You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@accumulo.apache.org by el...@apache.org on 2015/04/17 23:13:44 UTC
[2/2] accumulo git commit: ACCUMULO-3602 Added support for grouping
ranges per tablet when using AccumuloInputFormat
ACCUMULO-3602 Added support for grouping ranges per tablet when using AccumuloInputFormat
Introduces a new option to AccumuloInputFormat: batchScan. Comes with
a BatchInputSplit implementation which holds many Ranges against
the same Tablet to reduce the amount of overhead in InputSplits
generated by the InputFormat.
Project: http://git-wip-us.apache.org/repos/asf/accumulo/repo
Commit: http://git-wip-us.apache.org/repos/asf/accumulo/commit/956a50ec
Tree: http://git-wip-us.apache.org/repos/asf/accumulo/tree/956a50ec
Diff: http://git-wip-us.apache.org/repos/asf/accumulo/diff/956a50ec
Branch: refs/heads/master
Commit: 956a50ecb358ab2df3a705c05f3fa8c4ae8846a8
Parents: 315530f
Author: Eugene Cheipesh <ec...@gmail.com>
Authored: Fri Apr 17 15:42:43 2015 -0400
Committer: Josh Elser <el...@apache.org>
Committed: Fri Apr 17 16:26:55 2015 -0400
----------------------------------------------------------------------
core/src/main/findbugs/exclude-filter.xml | 1 +
.../core/client/mapred/AbstractInputFormat.java | 263 +++++++----
.../core/client/mapred/AccumuloInputFormat.java | 8 +-
.../core/client/mapred/InputFormatBase.java | 60 ++-
.../client/mapred/impl/BatchInputSplit.java | 42 ++
.../client/mapreduce/AbstractInputFormat.java | 249 +++++++----
.../client/mapreduce/AccumuloInputFormat.java | 12 +-
.../AccumuloMultiTableInputFormat.java | 12 +-
.../core/client/mapreduce/InputFormatBase.java | 65 ++-
.../core/client/mapreduce/RangeInputSplit.java | 424 ++----------------
.../mapreduce/impl/AccumuloInputSplit.java | 445 +++++++++++++++++++
.../client/mapreduce/impl/BatchInputSplit.java | 152 +++++++
.../mapreduce/lib/impl/InputConfigurator.java | 36 +-
.../mapreduce/AccumuloInputFormatTest.java | 30 +-
.../mapreduce/impl/BatchInputSplitTest.java | 122 +++++
.../test/functional/AccumuloInputFormatIT.java | 49 +-
16 files changed, 1348 insertions(+), 622 deletions(-)
----------------------------------------------------------------------
http://git-wip-us.apache.org/repos/asf/accumulo/blob/956a50ec/core/src/main/findbugs/exclude-filter.xml
----------------------------------------------------------------------
diff --git a/core/src/main/findbugs/exclude-filter.xml b/core/src/main/findbugs/exclude-filter.xml
index 0f66e06..3cc5e30 100644
--- a/core/src/main/findbugs/exclude-filter.xml
+++ b/core/src/main/findbugs/exclude-filter.xml
@@ -49,6 +49,7 @@
<Package name="org.apache.accumulo.core.iterators" />
<Package name="org.apache.accumulo.core.trace" />
<Class name="org.apache.accumulo.core.client.mapred.RangeInputSplit" />
+ <Class name="org.apache.accumulo.core.client.mapred.impl.BatchInputSplit" />
<Class name="org.apache.accumulo.core.util.AddressUtil" />
<Class name="org.apache.accumulo.core.zookeeper.ZooUtil" />
</Or>
http://git-wip-us.apache.org/repos/asf/accumulo/blob/956a50ec/core/src/main/java/org/apache/accumulo/core/client/mapred/AbstractInputFormat.java
----------------------------------------------------------------------
diff --git a/core/src/main/java/org/apache/accumulo/core/client/mapred/AbstractInputFormat.java b/core/src/main/java/org/apache/accumulo/core/client/mapred/AbstractInputFormat.java
index 1071683..df317e3 100644
--- a/core/src/main/java/org/apache/accumulo/core/client/mapred/AbstractInputFormat.java
+++ b/core/src/main/java/org/apache/accumulo/core/client/mapred/AbstractInputFormat.java
@@ -34,18 +34,22 @@ import org.apache.accumulo.core.client.ClientSideIteratorScanner;
import org.apache.accumulo.core.client.Connector;
import org.apache.accumulo.core.client.Instance;
import org.apache.accumulo.core.client.IsolatedScanner;
-import org.apache.accumulo.core.client.IteratorSetting;
import org.apache.accumulo.core.client.Scanner;
+import org.apache.accumulo.core.client.BatchScanner;
+import org.apache.accumulo.core.client.ScannerBase;
import org.apache.accumulo.core.client.TableDeletedException;
import org.apache.accumulo.core.client.TableNotFoundException;
import org.apache.accumulo.core.client.TableOfflineException;
+import org.apache.accumulo.core.client.IteratorSetting;
import org.apache.accumulo.core.client.admin.DelegationTokenConfig;
import org.apache.accumulo.core.client.impl.ClientContext;
import org.apache.accumulo.core.client.impl.OfflineScanner;
import org.apache.accumulo.core.client.impl.ScannerImpl;
import org.apache.accumulo.core.client.impl.Tables;
import org.apache.accumulo.core.client.impl.TabletLocator;
+import org.apache.accumulo.core.client.mapred.impl.BatchInputSplit;
import org.apache.accumulo.core.client.mapreduce.InputTableConfig;
+import org.apache.accumulo.core.client.mapreduce.impl.AccumuloInputSplit;
import org.apache.accumulo.core.client.mapreduce.lib.impl.ConfiguratorBase;
import org.apache.accumulo.core.client.mapreduce.lib.impl.InputConfigurator;
import org.apache.accumulo.core.client.mock.MockInstance;
@@ -384,7 +388,48 @@ public abstract class AbstractInputFormat<K,V> implements InputFormat<K,V> {
protected abstract static class AbstractRecordReader<K,V> implements RecordReader<K,V> {
protected long numKeysRead;
protected Iterator<Map.Entry<Key,Value>> scannerIterator;
- protected RangeInputSplit split;
+ protected org.apache.accumulo.core.client.mapreduce.impl.AccumuloInputSplit split;
+ protected ScannerBase scannerBase;
+
+
+ /**
+ * Extracts Iterators settings from the context to be used by RecordReader.
+ *
+ * @param job
+ * the Hadoop job configuration
+ * @param tableName
+ * the table name for which the scanner is configured
+ * @return List of iterator settings for given table
+ * @since 1.7.0
+ */
+ protected abstract List<IteratorSetting> jobIterators(JobConf job, String tableName);
+
+ /**
+ * Configures the iterators on a scanner for the given table name.
+ *
+ * @param job
+ * the Hadoop job configuration
+ * @param scanner
+ * the scanner for which to configure the iterators
+ * @param tableName
+ * the table name for which the scanner is configured
+ * @since 1.7.0
+ */
+ private void setupIterators(JobConf job, ScannerBase scanner, String tableName, AccumuloInputSplit split) {
+ List<IteratorSetting> iterators = null;
+
+ if (null == split) {
+ iterators = jobIterators(job, tableName);
+ } else {
+ iterators = split.getIterators();
+ if (null == iterators) {
+ iterators = jobIterators(job, tableName);
+ }
+ }
+
+ for (IteratorSetting iterator : iterators)
+ scanner.addScanIterator(iterator);
+ }
/**
* Configures the iterators on a scanner for the given table name.
@@ -396,16 +441,19 @@ public abstract class AbstractInputFormat<K,V> implements InputFormat<K,V> {
* @param tableName
* the table name for which the scanner is configured
* @since 1.6.0
+ * @deprecated since 1.7.0; Use {@link #jobIterators} instead.
*/
- protected abstract void setupIterators(JobConf job, Scanner scanner, String tableName, RangeInputSplit split);
+ @Deprecated
+ protected void setupIterators(JobConf job, Scanner scanner, String tableName, RangeInputSplit split) {
+ setupIterators(job, (ScannerBase) scanner, tableName, (AccumuloInputSplit) split);
+ }
/**
* Initialize a scanner over the given input split using this task attempt configuration.
*/
public void initialize(InputSplit inSplit, JobConf job) throws IOException {
- Scanner scanner;
- split = (RangeInputSplit) inSplit;
- log.debug("Initializing input split: " + split.getRange());
+ split = (AccumuloInputSplit) inSplit;
+ log.debug("Initializing input split: " + split.toString());
Instance instance = split.getInstance(getClientConfiguration(job));
if (null == instance) {
@@ -433,24 +481,74 @@ public abstract class AbstractInputFormat<K,V> implements InputFormat<K,V> {
// but the scanner will use the table id resolved at job setup time
InputTableConfig tableConfig = getInputTableConfig(job, split.getTableName());
- Boolean isOffline = split.isOffline();
- if (null == isOffline) {
- isOffline = tableConfig.isOfflineScan();
- }
+ log.debug("Creating connector with user: " + principal);
+ log.debug("Creating scanner for table: " + table);
+ log.debug("Authorizations are: " + authorizations);
- Boolean isIsolated = split.isIsolatedScan();
- if (null == isIsolated) {
- isIsolated = tableConfig.shouldUseIsolatedScanners();
- }
+ if (split instanceof org.apache.accumulo.core.client.mapreduce.RangeInputSplit) {
+ org.apache.accumulo.core.client.mapreduce.RangeInputSplit rangeSplit = (org.apache.accumulo.core.client.mapreduce.RangeInputSplit) split;
- Boolean usesLocalIterators = split.usesLocalIterators();
- if (null == usesLocalIterators) {
- usesLocalIterators = tableConfig.shouldUseLocalIterators();
- }
+ Boolean isOffline = rangeSplit.isOffline();
+ if (null == isOffline) {
+ isOffline = tableConfig.isOfflineScan();
+ }
+
+ Boolean isIsolated = rangeSplit.isIsolatedScan();
+ if (null == isIsolated) {
+ isIsolated = tableConfig.shouldUseIsolatedScanners();
+ }
- List<IteratorSetting> iterators = split.getIterators();
- if (null == iterators) {
- iterators = tableConfig.getIterators();
+ Boolean usesLocalIterators = rangeSplit.usesLocalIterators();
+ if (null == usesLocalIterators) {
+ usesLocalIterators = tableConfig.shouldUseLocalIterators();
+ }
+
+ Scanner scanner;
+
+ try {
+ if (isOffline) {
+ scanner = new OfflineScanner(instance, new Credentials(principal, token), split.getTableId(), authorizations);
+ } else if (instance instanceof MockInstance) {
+ scanner = instance.getConnector(principal, token).createScanner(split.getTableName(), authorizations);
+ } else {
+ ClientConfiguration clientConf = getClientConfiguration(job);
+ ClientContext context = new ClientContext(instance, new Credentials(principal, token), clientConf);
+ scanner = new ScannerImpl(context, split.getTableId(), authorizations);
+ }
+ if (isIsolated) {
+ log.info("Creating isolated scanner");
+ scanner = new IsolatedScanner(scanner);
+ }
+ if (usesLocalIterators) {
+ log.info("Using local iterators");
+ scanner = new ClientSideIteratorScanner(scanner);
+ }
+ setupIterators(job, scanner, split.getTableName(), split);
+ } catch (Exception e) {
+ throw new IOException(e);
+ }
+
+ scanner.setRange(rangeSplit.getRange());
+ scannerBase = scanner;
+
+ } else if (split instanceof BatchInputSplit) {
+ BatchScanner scanner;
+ BatchInputSplit multiRangeSplit = (BatchInputSplit) split;
+
+ try{
+ // Note: BatchScanner will use at most one thread per tablet, currently BatchInputSplit will not span tablets
+ int scanThreads = 1;
+ scanner = instance.getConnector(principal, token).createBatchScanner(split.getTableName(), authorizations, scanThreads);
+ setupIterators(job, scanner, split.getTableName(), split);
+ } catch (Exception e) {
+ throw new IOException(e);
+ }
+
+ scanner.setRanges(multiRangeSplit.getRanges());
+ scannerBase = scanner;
+
+ } else {
+ throw new IllegalArgumentException("Can not initialize from " + split.getClass().toString());
}
Collection<Pair<Text,Text>> columns = split.getFetchedColumns();
@@ -458,53 +556,27 @@ public abstract class AbstractInputFormat<K,V> implements InputFormat<K,V> {
columns = tableConfig.getFetchedColumns();
}
- try {
- log.debug("Creating connector with user: " + principal);
- log.debug("Creating scanner for table: " + table);
- log.debug("Authorizations are: " + authorizations);
- if (isOffline) {
- scanner = new OfflineScanner(instance, new Credentials(principal, token), split.getTableId(), authorizations);
- } else if (instance instanceof MockInstance) {
- scanner = instance.getConnector(principal, token).createScanner(split.getTableName(), authorizations);
- } else {
- ClientConfiguration clientConf = getClientConfiguration(job);
- ClientContext context = new ClientContext(instance, new Credentials(principal, token), clientConf);
- scanner = new ScannerImpl(context, split.getTableId(), authorizations);
- }
- if (isIsolated) {
- log.info("Creating isolated scanner");
- scanner = new IsolatedScanner(scanner);
- }
- if (usesLocalIterators) {
- log.info("Using local iterators");
- scanner = new ClientSideIteratorScanner(scanner);
- }
- setupIterators(job, scanner, split.getTableName(), split);
- } catch (Exception e) {
- throw new IOException(e);
- }
-
// setup a scanner within the bounds of this split
for (Pair<Text,Text> c : columns) {
if (c.getSecond() != null) {
log.debug("Fetching column " + c.getFirst() + ":" + c.getSecond());
- scanner.fetchColumn(c.getFirst(), c.getSecond());
+ scannerBase.fetchColumn(c.getFirst(), c.getSecond());
} else {
log.debug("Fetching column family " + c.getFirst());
- scanner.fetchColumnFamily(c.getFirst());
+ scannerBase.fetchColumnFamily(c.getFirst());
}
}
- scanner.setRange(split.getRange());
-
+ scannerIterator = scannerBase.iterator();
numKeysRead = 0;
-
- // do this last after setting all scanner options
- scannerIterator = scanner.iterator();
}
@Override
- public void close() {}
+ public void close() {
+ if (null != scannerBase) {
+ scannerBase.close();
+ }
+ }
@Override
public long getPos() throws IOException {
@@ -531,6 +603,7 @@ public abstract class AbstractInputFormat<K,V> implements InputFormat<K,V> {
return InputConfigurator.binOffline(tableId, ranges, instance, conn);
}
+
/**
* Gets the splits of the tables that have been set on the job by reading the metadata table for the specified ranges.
*
@@ -552,26 +625,32 @@ public abstract class AbstractInputFormat<K,V> implements InputFormat<K,V> {
InputTableConfig tableConfig = tableConfigEntry.getValue();
Instance instance = getInstance(job);
- boolean mockInstance;
String tableId;
// resolve table name to id once, and use id from this point forward
if (instance instanceof MockInstance) {
tableId = "";
- mockInstance = true;
} else {
try {
tableId = Tables.getTableId(instance, tableName);
} catch (TableNotFoundException e) {
throw new IOException(e);
}
- mockInstance = false;
}
Authorizations auths = getScanAuthorizations(job);
String principal = getPrincipal(job);
AuthenticationToken token = getAuthenticationToken(job);
+ boolean batchScan = InputConfigurator.isBatchScan(CLASS, job);
+ boolean supportBatchScan =
+ !(tableConfig.isOfflineScan() || tableConfig.shouldUseIsolatedScanners() || tableConfig.shouldUseLocalIterators());
+ if (batchScan && !supportBatchScan)
+ throw new IllegalArgumentException("BatchScanner optimization not available for offline scan, isolated, or local iterators");
+
boolean autoAdjust = tableConfig.shouldAutoAdjustRanges();
+ if (batchScan && !autoAdjust)
+ throw new IllegalArgumentException("AutoAdjustRanges must be enabled when using BatchScanner optimization");
+
List<Range> ranges = autoAdjust ? Range.mergeOverlapping(tableConfig.getRanges()) : tableConfig.getRanges();
if (ranges.isEmpty()) {
ranges = new ArrayList<Range>(1);
@@ -629,32 +708,36 @@ public abstract class AbstractInputFormat<K,V> implements InputFormat<K,V> {
}
for (Map.Entry<KeyExtent,List<Range>> extentRanges : tserverBin.getValue().entrySet()) {
Range ke = extentRanges.getKey().toDataRange();
- for (Range r : extentRanges.getValue()) {
- if (autoAdjust) {
- // divide ranges into smaller ranges, based on the tablets
- RangeInputSplit split = new RangeInputSplit(tableName, tableId, ke.clip(r), new String[] {location});
-
- split.setOffline(tableConfig.isOfflineScan());
- split.setIsolatedScan(tableConfig.shouldUseIsolatedScanners());
- split.setUsesLocalIterators(tableConfig.shouldUseLocalIterators());
- split.setMockInstance(mockInstance);
- split.setFetchedColumns(tableConfig.getFetchedColumns());
- split.setPrincipal(principal);
- split.setToken(token);
- split.setInstanceName(instance.getInstanceName());
- split.setZooKeepers(instance.getZooKeepers());
- split.setAuths(auths);
- split.setIterators(tableConfig.getIterators());
- split.setLogLevel(logLevel);
-
- splits.add(split);
- } else {
- // don't divide ranges
- ArrayList<String> locations = splitsToAdd.get(r);
- if (locations == null)
- locations = new ArrayList<String>(1);
- locations.add(location);
- splitsToAdd.put(r, locations);
+ if (batchScan) {
+ // group ranges by tablet to be read by a BatchScanner
+ ArrayList<Range> clippedRanges = new ArrayList<Range>();
+ for(Range r: extentRanges.getValue())
+ clippedRanges.add(ke.clip(r));
+
+ BatchInputSplit split = new BatchInputSplit(tableName, tableId, clippedRanges, new String[] {location});
+ AccumuloInputSplit.updateSplit(split, instance, tableConfig, principal, token, auths, logLevel);
+
+ splits.add(split);
+ } else {
+ // not grouping by tablet
+ for (Range r : extentRanges.getValue()) {
+ if (autoAdjust) {
+ // divide ranges into smaller ranges, based on the tablets
+ RangeInputSplit split = new RangeInputSplit(tableName, tableId, ke.clip(r), new String[] {location});
+ AccumuloInputSplit.updateSplit(split, instance, tableConfig, principal, token, auths, logLevel);
+ split.setOffline(tableConfig.isOfflineScan());
+ split.setIsolatedScan(tableConfig.shouldUseIsolatedScanners());
+ split.setUsesLocalIterators(tableConfig.shouldUseLocalIterators());
+
+ splits.add(split);
+ } else {
+ // don't divide ranges
+ ArrayList<String> locations = splitsToAdd.get(r);
+ if (locations == null)
+ locations = new ArrayList<String>(1);
+ locations.add(location);
+ splitsToAdd.put(r, locations);
+ }
}
}
}
@@ -663,19 +746,10 @@ public abstract class AbstractInputFormat<K,V> implements InputFormat<K,V> {
if (!autoAdjust)
for (Map.Entry<Range,ArrayList<String>> entry : splitsToAdd.entrySet()) {
RangeInputSplit split = new RangeInputSplit(tableName, tableId, entry.getKey(), entry.getValue().toArray(new String[0]));
-
+ AccumuloInputSplit.updateSplit(split, instance, tableConfig, principal, token, auths, logLevel);
split.setOffline(tableConfig.isOfflineScan());
split.setIsolatedScan(tableConfig.shouldUseIsolatedScanners());
split.setUsesLocalIterators(tableConfig.shouldUseLocalIterators());
- split.setMockInstance(mockInstance);
- split.setFetchedColumns(tableConfig.getFetchedColumns());
- split.setPrincipal(principal);
- split.setToken(token);
- split.setInstanceName(instance.getInstanceName());
- split.setZooKeepers(instance.getZooKeepers());
- split.setAuths(auths);
- split.setIterators(tableConfig.getIterators());
- split.setLogLevel(logLevel);
splits.add(split);
}
@@ -683,5 +757,4 @@ public abstract class AbstractInputFormat<K,V> implements InputFormat<K,V> {
return splits.toArray(new InputSplit[splits.size()]);
}
-
}
http://git-wip-us.apache.org/repos/asf/accumulo/blob/956a50ec/core/src/main/java/org/apache/accumulo/core/client/mapred/AccumuloInputFormat.java
----------------------------------------------------------------------
diff --git a/core/src/main/java/org/apache/accumulo/core/client/mapred/AccumuloInputFormat.java b/core/src/main/java/org/apache/accumulo/core/client/mapred/AccumuloInputFormat.java
index 2cdc236..2c23552 100644
--- a/core/src/main/java/org/apache/accumulo/core/client/mapred/AccumuloInputFormat.java
+++ b/core/src/main/java/org/apache/accumulo/core/client/mapred/AccumuloInputFormat.java
@@ -54,12 +54,14 @@ public class AccumuloInputFormat extends InputFormatBase<Key,Value> {
log.setLevel(getLogLevel(job));
// Override the log level from the configuration as if the RangeInputSplit has one it's the more correct one to use.
- if (split instanceof org.apache.accumulo.core.client.mapreduce.RangeInputSplit) {
- org.apache.accumulo.core.client.mapreduce.RangeInputSplit risplit = (org.apache.accumulo.core.client.mapreduce.RangeInputSplit) split;
- Level level = risplit.getLogLevel();
+ if (split instanceof org.apache.accumulo.core.client.mapreduce.impl.AccumuloInputSplit) {
+ org.apache.accumulo.core.client.mapreduce.impl.AccumuloInputSplit accSplit = (org.apache.accumulo.core.client.mapreduce.impl.AccumuloInputSplit) split;
+ Level level = accSplit.getLogLevel();
if (null != level) {
log.setLevel(level);
}
+ } else {
+ throw new IllegalArgumentException("No RecordReader for " + split.getClass().toString());
}
RecordReaderBase<Key,Value> recordReader = new RecordReaderBase<Key,Value>() {
http://git-wip-us.apache.org/repos/asf/accumulo/blob/956a50ec/core/src/main/java/org/apache/accumulo/core/client/mapred/InputFormatBase.java
----------------------------------------------------------------------
diff --git a/core/src/main/java/org/apache/accumulo/core/client/mapred/InputFormatBase.java b/core/src/main/java/org/apache/accumulo/core/client/mapred/InputFormatBase.java
index 5a8f592..517cd19 100644
--- a/core/src/main/java/org/apache/accumulo/core/client/mapred/InputFormatBase.java
+++ b/core/src/main/java/org/apache/accumulo/core/client/mapred/InputFormatBase.java
@@ -178,6 +178,7 @@ public abstract class InputFormatBase<K,V> extends AbstractInputFormat<K,V> {
/**
* Determines whether a configuration has auto-adjust ranges enabled.
+ * Must be enabled when {@link #setBatchScan(JobConf, boolean)} is true.
*
* @param job
* the Hadoop context for the configured job
@@ -297,6 +298,48 @@ public abstract class InputFormatBase<K,V> extends AbstractInputFormat<K,V> {
}
/**
+ * Controls the use of the {@link org.apache.accumulo.core.client.BatchScanner} in this job.
+ * Using this feature will group Ranges by their source tablet, producing an InputSplit per tablet
+ * rather than per Range. This batching helps to reduce overhead when querying a large number of small ranges.
+ * (ex: when doing quad-tree decomposition for spatial queries)
+ * <p>
+ * In order to achieve good locality of InputSplits this option always clips the input Ranges to tablet boundaries.
+ * This may result in one input Range contributing to several InputSplits.
+ * <p>
+ * Note: that the value of {@link #setAutoAdjustRanges(JobConf, boolean)} is ignored and is assumed to be true when BatchScan option is enabled.
+ * <p>
+ * This configuration is incompatible with:
+ * <ul>
+ * <li>{@link #setOfflineTableScan(JobConf, boolean)}</li>
+ * <li>{@link #setLocalIterators(JobConf, boolean)}</li>
+ * <li>{@link #setScanIsolation(JobConf, boolean)}</li>
+ * </ul>
+ * <p>
+ * By default, this feature is <b>disabled</b>.
+ *
+ * @param job
+ * the Hadoop job instance to be configured
+ * @param enableFeature
+ * the feature is enabled if true, disabled otherwise
+ * @since 1.7.0
+ */
+ public static void setBatchScan(JobConf job, boolean enableFeature) {
+ InputConfigurator.setBatchScan(CLASS, job, enableFeature);
+ }
+
+ /**
+ * Determines whether a configuration has the {@link org.apache.accumulo.core.client.BatchScanner} feature enabled.
+ *
+ * @param job
+ * the Hadoop context for the configured job
+ * @since 1.7.0
+ * @see #setBatchScan(JobConf, boolean)
+ */
+ public static boolean isBatchScan(JobConf job) {
+ return InputConfigurator.isBatchScan(CLASS, job);
+ }
+
+ /**
* Initializes an Accumulo {@link org.apache.accumulo.core.client.impl.TabletLocator} based on the configuration.
*
* @param job
@@ -315,19 +358,8 @@ public abstract class InputFormatBase<K,V> extends AbstractInputFormat<K,V> {
protected abstract static class RecordReaderBase<K,V> extends AbstractRecordReader<K,V> {
@Override
- protected void setupIterators(JobConf job, Scanner scanner, String tableName, org.apache.accumulo.core.client.mapred.RangeInputSplit split) {
- List<IteratorSetting> iterators = null;
-
- if (null == split) {
- iterators = getIterators(job);
- } else {
- iterators = split.getIterators();
- if (null == iterators) {
- iterators = getIterators(job);
- }
- }
-
- setupIterators(iterators, scanner);
+ protected List<IteratorSetting> jobIterators(JobConf job, String tableName) {
+ return getIterators(job);
}
/**
@@ -337,7 +369,9 @@ public abstract class InputFormatBase<K,V> extends AbstractInputFormat<K,V> {
* the iterators to set
* @param scanner
* the scanner to configure
+ * @deprecated since 1.7.0; Use {@link #jobIterators} instead.
*/
+ @Deprecated
protected void setupIterators(List<IteratorSetting> iterators, Scanner scanner) {
for (IteratorSetting iterator : iterators) {
scanner.addScanIterator(iterator);
http://git-wip-us.apache.org/repos/asf/accumulo/blob/956a50ec/core/src/main/java/org/apache/accumulo/core/client/mapred/impl/BatchInputSplit.java
----------------------------------------------------------------------
diff --git a/core/src/main/java/org/apache/accumulo/core/client/mapred/impl/BatchInputSplit.java b/core/src/main/java/org/apache/accumulo/core/client/mapred/impl/BatchInputSplit.java
new file mode 100644
index 0000000..619f9cd
--- /dev/null
+++ b/core/src/main/java/org/apache/accumulo/core/client/mapred/impl/BatchInputSplit.java
@@ -0,0 +1,42 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.accumulo.core.client.mapred.impl;
+
+import java.io.IOException;
+import java.util.Collection;
+
+import org.apache.accumulo.core.data.Range;
+import org.apache.hadoop.mapred.InputSplit;
+
+/**
+ * The Class BatchInputSplit. Encapsulates Accumulo ranges for use in Map Reduce jobs.
+ * Can contain several Ranges per InputSplit.
+ */
+public class BatchInputSplit extends org.apache.accumulo.core.client.mapreduce.impl.BatchInputSplit implements InputSplit {
+
+ public BatchInputSplit() {
+ super();
+ }
+
+ public BatchInputSplit(BatchInputSplit split) throws IOException {
+ super(split);
+ }
+
+ public BatchInputSplit(String table, String tableId, Collection<Range> ranges, String[] location) {
+ super(table, tableId, ranges, location);
+ }
+}
http://git-wip-us.apache.org/repos/asf/accumulo/blob/956a50ec/core/src/main/java/org/apache/accumulo/core/client/mapreduce/AbstractInputFormat.java
----------------------------------------------------------------------
diff --git a/core/src/main/java/org/apache/accumulo/core/client/mapreduce/AbstractInputFormat.java b/core/src/main/java/org/apache/accumulo/core/client/mapreduce/AbstractInputFormat.java
index 300e92b..e3c3c42 100644
--- a/core/src/main/java/org/apache/accumulo/core/client/mapreduce/AbstractInputFormat.java
+++ b/core/src/main/java/org/apache/accumulo/core/client/mapreduce/AbstractInputFormat.java
@@ -35,6 +35,8 @@ import org.apache.accumulo.core.client.Connector;
import org.apache.accumulo.core.client.Instance;
import org.apache.accumulo.core.client.IsolatedScanner;
import org.apache.accumulo.core.client.Scanner;
+import org.apache.accumulo.core.client.BatchScanner;
+import org.apache.accumulo.core.client.ScannerBase;
import org.apache.accumulo.core.client.TableDeletedException;
import org.apache.accumulo.core.client.TableNotFoundException;
import org.apache.accumulo.core.client.TableOfflineException;
@@ -44,6 +46,8 @@ import org.apache.accumulo.core.client.impl.OfflineScanner;
import org.apache.accumulo.core.client.impl.ScannerImpl;
import org.apache.accumulo.core.client.impl.Tables;
import org.apache.accumulo.core.client.impl.TabletLocator;
+import org.apache.accumulo.core.client.mapreduce.impl.AccumuloInputSplit;
+import org.apache.accumulo.core.client.mapreduce.impl.BatchInputSplit;
import org.apache.accumulo.core.client.mapreduce.lib.impl.ConfiguratorBase;
import org.apache.accumulo.core.client.mapreduce.lib.impl.InputConfigurator;
import org.apache.accumulo.core.client.mock.MockInstance;
@@ -61,6 +65,7 @@ import org.apache.accumulo.core.security.Authorizations;
import org.apache.accumulo.core.security.Credentials;
import org.apache.accumulo.core.util.Pair;
import org.apache.accumulo.core.util.UtilWaitThread;
+import org.apache.accumulo.core.client.IteratorSetting;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.InputFormat;
@@ -413,7 +418,50 @@ public abstract class AbstractInputFormat<K,V> extends InputFormat<K,V> {
protected abstract static class AbstractRecordReader<K,V> extends RecordReader<K,V> {
protected long numKeysRead;
protected Iterator<Map.Entry<Key,Value>> scannerIterator;
- protected RangeInputSplit split;
+ protected ScannerBase scannerBase;
+ protected AccumuloInputSplit split;
+
+ /**
+ * Extracts Iterators settings from the context to be used by RecordReader.
+ *
+ * @param context
+ * the Hadoop context for the configured job
+ * @param tableName
+ * the table name for which the scanner is configured
+ * @return List of iterator settings for given table
+ * @since 1.7.0
+ */
+ protected abstract List<IteratorSetting> contextIterators(TaskAttemptContext context, String tableName);
+
+ /**
+ * Configures the iterators on a scanner for the given table name.
+ * Will attempt to use configuration from the InputSplit, on failure will try to extract them from TaskAttemptContext.
+ *
+ * @param context
+ * the Hadoop context for the configured job
+ * @param tableName
+ * the table name for which the scanner is configured
+ * @param scanner
+ * the scanner for which to configure the iterators
+ * @param split
+ * InputSplit containing configurations
+ * @since 1.7.0
+ */
+ private void setupIterators(TaskAttemptContext context, ScannerBase scanner, String tableName, AccumuloInputSplit split) {
+ List<IteratorSetting> iterators = null;
+
+ if (null == split) {
+ iterators = contextIterators(context, tableName);
+ } else {
+ iterators = split.getIterators();
+ if (null == iterators) {
+ iterators = contextIterators(context, tableName);
+ }
+ }
+
+ for (IteratorSetting iterator : iterators)
+ scanner.addScanIterator(iterator);
+ }
/**
* Configures the iterators on a scanner for the given table name.
@@ -425,15 +473,18 @@ public abstract class AbstractInputFormat<K,V> extends InputFormat<K,V> {
* @param tableName
* the table name for which the scanner is configured
* @since 1.6.0
+ * @deprecated since 1.7.0; Use {@link #contextIterators} instead.
*/
- protected abstract void setupIterators(TaskAttemptContext context, Scanner scanner, String tableName, RangeInputSplit split);
+ @Deprecated
+ protected void setupIterators(TaskAttemptContext context, Scanner scanner, String tableName, RangeInputSplit split) {
+ setupIterators(context, (ScannerBase) scanner, tableName, (AccumuloInputSplit) split);
+ }
@Override
public void initialize(InputSplit inSplit, TaskAttemptContext attempt) throws IOException {
- Scanner scanner;
- split = (RangeInputSplit) inSplit;
- log.debug("Initializing input split: " + split.getRange());
+ split = (AccumuloInputSplit) inSplit;
+ log.debug("Initializing input split: " + split.toString());
Instance instance = split.getInstance(getClientConfiguration(attempt));
if (null == instance) {
@@ -461,46 +512,71 @@ public abstract class AbstractInputFormat<K,V> extends InputFormat<K,V> {
// but the scanner will use the table id resolved at job setup time
InputTableConfig tableConfig = getInputTableConfig(attempt, split.getTableName());
- Boolean isOffline = split.isOffline();
- if (null == isOffline) {
- isOffline = tableConfig.isOfflineScan();
- }
+ log.debug("Creating connector with user: " + principal);
+ log.debug("Creating scanner for table: " + table);
+ log.debug("Authorizations are: " + authorizations);
- Boolean isIsolated = split.isIsolatedScan();
- if (null == isIsolated) {
- isIsolated = tableConfig.shouldUseIsolatedScanners();
- }
+ if (split instanceof RangeInputSplit) {
+ RangeInputSplit rangeSplit = (RangeInputSplit) split;
+ Scanner scanner;
- Boolean usesLocalIterators = split.usesLocalIterators();
- if (null == usesLocalIterators) {
- usesLocalIterators = tableConfig.shouldUseLocalIterators();
- }
+ Boolean isOffline = rangeSplit.isOffline();
+ if (null == isOffline) {
+ isOffline = tableConfig.isOfflineScan();
+ }
- try {
- log.debug("Creating connector with user: " + principal);
- log.debug("Creating scanner for table: " + table);
- log.debug("Authorizations are: " + authorizations);
- if (isOffline) {
- scanner = new OfflineScanner(instance, new Credentials(principal, token), split.getTableId(), authorizations);
- } else if (instance instanceof MockInstance) {
- scanner = instance.getConnector(principal, token).createScanner(split.getTableName(), authorizations);
- } else {
- ClientConfiguration clientConf = getClientConfiguration(attempt);
- ClientContext context = new ClientContext(instance, new Credentials(principal, token), clientConf);
- scanner = new ScannerImpl(context, split.getTableId(), authorizations);
+ Boolean isIsolated = rangeSplit.isIsolatedScan();
+ if (null == isIsolated) {
+ isIsolated = tableConfig.shouldUseIsolatedScanners();
}
- if (isIsolated) {
- log.info("Creating isolated scanner");
- scanner = new IsolatedScanner(scanner);
+
+ Boolean usesLocalIterators = rangeSplit.usesLocalIterators();
+ if (null == usesLocalIterators) {
+ usesLocalIterators = tableConfig.shouldUseLocalIterators();
}
- if (usesLocalIterators) {
- log.info("Using local iterators");
- scanner = new ClientSideIteratorScanner(scanner);
+
+ try {
+ if (isOffline) {
+ scanner = new OfflineScanner(instance, new Credentials(principal, token), split.getTableId(), authorizations);
+ } else if (instance instanceof MockInstance) {
+ scanner = instance.getConnector(principal, token).createScanner(split.getTableName(), authorizations);
+ } else {
+ ClientConfiguration clientConf = getClientConfiguration(attempt);
+ ClientContext context = new ClientContext(instance, new Credentials(principal, token), clientConf);
+ scanner = new ScannerImpl(context, split.getTableId(), authorizations);
+ }
+ if (isIsolated) {
+ log.info("Creating isolated scanner");
+ scanner = new IsolatedScanner(scanner);
+ }
+ if (usesLocalIterators) {
+ log.info("Using local iterators");
+ scanner = new ClientSideIteratorScanner(scanner);
+ }
+
+ setupIterators(attempt, scanner, split.getTableName(), split);
+ } catch (Exception e) {
+ throw new IOException(e);
}
- setupIterators(attempt, scanner, split.getTableName(), split);
- } catch (Exception e) {
- throw new IOException(e);
+ scanner.setRange(rangeSplit.getRange());
+ scannerBase = scanner;
+
+ } else if (split instanceof BatchInputSplit) {
+ BatchInputSplit batchSplit = (BatchInputSplit) split;
+
+ BatchScanner scanner;
+ try{
+ // Note: BatchScanner will use at most one thread per tablet, currently BatchInputSplit will not span tablets
+ int scanThreads = 1;
+ scanner = instance.getConnector(principal, token).createBatchScanner(split.getTableName(), authorizations, scanThreads);
+ setupIterators(attempt, scanner, split.getTableName(), split);
+ } catch (Exception e) {
+ throw new IOException(e);
+ }
+
+ scanner.setRanges(batchSplit.getRanges());
+ scannerBase = scanner;
}
Collection<Pair<Text,Text>> columns = split.getFetchedColumns();
@@ -512,22 +588,23 @@ public abstract class AbstractInputFormat<K,V> extends InputFormat<K,V> {
for (Pair<Text,Text> c : columns) {
if (c.getSecond() != null) {
log.debug("Fetching column " + c.getFirst() + ":" + c.getSecond());
- scanner.fetchColumn(c.getFirst(), c.getSecond());
+ scannerBase.fetchColumn(c.getFirst(), c.getSecond());
} else {
log.debug("Fetching column family " + c.getFirst());
- scanner.fetchColumnFamily(c.getFirst());
+ scannerBase.fetchColumnFamily(c.getFirst());
}
}
- scanner.setRange(split.getRange());
+ scannerIterator = scannerBase.iterator();
numKeysRead = 0;
-
- // do this last after setting all scanner options
- scannerIterator = scanner.iterator();
}
@Override
- public void close() {}
+ public void close() {
+ if (null != scannerBase) {
+ scannerBase.close();
+ }
+ }
@Override
public float getProgress() throws IOException {
@@ -592,26 +669,32 @@ public abstract class AbstractInputFormat<K,V> extends InputFormat<K,V> {
InputTableConfig tableConfig = tableConfigEntry.getValue();
Instance instance = getInstance(context);
- boolean mockInstance;
String tableId;
// resolve table name to id once, and use id from this point forward
if (instance instanceof MockInstance) {
tableId = "";
- mockInstance = true;
} else {
try {
tableId = Tables.getTableId(instance, tableName);
} catch (TableNotFoundException e) {
throw new IOException(e);
}
- mockInstance = false;
}
Authorizations auths = getScanAuthorizations(context);
String principal = getPrincipal(context);
AuthenticationToken token = getAuthenticationToken(context);
+ boolean batchScan = InputConfigurator.isBatchScan(CLASS, context.getConfiguration());
+ boolean supportBatchScan =
+ !(tableConfig.isOfflineScan() || tableConfig.shouldUseIsolatedScanners() || tableConfig.shouldUseLocalIterators());
+ if (batchScan && !supportBatchScan)
+ throw new IllegalArgumentException("BatchScanner optimization not available for offline scan, isolated, or local iterators");
+
boolean autoAdjust = tableConfig.shouldAutoAdjustRanges();
+ if (batchScan && !autoAdjust)
+ throw new IllegalArgumentException("AutoAdjustRanges must be enabled when using BatchScanner optimization");
+
List<Range> ranges = autoAdjust ? Range.mergeOverlapping(tableConfig.getRanges()) : tableConfig.getRanges();
if (ranges.isEmpty()) {
ranges = new ArrayList<Range>(1);
@@ -654,6 +737,8 @@ public abstract class AbstractInputFormat<K,V> extends InputFormat<K,V> {
throw new IOException(e);
}
+ // all of this code will add either range per each locations or split ranges and add range-location split
+ // Map from Range to Array of Locations, we only use this if we're don't split
HashMap<Range,ArrayList<String>> splitsToAdd = null;
if (!autoAdjust)
@@ -670,32 +755,35 @@ public abstract class AbstractInputFormat<K,V> extends InputFormat<K,V> {
}
for (Map.Entry<KeyExtent,List<Range>> extentRanges : tserverBin.getValue().entrySet()) {
Range ke = extentRanges.getKey().toDataRange();
- for (Range r : extentRanges.getValue()) {
- if (autoAdjust) {
- // divide ranges into smaller ranges, based on the tablets
- RangeInputSplit split = new RangeInputSplit(tableName, tableId, ke.clip(r), new String[] {location});
-
- split.setOffline(tableConfig.isOfflineScan());
- split.setIsolatedScan(tableConfig.shouldUseIsolatedScanners());
- split.setUsesLocalIterators(tableConfig.shouldUseLocalIterators());
- split.setMockInstance(mockInstance);
- split.setFetchedColumns(tableConfig.getFetchedColumns());
- split.setPrincipal(principal);
- split.setToken(token);
- split.setInstanceName(instance.getInstanceName());
- split.setZooKeepers(instance.getZooKeepers());
- split.setAuths(auths);
- split.setIterators(tableConfig.getIterators());
- split.setLogLevel(logLevel);
-
- splits.add(split);
- } else {
- // don't divide ranges
- ArrayList<String> locations = splitsToAdd.get(r);
- if (locations == null)
- locations = new ArrayList<String>(1);
- locations.add(location);
- splitsToAdd.put(r, locations);
+ if (batchScan) {
+ // group ranges by tablet to be read by a BatchScanner
+ ArrayList<Range> clippedRanges = new ArrayList<Range>();
+ for(Range r: extentRanges.getValue())
+ clippedRanges.add(ke.clip(r));
+ BatchInputSplit split = new BatchInputSplit(tableName, tableId, clippedRanges, new String[] {location});
+ AccumuloInputSplit.updateSplit(split, instance, tableConfig, principal, token, auths, logLevel);
+
+ splits.add(split);
+ } else {
+ // not grouping by tablet
+ for (Range r : extentRanges.getValue()) {
+ if (autoAdjust) {
+ // divide ranges into smaller ranges, based on the tablets
+ RangeInputSplit split = new RangeInputSplit(tableName, tableId, ke.clip(r), new String[] {location});
+ AccumuloInputSplit.updateSplit(split, instance, tableConfig, principal, token, auths, logLevel);
+ split.setOffline(tableConfig.isOfflineScan());
+ split.setIsolatedScan(tableConfig.shouldUseIsolatedScanners());
+ split.setUsesLocalIterators(tableConfig.shouldUseLocalIterators());
+
+ splits.add(split);
+ } else {
+ // don't divide ranges
+ ArrayList<String> locations = splitsToAdd.get(r);
+ if (locations == null)
+ locations = new ArrayList<String>(1);
+ locations.add(location);
+ splitsToAdd.put(r, locations);
+ }
}
}
}
@@ -704,23 +792,14 @@ public abstract class AbstractInputFormat<K,V> extends InputFormat<K,V> {
if (!autoAdjust)
for (Map.Entry<Range,ArrayList<String>> entry : splitsToAdd.entrySet()) {
RangeInputSplit split = new RangeInputSplit(tableName, tableId, entry.getKey(), entry.getValue().toArray(new String[0]));
-
+ AccumuloInputSplit.updateSplit(split, instance, tableConfig, principal, token, auths, logLevel);
split.setOffline(tableConfig.isOfflineScan());
split.setIsolatedScan(tableConfig.shouldUseIsolatedScanners());
split.setUsesLocalIterators(tableConfig.shouldUseLocalIterators());
- split.setMockInstance(mockInstance);
- split.setFetchedColumns(tableConfig.getFetchedColumns());
- split.setPrincipal(principal);
- split.setToken(token);
- split.setInstanceName(instance.getInstanceName());
- split.setZooKeepers(instance.getZooKeepers());
- split.setAuths(auths);
- split.setIterators(tableConfig.getIterators());
- split.setLogLevel(logLevel);
splits.add(split);
}
}
return splits;
}
-}
+}
\ No newline at end of file
http://git-wip-us.apache.org/repos/asf/accumulo/blob/956a50ec/core/src/main/java/org/apache/accumulo/core/client/mapreduce/AccumuloInputFormat.java
----------------------------------------------------------------------
diff --git a/core/src/main/java/org/apache/accumulo/core/client/mapreduce/AccumuloInputFormat.java b/core/src/main/java/org/apache/accumulo/core/client/mapreduce/AccumuloInputFormat.java
index b98cb77..7af5e66 100644
--- a/core/src/main/java/org/apache/accumulo/core/client/mapreduce/AccumuloInputFormat.java
+++ b/core/src/main/java/org/apache/accumulo/core/client/mapreduce/AccumuloInputFormat.java
@@ -21,6 +21,7 @@ import java.util.Map.Entry;
import org.apache.accumulo.core.client.ClientConfiguration;
import org.apache.accumulo.core.client.security.tokens.AuthenticationToken;
+import org.apache.accumulo.core.client.mapreduce.impl.AccumuloInputSplit;
import org.apache.accumulo.core.data.Key;
import org.apache.accumulo.core.data.Value;
import org.apache.accumulo.core.security.Authorizations;
@@ -52,17 +53,18 @@ public class AccumuloInputFormat extends InputFormatBase<Key,Value> {
public RecordReader<Key,Value> createRecordReader(InputSplit split, TaskAttemptContext context) throws IOException, InterruptedException {
log.setLevel(getLogLevel(context));
- // Override the log level from the configuration as if the RangeInputSplit has one it's the more correct one to use.
- if (split instanceof org.apache.accumulo.core.client.mapreduce.RangeInputSplit) {
- org.apache.accumulo.core.client.mapreduce.RangeInputSplit risplit = (org.apache.accumulo.core.client.mapreduce.RangeInputSplit) split;
- Level level = risplit.getLogLevel();
+ // Override the log level from the configuration as if the InputSplit has one it's the more correct one to use.
+ if (split instanceof AccumuloInputSplit) {
+ AccumuloInputSplit accSplit = (AccumuloInputSplit) split;
+ Level level = accSplit.getLogLevel();
if (null != level) {
log.setLevel(level);
}
+ } else {
+ throw new IllegalArgumentException("No RecordReader for " + split.getClass().toString());
}
return new RecordReaderBase<Key,Value>() {
-
@Override
public boolean nextKeyValue() throws IOException, InterruptedException {
if (scannerIterator.hasNext()) {
http://git-wip-us.apache.org/repos/asf/accumulo/blob/956a50ec/core/src/main/java/org/apache/accumulo/core/client/mapreduce/AccumuloMultiTableInputFormat.java
----------------------------------------------------------------------
diff --git a/core/src/main/java/org/apache/accumulo/core/client/mapreduce/AccumuloMultiTableInputFormat.java b/core/src/main/java/org/apache/accumulo/core/client/mapreduce/AccumuloMultiTableInputFormat.java
index bed0def..679256b 100644
--- a/core/src/main/java/org/apache/accumulo/core/client/mapreduce/AccumuloMultiTableInputFormat.java
+++ b/core/src/main/java/org/apache/accumulo/core/client/mapreduce/AccumuloMultiTableInputFormat.java
@@ -24,7 +24,6 @@ import java.util.Map;
import org.apache.accumulo.core.client.ClientConfiguration;
import org.apache.accumulo.core.client.IteratorSetting;
-import org.apache.accumulo.core.client.Scanner;
import org.apache.accumulo.core.client.mapreduce.lib.impl.InputConfigurator;
import org.apache.accumulo.core.client.security.tokens.AuthenticationToken;
import org.apache.accumulo.core.data.Key;
@@ -86,15 +85,8 @@ public class AccumuloMultiTableInputFormat extends AbstractInputFormat<Key,Value
}
@Override
- protected void setupIterators(TaskAttemptContext context, Scanner scanner, String tableName, RangeInputSplit split) {
- List<IteratorSetting> iterators = split.getIterators();
- if (null == iterators) {
- iterators = getInputTableConfig(context, tableName).getIterators();
- }
-
- for (IteratorSetting setting : iterators) {
- scanner.addScanIterator(setting);
- }
+ protected List<IteratorSetting> contextIterators(TaskAttemptContext context, String tableName) {
+ return getInputTableConfig(context, tableName).getIterators();
}
};
}
http://git-wip-us.apache.org/repos/asf/accumulo/blob/956a50ec/core/src/main/java/org/apache/accumulo/core/client/mapreduce/InputFormatBase.java
----------------------------------------------------------------------
diff --git a/core/src/main/java/org/apache/accumulo/core/client/mapreduce/InputFormatBase.java b/core/src/main/java/org/apache/accumulo/core/client/mapreduce/InputFormatBase.java
index 76ca401..dcc4fd5 100644
--- a/core/src/main/java/org/apache/accumulo/core/client/mapreduce/InputFormatBase.java
+++ b/core/src/main/java/org/apache/accumulo/core/client/mapreduce/InputFormatBase.java
@@ -177,6 +177,7 @@ public abstract class InputFormatBase<K,V> extends AbstractInputFormat<K,V> {
/**
* Determines whether a configuration has auto-adjust ranges enabled.
+ * Must be enabled when {@link #setBatchScan(Job, boolean)} is true.
*
* @param context
* the Hadoop context for the configured job
@@ -296,6 +297,48 @@ public abstract class InputFormatBase<K,V> extends AbstractInputFormat<K,V> {
}
/**
+ * Controls the use of the {@link org.apache.accumulo.core.client.BatchScanner} in this job.
+ * Using this feature will group Ranges by their source tablet, producing an InputSplit per tablet
+ * rather than per Range. This batching helps to reduce overhead when querying a large number of small ranges.
+ * (ex: when doing quad-tree decomposition for spatial queries)
+ * <p>
+ * In order to achieve good locality of InputSplits this option always clips the input Ranges to tablet boundaries.
+ * This may result in one input Range contributing to several InputSplits.
+ * <p>
+ * Note: that the value of {@link #setAutoAdjustRanges(Job, boolean)} is ignored and is assumed to be true when BatchScan option is enabled.
+ * <p>
+ * This configuration is incompatible with:
+ * <ul>
+ * <li>{@link #setOfflineTableScan(org.apache.hadoop.mapreduce.Job, boolean)}</li>
+ * <li>{@link #setLocalIterators(org.apache.hadoop.mapreduce.Job, boolean)}</li>
+ * <li>{@link #setScanIsolation(org.apache.hadoop.mapreduce.Job, boolean)}</li>
+ * </ul>
+ * <p>
+ * By default, this feature is <b>disabled</b>.
+ *
+ * @param job
+ * the Hadoop job instance to be configured
+ * @param enableFeature
+ * the feature is enabled if true, disabled otherwise
+ * @since 1.7.0
+ */
+ public static void setBatchScan(Job job, boolean enableFeature) {
+ InputConfigurator.setBatchScan(CLASS, job.getConfiguration(), enableFeature);
+ }
+
+ /**
+ * Determines whether a configuration has the {@link org.apache.accumulo.core.client.BatchScanner} feature enabled.
+ *
+ * @param context
+ * the Hadoop context for the configured job
+ * @since 1.7.0
+ * @see #setBatchScan(Job, boolean)
+ */
+ public static boolean isBatchScan(JobContext context) {
+ return InputConfigurator.isBatchScan(CLASS, context.getConfiguration());
+ }
+
+ /**
* Initializes an Accumulo {@link org.apache.accumulo.core.client.impl.TabletLocator} based on the configuration.
*
* @param context
@@ -314,8 +357,8 @@ public abstract class InputFormatBase<K,V> extends AbstractInputFormat<K,V> {
protected abstract static class RecordReaderBase<K,V> extends AbstractRecordReader<K,V> {
@Override
- protected void setupIterators(TaskAttemptContext context, Scanner scanner, String tableName, org.apache.accumulo.core.client.mapreduce.RangeInputSplit split) {
- setupIterators(context, scanner, split);
+ protected List<IteratorSetting> contextIterators(TaskAttemptContext context, String tableName) {
+ return getIterators(context);
}
/**
@@ -325,27 +368,21 @@ public abstract class InputFormatBase<K,V> extends AbstractInputFormat<K,V> {
* the Hadoop context for the configured job
* @param scanner
* the scanner to configure
+ * @deprecated since 1.7.0; Use {@link #contextIterators} instead.
*/
@Deprecated
protected void setupIterators(TaskAttemptContext context, Scanner scanner) {
- setupIterators(context, scanner, null);
+ // tableName is given as null as it will be ignored in eventual call to #contextIterators
+ setupIterators(context, scanner, null, null);
}
/**
* Initialize a scanner over the given input split using this task attempt configuration.
+ * @deprecated since 1.7.0; Use {@link #contextIterators} instead.
*/
+ @Deprecated
protected void setupIterators(TaskAttemptContext context, Scanner scanner, org.apache.accumulo.core.client.mapreduce.RangeInputSplit split) {
- List<IteratorSetting> iterators = null;
- if (null == split) {
- iterators = getIterators(context);
- } else {
- iterators = split.getIterators();
- if (null == iterators) {
- iterators = getIterators(context);
- }
- }
- for (IteratorSetting iterator : iterators)
- scanner.addScanIterator(iterator);
+ setupIterators(context, scanner, null, split);
}
}
http://git-wip-us.apache.org/repos/asf/accumulo/blob/956a50ec/core/src/main/java/org/apache/accumulo/core/client/mapreduce/RangeInputSplit.java
----------------------------------------------------------------------
diff --git a/core/src/main/java/org/apache/accumulo/core/client/mapreduce/RangeInputSplit.java b/core/src/main/java/org/apache/accumulo/core/client/mapreduce/RangeInputSplit.java
index 872ee7b..6c870a0 100644
--- a/core/src/main/java/org/apache/accumulo/core/client/mapreduce/RangeInputSplit.java
+++ b/core/src/main/java/org/apache/accumulo/core/client/mapreduce/RangeInputSplit.java
@@ -16,114 +16,53 @@
*/
package org.apache.accumulo.core.client.mapreduce;
-import static java.nio.charset.StandardCharsets.UTF_8;
-
import java.io.DataInput;
import java.io.DataOutput;
import java.io.IOException;
-import java.math.BigInteger;
-import java.util.ArrayList;
import java.util.Arrays;
-import java.util.Collection;
-import java.util.HashSet;
-import java.util.List;
-import java.util.Set;
-import org.apache.accumulo.core.client.ClientConfiguration;
-import org.apache.accumulo.core.client.Instance;
-import org.apache.accumulo.core.client.IteratorSetting;
-import org.apache.accumulo.core.client.ZooKeeperInstance;
-import org.apache.accumulo.core.client.mapreduce.lib.impl.ConfiguratorBase.TokenSource;
-import org.apache.accumulo.core.client.mapreduce.lib.impl.InputConfigurator;
-import org.apache.accumulo.core.client.mock.MockInstance;
-import org.apache.accumulo.core.client.security.tokens.AuthenticationToken;
-import org.apache.accumulo.core.client.security.tokens.AuthenticationToken.AuthenticationTokenSerializer;
-import org.apache.accumulo.core.data.ByteSequence;
+import org.apache.accumulo.core.client.mapreduce.impl.AccumuloInputSplit;
import org.apache.accumulo.core.data.Key;
import org.apache.accumulo.core.data.PartialKey;
import org.apache.accumulo.core.data.Range;
-import org.apache.accumulo.core.security.Authorizations;
-import org.apache.accumulo.core.util.Base64;
-import org.apache.accumulo.core.util.Pair;
-import org.apache.hadoop.io.Text;
-import org.apache.hadoop.io.Writable;
-import org.apache.hadoop.mapreduce.InputSplit;
-import org.apache.log4j.Level;
/**
* The Class RangeInputSplit. Encapsulates an Accumulo range for use in Map Reduce jobs.
*/
-public class RangeInputSplit extends InputSplit implements Writable {
+public class RangeInputSplit extends AccumuloInputSplit {
private Range range;
- private String[] locations;
- private String tableId, tableName, instanceName, zooKeepers, principal;
- private TokenSource tokenSource;
- private String tokenFile;
- private AuthenticationToken token;
- private Boolean offline, mockInstance, isolatedScan, localIterators;
- private Authorizations auths;
- private Set<Pair<Text,Text>> fetchedColumns;
- private List<IteratorSetting> iterators;
- private Level level;
+ private Boolean offline, isolatedScan, localIterators;
public RangeInputSplit() {
range = new Range();
- locations = new String[0];
- tableName = "";
- tableId = "";
}
public RangeInputSplit(RangeInputSplit split) throws IOException {
+ super(split);
this.setRange(split.getRange());
- this.setLocations(split.getLocations());
- this.setTableName(split.getTableName());
- this.setTableId(split.getTableId());
}
protected RangeInputSplit(String table, String tableId, Range range, String[] locations) {
+ super(table, tableId, locations);
this.range = range;
- setLocations(locations);
- this.tableName = table;
- this.tableId = tableId;
- }
-
- public Range getRange() {
- return range;
- }
-
- private static byte[] extractBytes(ByteSequence seq, int numBytes) {
- byte[] bytes = new byte[numBytes + 1];
- bytes[0] = 0;
- for (int i = 0; i < numBytes; i++) {
- if (i >= seq.length())
- bytes[i + 1] = 0;
- else
- bytes[i + 1] = seq.byteAt(i);
- }
- return bytes;
- }
-
- public static float getProgress(ByteSequence start, ByteSequence end, ByteSequence position) {
- int maxDepth = Math.min(Math.max(end.length(), start.length()), position.length());
- BigInteger startBI = new BigInteger(extractBytes(start, maxDepth));
- BigInteger endBI = new BigInteger(extractBytes(end, maxDepth));
- BigInteger positionBI = new BigInteger(extractBytes(position, maxDepth));
- return (float) (positionBI.subtract(startBI).doubleValue() / endBI.subtract(startBI).doubleValue());
}
public float getProgress(Key currentKey) {
if (currentKey == null)
return 0f;
- if (range.getStartKey() != null && range.getEndKey() != null) {
- if (range.getStartKey().compareTo(range.getEndKey(), PartialKey.ROW) != 0) {
- // just look at the row progress
- return getProgress(range.getStartKey().getRowData(), range.getEndKey().getRowData(), currentKey.getRowData());
- } else if (range.getStartKey().compareTo(range.getEndKey(), PartialKey.ROW_COLFAM) != 0) {
- // just look at the column family progress
- return getProgress(range.getStartKey().getColumnFamilyData(), range.getEndKey().getColumnFamilyData(), currentKey.getColumnFamilyData());
- } else if (range.getStartKey().compareTo(range.getEndKey(), PartialKey.ROW_COLFAM_COLQUAL) != 0) {
- // just look at the column qualifier progress
- return getProgress(range.getStartKey().getColumnQualifierData(), range.getEndKey().getColumnQualifierData(), currentKey.getColumnQualifierData());
+ if (range.contains(currentKey)) {
+ // find the current range and report as if that is the single range
+ if (range.getStartKey() != null && range.getEndKey() != null) {
+ if (range.getStartKey().compareTo(range.getEndKey(), PartialKey.ROW) != 0) {
+ // just look at the row progress
+ return getProgress(range.getStartKey().getRowData(), range.getEndKey().getRowData(), currentKey.getRowData());
+ } else if (range.getStartKey().compareTo(range.getEndKey(), PartialKey.ROW_COLFAM) != 0) {
+ // just look at the column family progress
+ return getProgress(range.getStartKey().getColumnFamilyData(), range.getEndKey().getColumnFamilyData(), currentKey.getColumnFamilyData());
+ } else if (range.getStartKey().compareTo(range.getEndKey(), PartialKey.ROW_COLFAM_COLQUAL) != 0) {
+ // just look at the column qualifier progress
+ return getProgress(range.getStartKey().getColumnQualifierData(), range.getEndKey().getColumnQualifierData(), currentKey.getColumnQualifierData());
+ }
}
}
// if we can't figure it out, then claim no progress
@@ -135,38 +74,15 @@ public class RangeInputSplit extends InputSplit implements Writable {
*/
@Override
public long getLength() throws IOException {
- Text startRow = range.isInfiniteStartKey() ? new Text(new byte[] {Byte.MIN_VALUE}) : range.getStartKey().getRow();
- Text stopRow = range.isInfiniteStopKey() ? new Text(new byte[] {Byte.MAX_VALUE}) : range.getEndKey().getRow();
- int maxCommon = Math.min(7, Math.min(startRow.getLength(), stopRow.getLength()));
- long diff = 0;
-
- byte[] start = startRow.getBytes();
- byte[] stop = stopRow.getBytes();
- for (int i = 0; i < maxCommon; ++i) {
- diff |= 0xff & (start[i] ^ stop[i]);
- diff <<= Byte.SIZE;
- }
-
- if (startRow.getLength() != stopRow.getLength())
- diff |= 0xff;
-
- return diff + 1;
+ return getRangeLength(range);
}
- @Override
- public String[] getLocations() throws IOException {
- return Arrays.copyOf(locations, locations.length);
- }
@Override
public void readFields(DataInput in) throws IOException {
+ super.readFields(in);
+
range.readFields(in);
- tableName = in.readUTF();
- tableId = in.readUTF();
- int numLocs = in.readInt();
- locations = new String[numLocs];
- for (int i = 0; i < numLocs; ++i)
- locations[i] = in.readUTF();
if (in.readBoolean()) {
isolatedScan = in.readBoolean();
@@ -179,81 +95,13 @@ public class RangeInputSplit extends InputSplit implements Writable {
if (in.readBoolean()) {
localIterators = in.readBoolean();
}
-
- if (in.readBoolean()) {
- mockInstance = in.readBoolean();
- }
-
- if (in.readBoolean()) {
- int numColumns = in.readInt();
- List<String> columns = new ArrayList<String>(numColumns);
- for (int i = 0; i < numColumns; i++) {
- columns.add(in.readUTF());
- }
-
- fetchedColumns = InputConfigurator.deserializeFetchedColumns(columns);
- }
-
- if (in.readBoolean()) {
- String strAuths = in.readUTF();
- auths = new Authorizations(strAuths.getBytes(UTF_8));
- }
-
- if (in.readBoolean()) {
- principal = in.readUTF();
- }
-
- if (in.readBoolean()) {
- int ordinal = in.readInt();
- this.tokenSource = TokenSource.values()[ordinal];
-
- switch (this.tokenSource) {
- case INLINE:
- String tokenClass = in.readUTF();
- byte[] base64TokenBytes = in.readUTF().getBytes(UTF_8);
- byte[] tokenBytes = Base64.decodeBase64(base64TokenBytes);
-
- this.token = AuthenticationTokenSerializer.deserialize(tokenClass, tokenBytes);
- break;
-
- case FILE:
- this.tokenFile = in.readUTF();
-
- break;
- default:
- throw new IOException("Cannot parse unknown TokenSource ordinal");
- }
- }
-
- if (in.readBoolean()) {
- instanceName = in.readUTF();
- }
-
- if (in.readBoolean()) {
- zooKeepers = in.readUTF();
- }
-
- if (in.readBoolean()) {
- int numIterators = in.readInt();
- iterators = new ArrayList<IteratorSetting>(numIterators);
- for (int i = 0; i < numIterators; i++) {
- iterators.add(new IteratorSetting(in));
- }
- }
-
- if (in.readBoolean()) {
- level = Level.toLevel(in.readInt());
- }
}
@Override
public void write(DataOutput out) throws IOException {
+ super.write(out);
+
range.write(out);
- out.writeUTF(tableName);
- out.writeUTF(tableId);
- out.writeInt(locations.length);
- for (int i = 0; i < locations.length; ++i)
- out.writeUTF(locations[i]);
out.writeBoolean(null != isolatedScan);
if (null != isolatedScan) {
@@ -269,73 +117,13 @@ public class RangeInputSplit extends InputSplit implements Writable {
if (null != localIterators) {
out.writeBoolean(localIterators);
}
-
- out.writeBoolean(null != mockInstance);
- if (null != mockInstance) {
- out.writeBoolean(mockInstance);
- }
-
- out.writeBoolean(null != fetchedColumns);
- if (null != fetchedColumns) {
- String[] cols = InputConfigurator.serializeColumns(fetchedColumns);
- out.writeInt(cols.length);
- for (String col : cols) {
- out.writeUTF(col);
- }
- }
-
- out.writeBoolean(null != auths);
- if (null != auths) {
- out.writeUTF(auths.serialize());
- }
-
- out.writeBoolean(null != principal);
- if (null != principal) {
- out.writeUTF(principal);
- }
-
- out.writeBoolean(null != tokenSource);
- if (null != tokenSource) {
- out.writeInt(tokenSource.ordinal());
-
- if (null != token && null != tokenFile) {
- throw new IOException("Cannot use both inline AuthenticationToken and file-based AuthenticationToken");
- } else if (null != token) {
- out.writeUTF(token.getClass().getCanonicalName());
- out.writeUTF(Base64.encodeBase64String(AuthenticationTokenSerializer.serialize(token)));
- } else {
- out.writeUTF(tokenFile);
- }
- }
-
- out.writeBoolean(null != instanceName);
- if (null != instanceName) {
- out.writeUTF(instanceName);
- }
-
- out.writeBoolean(null != zooKeepers);
- if (null != zooKeepers) {
- out.writeUTF(zooKeepers);
- }
-
- out.writeBoolean(null != iterators);
- if (null != iterators) {
- out.writeInt(iterators.size());
- for (IteratorSetting iterator : iterators) {
- iterator.write(out);
- }
- }
-
- out.writeBoolean(null != level);
- if (null != level) {
- out.writeInt(level.toInt());
- }
}
@Override
public String toString() {
StringBuilder sb = new StringBuilder(256);
- sb.append("Range: ").append(range);
+ sb.append("RangeInputSplit:");
+ sb.append(" Range: ").append(range);
sb.append(" Locations: ").append(Arrays.asList(locations));
sb.append(" Table: ").append(tableName);
sb.append(" TableID: ").append(tableId);
@@ -356,123 +144,12 @@ public class RangeInputSplit extends InputSplit implements Writable {
return sb.toString();
}
- /**
- * Use {@link #getTableName}
- *
- * @deprecated since 1.6.1, use getTableName() instead.
- */
- @Deprecated
- public String getTable() {
- return getTableName();
- }
-
- public String getTableName() {
- return tableName;
- }
-
- /**
- * Use {@link #setTableName}
- *
- * @deprecated since 1.6.1, use setTableName() instead.
- */
- @Deprecated
- public void setTable(String table) {
- setTableName(table);
- }
-
- public void setTableName(String table) {
- this.tableName = table;
- }
-
- public void setTableId(String tableId) {
- this.tableId = tableId;
- }
-
- public String getTableId() {
- return tableId;
- }
-
- /**
- * @see #getInstance(ClientConfiguration)
- * @deprecated since 1.7.0, use getInstance(ClientConfiguration) instead.
- */
- @Deprecated
- public Instance getInstance() {
- return getInstance(ClientConfiguration.loadDefault());
- }
-
- public Instance getInstance(ClientConfiguration base) {
- if (null == instanceName) {
- return null;
- }
-
- if (isMockInstance()) {
- return new MockInstance(getInstanceName());
- }
-
- if (null == zooKeepers) {
- return null;
- }
-
- return new ZooKeeperInstance(base.withInstance(getInstanceName()).withZkHosts(getZooKeepers()));
- }
-
- public String getInstanceName() {
- return instanceName;
- }
-
- public void setInstanceName(String instanceName) {
- this.instanceName = instanceName;
- }
-
- public String getZooKeepers() {
- return zooKeepers;
- }
-
- public void setZooKeepers(String zooKeepers) {
- this.zooKeepers = zooKeepers;
- }
-
- public String getPrincipal() {
- return principal;
- }
-
- public void setPrincipal(String principal) {
- this.principal = principal;
- }
-
- public AuthenticationToken getToken() {
- return token;
- }
-
- public void setToken(AuthenticationToken token) {
- this.tokenSource = TokenSource.INLINE;
- this.token = token;
- }
-
- public void setToken(String tokenFile) {
- this.tokenSource = TokenSource.FILE;
- this.tokenFile = tokenFile;
- }
-
- public Boolean isOffline() {
- return offline;
- }
-
- public void setOffline(Boolean offline) {
- this.offline = offline;
- }
-
- public void setLocations(String[] locations) {
- this.locations = Arrays.copyOf(locations, locations.length);
- }
-
- public Boolean isMockInstance() {
- return mockInstance;
+ public Range getRange() {
+ return range;
}
- public void setMockInstance(Boolean mockInstance) {
- this.mockInstance = mockInstance;
+ public void setRange(Range range) {
+ this.range = range;
}
public Boolean isIsolatedScan() {
@@ -483,16 +160,12 @@ public class RangeInputSplit extends InputSplit implements Writable {
this.isolatedScan = isolatedScan;
}
- public Authorizations getAuths() {
- return auths;
- }
-
- public void setAuths(Authorizations auths) {
- this.auths = auths;
+ public Boolean isOffline() {
+ return offline;
}
- public void setRange(Range range) {
- this.range = range;
+ public void setOffline(Boolean offline) {
+ this.offline = offline;
}
public Boolean usesLocalIterators() {
@@ -502,35 +175,4 @@ public class RangeInputSplit extends InputSplit implements Writable {
public void setUsesLocalIterators(Boolean localIterators) {
this.localIterators = localIterators;
}
-
- public Set<Pair<Text,Text>> getFetchedColumns() {
- return fetchedColumns;
- }
-
- public void setFetchedColumns(Collection<Pair<Text,Text>> fetchedColumns) {
- this.fetchedColumns = new HashSet<Pair<Text,Text>>();
- for (Pair<Text,Text> columns : fetchedColumns) {
- this.fetchedColumns.add(columns);
- }
- }
-
- public void setFetchedColumns(Set<Pair<Text,Text>> fetchedColumns) {
- this.fetchedColumns = fetchedColumns;
- }
-
- public List<IteratorSetting> getIterators() {
- return iterators;
- }
-
- public void setIterators(List<IteratorSetting> iterators) {
- this.iterators = iterators;
- }
-
- public Level getLogLevel() {
- return level;
- }
-
- public void setLogLevel(Level level) {
- this.level = level;
- }
}
http://git-wip-us.apache.org/repos/asf/accumulo/blob/956a50ec/core/src/main/java/org/apache/accumulo/core/client/mapreduce/impl/AccumuloInputSplit.java
----------------------------------------------------------------------
diff --git a/core/src/main/java/org/apache/accumulo/core/client/mapreduce/impl/AccumuloInputSplit.java b/core/src/main/java/org/apache/accumulo/core/client/mapreduce/impl/AccumuloInputSplit.java
new file mode 100644
index 0000000..94d0026
--- /dev/null
+++ b/core/src/main/java/org/apache/accumulo/core/client/mapreduce/impl/AccumuloInputSplit.java
@@ -0,0 +1,445 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.accumulo.core.client.mapreduce.impl;
+
+import static java.nio.charset.StandardCharsets.UTF_8;
+
+import java.io.DataInput;
+import java.io.DataOutput;
+import java.io.IOException;
+import java.math.BigInteger;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.Collection;
+import java.util.HashSet;
+import java.util.List;
+import java.util.Set;
+
+import org.apache.accumulo.core.client.ClientConfiguration;
+import org.apache.accumulo.core.client.Instance;
+import org.apache.accumulo.core.client.IteratorSetting;
+import org.apache.accumulo.core.client.ZooKeeperInstance;
+import org.apache.accumulo.core.client.mapreduce.InputTableConfig;
+import org.apache.accumulo.core.client.mapreduce.lib.impl.ConfiguratorBase.TokenSource;
+import org.apache.accumulo.core.client.mapreduce.lib.impl.InputConfigurator;
+import org.apache.accumulo.core.client.mock.MockInstance;
+import org.apache.accumulo.core.client.security.tokens.AuthenticationToken;
+import org.apache.accumulo.core.client.security.tokens.AuthenticationToken.AuthenticationTokenSerializer;
+import org.apache.accumulo.core.data.ByteSequence;
+import org.apache.accumulo.core.data.Range;
+import org.apache.accumulo.core.security.Authorizations;
+import org.apache.accumulo.core.util.Base64;
+import org.apache.accumulo.core.util.Pair;
+import org.apache.accumulo.core.data.Key;
+import org.apache.hadoop.io.Text;
+import org.apache.hadoop.io.Writable;
+import org.apache.hadoop.mapreduce.InputSplit;
+import org.apache.log4j.Level;
+
+/**
+ * Abstracts over configurations common to all InputSplits. Specifically it leaves out methods
+ * related to number of ranges and locations per InputSplit as those vary by implementation.
+ *
+ * @see org.apache.accumulo.core.client.mapreduce.RangeInputSplit
+ * @see org.apache.accumulo.core.client.mapreduce.impl.BatchInputSplit
+ */
+public abstract class AccumuloInputSplit extends InputSplit implements Writable {
+ protected String[] locations;
+ protected String tableId, tableName, instanceName, zooKeepers, principal;
+ protected TokenSource tokenSource;
+ protected String tokenFile;
+ protected AuthenticationToken token;
+ protected Boolean mockInstance;
+ protected Authorizations auths;
+ protected Set<Pair<Text,Text>> fetchedColumns;
+ protected List<IteratorSetting> iterators;
+ protected Level level;
+
+ public abstract float getProgress(Key currentKey);
+
+ public AccumuloInputSplit() {
+ locations = new String[0];
+ tableName = "";
+ tableId = "";
+ }
+
+ public AccumuloInputSplit(AccumuloInputSplit split) throws IOException {
+ this.setLocations(split.getLocations());
+ this.setTableName(split.getTableName());
+ this.setTableId(split.getTableId());
+ }
+
+ protected AccumuloInputSplit(String table, String tableId, String[] locations) {
+ setLocations(locations);
+ this.tableName = table;
+ this.tableId = tableId;
+ }
+
+ /**
+ * Central place to set common split configuration not handled by split constructors.
+ * The intention is to make it harder to miss optional setters in future refactor.
+ */
+ public static void updateSplit(AccumuloInputSplit split, Instance instance, InputTableConfig tableConfig,
+ String principal, AuthenticationToken token, Authorizations auths, Level logLevel) {
+ split.setInstanceName(instance.getInstanceName());
+ split.setZooKeepers(instance.getZooKeepers());
+ split.setMockInstance(instance instanceof MockInstance);
+
+ split.setPrincipal(principal);
+ split.setToken(token);
+ split.setAuths(auths);
+
+ split.setFetchedColumns(tableConfig.getFetchedColumns());
+ split.setIterators(tableConfig.getIterators());
+ split.setLogLevel(logLevel);
+ }
+
+ private static byte[] extractBytes(ByteSequence seq, int numBytes) {
+ byte[] bytes = new byte[numBytes + 1];
+ bytes[0] = 0;
+ for (int i = 0; i < numBytes; i++) {
+ if (i >= seq.length())
+ bytes[i + 1] = 0;
+ else
+ bytes[i + 1] = seq.byteAt(i);
+ }
+ return bytes;
+ }
+
+ public static float getProgress(ByteSequence start, ByteSequence end, ByteSequence position) {
+ int maxDepth = Math.min(Math.max(end.length(), start.length()), position.length());
+ BigInteger startBI = new BigInteger(extractBytes(start, maxDepth));
+ BigInteger endBI = new BigInteger(extractBytes(end, maxDepth));
+ BigInteger positionBI = new BigInteger(extractBytes(position, maxDepth));
+ return (float) (positionBI.subtract(startBI).doubleValue() / endBI.subtract(startBI).doubleValue());
+ }
+
+ public long getRangeLength(Range range) throws IOException {
+ Text startRow = range.isInfiniteStartKey() ? new Text(new byte[] {Byte.MIN_VALUE}) : range.getStartKey().getRow();
+ Text stopRow = range.isInfiniteStopKey() ? new Text(new byte[] {Byte.MAX_VALUE}) : range.getEndKey().getRow();
+ int maxCommon = Math.min(7, Math.min(startRow.getLength(), stopRow.getLength()));
+ long diff = 0;
+
+ byte[] start = startRow.getBytes();
+ byte[] stop = stopRow.getBytes();
+ for (int i = 0; i < maxCommon; ++i) {
+ diff |= 0xff & (start[i] ^ stop[i]);
+ diff <<= Byte.SIZE;
+ }
+
+ if (startRow.getLength() != stopRow.getLength())
+ diff |= 0xff;
+
+ return diff + 1;
+ }
+
+ @Override
+ public String[] getLocations() throws IOException {
+ return Arrays.copyOf(locations, locations.length);
+ }
+
+ @Override
+ public void readFields(DataInput in) throws IOException {
+ tableName = in.readUTF();
+ tableId = in.readUTF();
+ int numLocs = in.readInt();
+ locations = new String[numLocs];
+ for (int i = 0; i < numLocs; ++i)
+ locations[i] = in.readUTF();
+
+ if (in.readBoolean()) {
+ mockInstance = in.readBoolean();
+ }
+
+ if (in.readBoolean()) {
+ int numColumns = in.readInt();
+ List<String> columns = new ArrayList<String>(numColumns);
+ for (int i = 0; i < numColumns; i++) {
+ columns.add(in.readUTF());
+ }
+
+ fetchedColumns = InputConfigurator.deserializeFetchedColumns(columns);
+ }
+
+ if (in.readBoolean()) {
+ String strAuths = in.readUTF();
+ auths = new Authorizations(strAuths.getBytes(UTF_8));
+ }
+
+ if (in.readBoolean()) {
+ principal = in.readUTF();
+ }
+
+ if (in.readBoolean()) {
+ int ordinal = in.readInt();
+ this.tokenSource = TokenSource.values()[ordinal];
+
+ switch (this.tokenSource) {
+ case INLINE:
+ String tokenClass = in.readUTF();
+ byte[] base64TokenBytes = in.readUTF().getBytes(UTF_8);
+ byte[] tokenBytes = Base64.decodeBase64(base64TokenBytes);
+
+ this.token = AuthenticationTokenSerializer.deserialize(tokenClass, tokenBytes);
+ break;
+
+ case FILE:
+ this.tokenFile = in.readUTF();
+
+ break;
+ default:
+ throw new IOException("Cannot parse unknown TokenSource ordinal");
+ }
+ }
+
+ if (in.readBoolean()) {
+ instanceName = in.readUTF();
+ }
+
+ if (in.readBoolean()) {
+ zooKeepers = in.readUTF();
+ }
+
+ if (in.readBoolean()) {
+ int numIterators = in.readInt();
+ iterators = new ArrayList<IteratorSetting>(numIterators);
+ for (int i = 0; i < numIterators; i++) {
+ iterators.add(new IteratorSetting(in));
+ }
+ }
+
+ if (in.readBoolean()) {
+ level = Level.toLevel(in.readInt());
+ }
+ }
+
+ @Override
+ public void write(DataOutput out) throws IOException {
+ out.writeUTF(tableName);
+ out.writeUTF(tableId);
+ out.writeInt(locations.length);
+ for (int i = 0; i < locations.length; ++i)
+ out.writeUTF(locations[i]);
+
+ out.writeBoolean(null != mockInstance);
+ if (null != mockInstance) {
+ out.writeBoolean(mockInstance);
+ }
+
+ out.writeBoolean(null != fetchedColumns);
+ if (null != fetchedColumns) {
+ String[] cols = InputConfigurator.serializeColumns(fetchedColumns);
+ out.writeInt(cols.length);
+ for (String col : cols) {
+ out.writeUTF(col);
+ }
+ }
+
+ out.writeBoolean(null != auths);
+ if (null != auths) {
+ out.writeUTF(auths.serialize());
+ }
+
+ out.writeBoolean(null != principal);
+ if (null != principal) {
+ out.writeUTF(principal);
+ }
+
+ out.writeBoolean(null != tokenSource);
+ if (null != tokenSource) {
+ out.writeInt(tokenSource.ordinal());
+
+ if (null != token && null != tokenFile) {
+ throw new IOException("Cannot use both inline AuthenticationToken and file-based AuthenticationToken");
+ } else if (null != token) {
+ out.writeUTF(token.getClass().getCanonicalName());
+ out.writeUTF(Base64.encodeBase64String(AuthenticationTokenSerializer.serialize(token)));
+ } else {
+ out.writeUTF(tokenFile);
+ }
+ }
+
+ out.writeBoolean(null != instanceName);
+ if (null != instanceName) {
+ out.writeUTF(instanceName);
+ }
+
+ out.writeBoolean(null != zooKeepers);
+ if (null != zooKeepers) {
+ out.writeUTF(zooKeepers);
+ }
+
+ out.writeBoolean(null != iterators);
+ if (null != iterators) {
+ out.writeInt(iterators.size());
+ for (IteratorSetting iterator : iterators) {
+ iterator.write(out);
+ }
+ }
+
+ out.writeBoolean(null != level);
+ if (null != level) {
+ out.writeInt(level.toInt());
+ }
+ }
+
+ /**
+ * Use {@link #getTableName}
+ */
+ @Deprecated
+ public String getTable() {
+ return getTableName();
+ }
+
+ public String getTableName() {
+ return tableName;
+ }
+
+ /**
+ * Use {@link #setTableName}
+ */
+ @Deprecated
+ public void setTable(String table) {
+ setTableName(table);
+ }
+
+ public void setTableName(String table) {
+ this.tableName = table;
+ }
+
+ public void setTableId(String tableId) {
+ this.tableId = tableId;
+ }
+
+ public String getTableId() {
+ return tableId;
+ }
+
+ /**
+ * @see #getInstance(ClientConfiguration)
+ */
+ @Deprecated
+ public Instance getInstance() {
+ return getInstance(ClientConfiguration.loadDefault());
+ }
+
+ public Instance getInstance(ClientConfiguration base) {
+ if (null == instanceName) {
+ return null;
+ }
+
+ if (isMockInstance()) {
+ return new MockInstance(getInstanceName());
+ }
+
+ if (null == zooKeepers) {
+ return null;
+ }
+
+ return new ZooKeeperInstance(base.withInstance(getInstanceName()).withZkHosts(getZooKeepers()));
+ }
+
+ public String getInstanceName() {
+ return instanceName;
+ }
+
+ public void setInstanceName(String instanceName) {
+ this.instanceName = instanceName;
+ }
+
+ public String getZooKeepers() {
+ return zooKeepers;
+ }
+
+ public void setZooKeepers(String zooKeepers) {
+ this.zooKeepers = zooKeepers;
+ }
+
+ public String getPrincipal() {
+ return principal;
+ }
+
+ public void setPrincipal(String principal) {
+ this.principal = principal;
+ }
+
+ public AuthenticationToken getToken() {
+ return token;
+ }
+
+ public void setToken(AuthenticationToken token) {
+ this.tokenSource = TokenSource.INLINE;
+ this.token = token;
+ }
+
+ public void setToken(String tokenFile) {
+ this.tokenSource = TokenSource.FILE;
+ this.tokenFile = tokenFile;
+ }
+
+ public void setLocations(String[] locations) {
+ this.locations = Arrays.copyOf(locations, locations.length);
+ }
+
+ public Boolean isMockInstance() {
+ return mockInstance;
+ }
+
+ public void setMockInstance(Boolean mockInstance) {
+ this.mockInstance = mockInstance;
+ }
+
+ public Authorizations getAuths() {
+ return auths;
+ }
+
+ public void setAuths(Authorizations auths) {
+ this.auths = auths;
+ }
+
+
+ public Set<Pair<Text,Text>> getFetchedColumns() {
+ return fetchedColumns;
+ }
+
+ public void setFetchedColumns(Collection<Pair<Text,Text>> fetchedColumns) {
+ this.fetchedColumns = new HashSet<Pair<Text,Text>>();
+ for (Pair<Text,Text> columns : fetchedColumns) {
+ this.fetchedColumns.add(columns);
+ }
+ }
+
+ public void setFetchedColumns(Set<Pair<Text,Text>> fetchedColumns) {
+ this.fetchedColumns = fetchedColumns;
+ }
+
+ public List<IteratorSetting> getIterators() {
+ return iterators;
+ }
+
+ public void setIterators(List<IteratorSetting> iterators) {
+ this.iterators = iterators;
+ }
+
+ public Level getLogLevel() {
+ return level;
+ }
+
+ public void setLogLevel(Level level) {
+ this.level = level;
+ }
+}