org.apache.hadoop.hbase.client

Class ClientScanner

java.lang.Object

org.apache.hadoop.hbase.client.AbstractClientScanner

org.apache.hadoop.hbase.client.ClientScanner

All Implemented Interfaces:

Closeable, AutoCloseable, Iterable<Result>, ResultScanner

Direct Known Subclasses:

ClientSmallScanner, ReversedClientScanner

@InterfaceAudience.Private
public class ClientScanner
extends AbstractClientScanner

Implements the scanner interface for the HBase client. If there are multiple regions in a table, this scanner will iterate through them all.

Field Summary

Fields

Modifier and Type	Field and Description
protected LinkedList<Result>	cache
protected int	caching
protected org.apache.hadoop.hbase.client.ScannerCallableWithReplicas	callable
protected RpcRetryingCaller<Result[]>	caller
protected boolean	closed
protected org.apache.hadoop.conf.Configuration	conf
protected HRegionInfo	currentRegion
protected Cell	lastCellLoadedToCache The last cell from a not full Row which is added to cache
protected long	lastNext
protected Result	lastResult
protected long	maxScannerResultSize
protected LinkedList<Result>	partialResults A list of partial results that have been returned from the server.
protected byte[]	partialResultsRow The row for which we are accumulating partial Results (i.e.
protected ExecutorService	pool
protected int	primaryOperationTimeout
protected RpcControllerFactory	rpcControllerFactory
protected Scan	scan
protected boolean	scanMetricsPublished
protected int	scannerTimeout

Fields inherited from class org.apache.hadoop.hbase.client.AbstractClientScanner

scanMetrics

Constructor Summary

Constructors

Constructor and Description
ClientScanner(org.apache.hadoop.conf.Configuration conf, Scan scan, TableName tableName, ClusterConnection connection, RpcRetryingCallerFactory rpcFactory, RpcControllerFactory controllerFactory, ExecutorService pool, int primaryOperationTimeout) Create a new ClientScanner for the specified table Note that the passed Scan's start row maybe changed changed.

Method Summary

Methods

Modifier and Type	Method and Description
protected boolean	checkScanStopRow(byte[] endKey)
void	close() Closes the scanner and releases any resources it has allocated
protected static byte[]	createClosestRowBefore(byte[] row) Create the closest row before the specified row
int	getCacheSize()
protected int	getCaching()
protected org.apache.hadoop.conf.Configuration	getConf()
protected ClusterConnection	getConnection()
protected long	getMaxResultSize()
protected ExecutorService	getPool()
protected int	getPrimaryOperationTimeout()
protected List<Result>	getResultsToAddToCache(Result[] resultsFromServer, boolean heartbeatMessage) This method ensures all of our book keeping regarding partial results is kept up to date.
protected int	getRetries()
protected Scan	getScan()
protected org.apache.hadoop.hbase.client.ScannerCallableWithReplicas	getScannerCallable(byte[] localStartKey, int nbRows)
protected int	getScannerTimeout()
protected TableName	getTable()
protected byte[]	getTableName() Deprecated. As of release 0.96 (HBASE-9508). This will be removed in HBase 2.0.0. Use getTable().
protected long	getTimestamp()
protected void	initializeScannerInConstruction()
protected void	loadCache() Contact the servers to load more Results in the cache.
Result	next() Grab the next row's worth of values.
protected boolean	nextScanner(int nbRows, boolean done)
boolean	renewLease() Allow the client to renew the scanner's lease on the server.
protected void	updateLastCellLoadedToCache(Result result)
protected void	writeScanMetrics() Publish the scan metrics.

Methods inherited from class org.apache.hadoop.hbase.client.AbstractClientScanner

getScanMetrics, initScanMetrics, iterator, next

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

scan

protected Scan scan

closed

protected boolean closed

currentRegion

protected HRegionInfo currentRegion

callable

protected org.apache.hadoop.hbase.client.ScannerCallableWithReplicas callable

cache

protected final LinkedList<Result> cache

partialResults

protected final LinkedList<Result> partialResults

A list of partial results that have been returned from the server. This list should only contain results if this scanner does not have enough partial results to form the complete result.

partialResultsRow

protected byte[] partialResultsRow

The row for which we are accumulating partial Results (i.e. the row of the Results stored inside partialResults). Changes to partialResultsRow and partialResults are kept in sync via the methods addToPartialResults(Result) and clearPartialResults()

lastCellLoadedToCache

protected Cell lastCellLoadedToCache

The last cell from a not full Row which is added to cache

caching

protected final int caching

lastNext

protected long lastNext

lastResult

protected Result lastResult

maxScannerResultSize

protected final long maxScannerResultSize

scannerTimeout

protected final int scannerTimeout

scanMetricsPublished

protected boolean scanMetricsPublished

caller

protected RpcRetryingCaller<Result[]> caller

rpcControllerFactory

protected RpcControllerFactory rpcControllerFactory

conf

protected org.apache.hadoop.conf.Configuration conf

primaryOperationTimeout

protected final int primaryOperationTimeout

pool

protected final ExecutorService pool

Constructor Detail

ClientScanner

public ClientScanner(org.apache.hadoop.conf.Configuration conf,
             Scan scan,
             TableName tableName,
             ClusterConnection connection,
             RpcRetryingCallerFactory rpcFactory,
             RpcControllerFactory controllerFactory,
             ExecutorService pool,
             int primaryOperationTimeout)
              throws IOException

Create a new ClientScanner for the specified table Note that the passed Scan's start row maybe changed changed.

Parameters:

conf - The Configuration to use.

scan - Scan to use in this scanner

tableName - The table that we wish to scan

connection - Connection identifying the cluster

Throws:

IOException

Method Detail

initializeScannerInConstruction

protected void initializeScannerInConstruction()
                                        throws IOException

Throws:

IOException

getConnection

protected ClusterConnection getConnection()

getTableName

@Deprecated
protected byte[] getTableName()

Deprecated. As of release 0.96 (HBASE-9508). This will be removed in HBase 2.0.0. Use getTable().

Returns:

Table name

getTable

protected TableName getTable()

getRetries

protected int getRetries()

getScannerTimeout

protected int getScannerTimeout()

getConf

protected org.apache.hadoop.conf.Configuration getConf()

getScan

protected Scan getScan()

getPool

protected ExecutorService getPool()

getPrimaryOperationTimeout

protected int getPrimaryOperationTimeout()

getCaching

protected int getCaching()

getTimestamp

protected long getTimestamp()

getMaxResultSize

protected long getMaxResultSize()

checkScanStopRow

protected boolean checkScanStopRow(byte[] endKey)

nextScanner

protected boolean nextScanner(int nbRows,
                  boolean done)
                       throws IOException

Throws:

IOException

getScannerCallable

@InterfaceAudience.Private
protected org.apache.hadoop.hbase.client.ScannerCallableWithReplicas getScannerCallable(byte[] localStartKey,
                                                                                                      int nbRows)

writeScanMetrics

protected void writeScanMetrics()

Publish the scan metrics. For now, we use scan.setAttribute to pass the metrics back to the application or TableInputFormat.Later, we could push it to other systems. We don't use metrics framework because it doesn't support multi-instances of the same metrics on the same machine; for scan/map reduce scenarios, we will have multiple scans running at the same time. By default, scan metrics are disabled; if the application wants to collect them, this behavior can be turned on by calling calling Scan.setScanMetricsEnabled(boolean)

This invocation clears the scan metrics. Metrics are aggregated in the Scan instance.

next

public Result next()
            throws IOException

Description copied from interface: ResultScanner

Grab the next row's worth of values. The scanner will return a Result.

Returns:

Result object if there is another row, null if the scanner is exhausted.

Throws:

IOException - e

getCacheSize

public int getCacheSize()

loadCache

protected void loadCache()
                  throws IOException

Contact the servers to load more Results in the cache.

Throws:

IOException

getResultsToAddToCache

protected List<Result> getResultsToAddToCache(Result[] resultsFromServer,
                                  boolean heartbeatMessage)
                                       throws IOException

This method ensures all of our book keeping regarding partial results is kept up to date. This method should be called once we know that the results we received back from the RPC request do not contain errors. We return a list of results that should be added to the cache. In general, this list will contain all NON-partial results from the input array (unless the client has specified that they are okay with receiving partial results)

Parameters:

resultsFromServer - The array of Results returned from the server

heartbeatMessage - Flag indicating whether or not the response received from the server represented a complete response, or a heartbeat message that was sent to keep the client-server connection alive

Returns:

the list of results that should be added to the cache.

Throws:

IOException

close

public void close()

Description copied from interface: ResultScanner

Closes the scanner and releases any resources it has allocated

createClosestRowBefore

protected static byte[] createClosestRowBefore(byte[] row)

Create the closest row before the specified row

Parameters:

row -

Returns:

a new byte array which is the closest front row of the specified one

renewLease

public boolean renewLease()

Description copied from class: AbstractClientScanner

Allow the client to renew the scanner's lease on the server.

Specified by:

renewLease in class AbstractClientScanner

Returns:

true if the lease was successfully renewed, false otherwise.

updateLastCellLoadedToCache

protected void updateLastCellLoadedToCache(Result result)