
        conf.set("dfs.client.read.shortcircuit", "false");
        HCatSchema tableSchema = HCatUtil.extractSchema(table);
        StorerInfo storerInfo = InternalUtil.extractStorerInfo(table.getTTable().getSd(), table.getParameters());
        List<String> partitionCols = new ArrayList<String>();
        for (FieldSchema schema : table.getPartitionKeys()) {
            partitionCols.add(schema.getName());
        }
        HiveStorageHandler storageHandler = HCatUtil.getStorageHandler(conf, storerInfo);
        // Serialize the output info into the configuration
        outputJobInfo.setTableInfo(HCatTableInfo.valueOf(table.getTTable()));
        outputJobInfo.setOutputSchema(tableSchema);
        harRequested = getHarRequested(hiveConf);
        outputJobInfo.setHarRequested(harRequested);
        maxDynamicPartitions = getMaxDynamicPartitions(hiveConf);
        outputJobInfo.setMaximumDynamicPartitions(maxDynamicPartitions);
        HCatUtil.configureOutputStorageHandler(storageHandler, conf, outputJobInfo);
        Path tblPath = new Path(table.getTTable().getSd().getLocation());
        /*  Set the umask in conf such that files/dirs get created with table-dir
      * permissions. Following three assumptions are made:
      * 1. Actual files/dirs creation is done by RecordWriter of underlying
      * output format. It is assumed that they use default permissions while creation.
      * 2. Default Permissions = FsPermission.getDefault() = 777.
      * 3. UMask is honored by underlying filesystem.
      */
        FsPermission.setUMask(conf, FsPermission.getDefault().applyUMask(tblPath.getFileSystem(conf).getFileStatus(tblPath).getPermission()));
        if (Security.getInstance().isSecurityEnabled()) {
            Security.getInstance().handleSecurity(credentials, outputJobInfo, client, conf, harRequested);
        }
    } catch (Exception e) {
        if (e instanceof HCatException) {
            throw (HCatException) e;
        } else {
            throw new HCatException(ErrorType.ERROR_SET_OUTPUT, e);
        }
    } finally {
        HCatUtil.closeHiveClientQuietly(client);
    }
}

Example 24 with IMetaStoreClient

use of org.apache.hadoop.hive.metastore.IMetaStoreClient in project hive by apache.

the class TestHiveClientCache method testMultipleThreadAccess.

/**
 * Check that a *new* client is created if asked from different threads even with
 * the same hive configuration
 * @throws ExecutionException
 * @throws InterruptedException
 */
@Test
public void testMultipleThreadAccess() throws ExecutionException, InterruptedException {
    final HiveClientCache cache = new HiveClientCache(1000);
    class GetHiveClient implements Callable<IMetaStoreClient> {

        @Override
        public IMetaStoreClient call() throws IOException, MetaException, LoginException {
            return cache.get(hiveConf);
        }
    }
    ExecutorService executor = Executors.newFixedThreadPool(2);
    Callable<IMetaStoreClient> worker1 = new GetHiveClient();
    Callable<IMetaStoreClient> worker2 = new GetHiveClient();
    Future<IMetaStoreClient> clientFuture1 = executor.submit(worker1);
    Future<IMetaStoreClient> clientFuture2 = executor.submit(worker2);
    IMetaStoreClient client1 = clientFuture1.get();
    IMetaStoreClient client2 = clientFuture2.get();
    assertNotNull(client1);
    assertNotNull(client2);
    assertNotSame(client1, client2);
}

Example 25 with IMetaStoreClient

use of org.apache.hadoop.hive.metastore.IMetaStoreClient in project hive by apache.

the class SmokeTest method main.

public static void main(String[] args) throws Exception {
    SmokeTest test = new SmokeTest();
    Configuration conf = MetastoreConf.newMetastoreConf();
    IMetaStoreClient client = new HiveMetaStoreClient(conf);
    test.runTest(client);
}