use of org.apache.accumulo.core.client.AccumuloSecurityException in project hive by apache.
the class HiveAccumuloTableInputFormat method getSplits.
@Override
public InputSplit[] getSplits(JobConf jobConf, int numSplits) throws IOException {
final AccumuloConnectionParameters accumuloParams = new AccumuloConnectionParameters(jobConf);
final Instance instance = accumuloParams.getInstance();
final ColumnMapper columnMapper;
try {
columnMapper = getColumnMapper(jobConf);
} catch (TooManyAccumuloColumnsException e) {
throw new IOException(e);
}
JobContext context = ShimLoader.getHadoopShims().newJobContext(Job.getInstance(jobConf));
Path[] tablePaths = FileInputFormat.getInputPaths(context);
try {
UserGroupInformation ugi = UserGroupInformation.getCurrentUser();
final Connector connector;
// Need to get a Connector so we look up the user's authorizations if not otherwise specified
if (accumuloParams.useSasl() && !ugi.hasKerberosCredentials()) {
// In a YARN/Tez job, don't have the Kerberos credentials anymore, use the delegation token
AuthenticationToken token = ConfiguratorBase.getAuthenticationToken(AccumuloInputFormat.class, jobConf);
// Convert the stub from the configuration back into a normal Token
// More reflection to support 1.6
token = helper.unwrapAuthenticationToken(jobConf, token);
connector = instance.getConnector(accumuloParams.getAccumuloUserName(), token);
} else {
// Still in the local JVM, use the username+password or Kerberos credentials
connector = accumuloParams.getConnector(instance);
}
final List<ColumnMapping> columnMappings = columnMapper.getColumnMappings();
final List<IteratorSetting> iterators = predicateHandler.getIterators(jobConf, columnMapper);
final Collection<Range> ranges = predicateHandler.getRanges(jobConf, columnMapper);
// We don't want that.
if (null != ranges && ranges.isEmpty()) {
return new InputSplit[0];
}
// Set the relevant information in the Configuration for the AccumuloInputFormat
configure(jobConf, instance, connector, accumuloParams, columnMapper, iterators, ranges);
int numColumns = columnMappings.size();
List<Integer> readColIds = ColumnProjectionUtils.getReadColumnIDs(jobConf);
// Sanity check
if (numColumns < readColIds.size())
throw new IOException("Number of column mappings (" + numColumns + ")" + " numbers less than the hive table columns. (" + readColIds.size() + ")");
// get splits from Accumulo
InputSplit[] splits = accumuloInputFormat.getSplits(jobConf, numSplits);
HiveAccumuloSplit[] hiveSplits = new HiveAccumuloSplit[splits.length];
for (int i = 0; i < splits.length; i++) {
RangeInputSplit ris = (RangeInputSplit) splits[i];
hiveSplits[i] = new HiveAccumuloSplit(ris, tablePaths[0]);
}
return hiveSplits;
} catch (AccumuloException e) {
log.error("Could not configure AccumuloInputFormat", e);
throw new IOException(StringUtils.stringifyException(e));
} catch (AccumuloSecurityException e) {
log.error("Could not configure AccumuloInputFormat", e);
throw new IOException(StringUtils.stringifyException(e));
} catch (SerDeException e) {
log.error("Could not configure AccumuloInputFormat", e);
throw new IOException(StringUtils.stringifyException(e));
}
}
use of org.apache.accumulo.core.client.AccumuloSecurityException in project hive by apache.
the class HiveAccumuloTableOutputFormat method configureAccumuloOutputFormat.
protected void configureAccumuloOutputFormat(JobConf job) throws IOException {
AccumuloConnectionParameters cnxnParams = getConnectionParams(job);
final String tableName = job.get(AccumuloSerDeParameters.TABLE_NAME);
// Make sure we actually go the table name
Preconditions.checkNotNull(tableName, "Expected Accumulo table name to be provided in job configuration");
// Set the necessary Accumulo information
try {
if (cnxnParams.useMockInstance()) {
setMockInstanceWithErrorChecking(job, cnxnParams.getAccumuloInstanceName());
} else {
// Accumulo instance name with ZK quorum
setZooKeeperInstanceWithErrorChecking(job, cnxnParams.getAccumuloInstanceName(), cnxnParams.getZooKeepers(), cnxnParams.useSasl());
}
// The AccumuloOutputFormat will look for it there.
if (cnxnParams.useSasl()) {
UserGroupInformation ugi = getCurrentUser();
if (!hasKerberosCredentials(ugi)) {
getHelper().addTokenFromUserToJobConf(ugi, job);
} else {
// Still in the local JVM, can use Kerberos credentials
try {
Connector connector = cnxnParams.getConnector();
AuthenticationToken token = getHelper().getDelegationToken(connector);
// Send the DelegationToken down to the Configuration for Accumulo to use
setConnectorInfoWithErrorChecking(job, cnxnParams.getAccumuloUserName(), token);
// Convert the Accumulo token in a Hadoop token
Token<? extends TokenIdentifier> accumuloToken = getHelper().getHadoopToken(token);
log.info("Adding Hadoop Token for Accumulo to Job's Credentials");
// Add the Hadoop token to the JobConf
getHelper().mergeTokenIntoJobConf(job, accumuloToken);
// Make sure the UGI contains the token too for good measure
if (!ugi.addToken(accumuloToken)) {
throw new IOException("Failed to add Accumulo Token to UGI");
}
} catch (AccumuloException | AccumuloSecurityException e) {
throw new IOException("Failed to acquire Accumulo DelegationToken", e);
}
}
} else {
setConnectorInfoWithErrorChecking(job, cnxnParams.getAccumuloUserName(), new PasswordToken(cnxnParams.getAccumuloPassword()));
}
// Set the table where we're writing this data
setDefaultAccumuloTableName(job, tableName);
} catch (AccumuloSecurityException e) {
log.error("Could not connect to Accumulo with provided credentials", e);
throw new IOException(e);
}
}
use of org.apache.accumulo.core.client.AccumuloSecurityException in project YCSB by brianfrankcooper.
the class AccumuloClient method init.
@Override
public void init() throws DBException {
colFam = new Text(getProperties().getProperty("accumulo.columnFamily"));
inst = new ZooKeeperInstance(getProperties().getProperty("accumulo.instanceName"), getProperties().getProperty("accumulo.zooKeepers"));
try {
String principal = getProperties().getProperty("accumulo.username");
AuthenticationToken token = new PasswordToken(getProperties().getProperty("accumulo.password"));
connector = inst.getConnector(principal, token);
} catch (AccumuloException e) {
throw new DBException(e);
} catch (AccumuloSecurityException e) {
throw new DBException(e);
}
if (!(getProperties().getProperty("accumulo.pcFlag", "none").equals("none"))) {
System.err.println("Sorry, the ZK based producer/consumer implementation has been removed. " + "Please see YCSB issue #416 for work on adding a general solution to coordinated work.");
}
}
use of org.apache.accumulo.core.client.AccumuloSecurityException in project Gaffer by gchq.
the class IngestUtils method createSplitsFile.
/**
* Get the existing splits from a table in Accumulo and write a splits file.
* The number of splits is returned.
*
* @param conn - An existing connection to an Accumulo instance
* @param table - The table name
* @param fs - The FileSystem in which to create the splits file
* @param splitsFile - A Path for the output splits file
* @param maxSplits - The maximum number of splits
* @return The number of splits in the table
* @throws IOException for any IO issues reading from the file system. Other accumulo exceptions are caught and wrapped in an IOException.
*/
public static int createSplitsFile(final Connector conn, final String table, final FileSystem fs, final Path splitsFile, final int maxSplits) throws IOException {
LOGGER.info("Creating splits file in location {} from table {} with maximum splits {}", splitsFile, table, maxSplits);
// Get the splits from the table
Collection<Text> splits;
try {
splits = conn.tableOperations().listSplits(table, maxSplits);
} catch (TableNotFoundException | AccumuloSecurityException | AccumuloException e) {
throw new IOException(e.getMessage(), e);
}
// This should have returned at most maxSplits splits, but this is not implemented properly in MockInstance.
if (splits.size() > maxSplits) {
if (conn instanceof MockConnector) {
LOGGER.info("Manually reducing the number of splits to {} due to MockInstance not implementing" + " listSplits(table, maxSplits) properly", maxSplits);
} else {
LOGGER.info("Manually reducing the number of splits to {} (number of splits was {})", maxSplits, splits.size());
}
final Collection<Text> filteredSplits = new TreeSet<>();
final int outputEveryNth = splits.size() / maxSplits;
LOGGER.info("Outputting every {}-th split from {} total", outputEveryNth, splits.size());
int i = 0;
for (final Text text : splits) {
if (i % outputEveryNth == 0) {
filteredSplits.add(text);
}
i++;
if (filteredSplits.size() >= maxSplits) {
break;
}
}
splits = filteredSplits;
}
LOGGER.info("Found {} splits from table {}", splits.size(), table);
try (final PrintStream out = new PrintStream(new BufferedOutputStream(fs.create(splitsFile, true)), false, CommonConstants.UTF_8)) {
// Write the splits to file
if (splits.isEmpty()) {
out.close();
return 0;
}
for (final Text split : splits) {
out.println(new String(Base64.encodeBase64(split.getBytes()), CommonConstants.UTF_8));
}
}
return splits.size();
}
use of org.apache.accumulo.core.client.AccumuloSecurityException in project Gaffer by gchq.
the class TableUtils method setLocalityGroups.
public static void setLocalityGroups(final AccumuloStore store) throws StoreException {
final String tableName = store.getProperties().getTable();
Map<String, Set<Text>> localityGroups = new HashMap<>();
for (final String group : store.getSchema().getGroups()) {
HashSet<Text> localityGroup = new HashSet<>();
localityGroup.add(new Text(group));
localityGroups.put(group, localityGroup);
}
LOGGER.info("Setting locality groups on table {}", tableName);
try {
store.getConnection().tableOperations().setLocalityGroups(tableName, localityGroups);
} catch (AccumuloException | AccumuloSecurityException | TableNotFoundException e) {
throw new StoreException(e.getMessage(), e);
}
}
Aggregations