Class InputFormatBase<K,V>

java.lang.Object
org.apache.hadoop.mapreduce.InputFormat<K,V>
org.apache.accumulo.core.client.mapreduce.AbstractInputFormat<K,V>
org.apache.accumulo.core.client.mapreduce.InputFormatBase<K,V>
Direct Known Subclasses:
AccumuloInputFormat, AccumuloRowInputFormat

public abstract class InputFormatBase<K,V> extends AbstractInputFormat<K,V>
This abstract InputFormat class allows MapReduce jobs to use Accumulo as the source of K,V pairs.

Subclasses must implement a InputFormat.createRecordReader(InputSplit, TaskAttemptContext) to provide a RecordReader for K,V.

A static base class, RecordReaderBase, is provided to retrieve Accumulo Key/Value pairs, but one must implement its RecordReader.nextKeyValue() to transform them to the desired generic types K,V.

See AccumuloInputFormat for an example implementation.

  • Constructor Details

    • InputFormatBase

      public InputFormatBase()
  • Method Details

    • getInputTableName

      protected static String getInputTableName(org.apache.hadoop.mapreduce.JobContext context)
      Gets the table name from the configuration.
      Parameters:
      context - the Hadoop context for the configured job
      Returns:
      the table name
      Since:
      1.5.0
      See Also:
    • setInputTableName

      public static void setInputTableName(org.apache.hadoop.mapreduce.Job job, String tableName)
      Sets the name of the input table, over which this job will scan.
      Parameters:
      job - the Hadoop job instance to be configured
      tableName - the table to use when the tablename is null in the write call
      Since:
      1.5.0
    • setRanges

      public static void setRanges(org.apache.hadoop.mapreduce.Job job, Collection<Range> ranges)
      Sets the input ranges to scan for the single input table associated with this job.
      Parameters:
      job - the Hadoop job instance to be configured
      ranges - the ranges that will be mapped over
      Since:
      1.5.0
    • getRanges

      protected static List<Range> getRanges(org.apache.hadoop.mapreduce.JobContext context) throws IOException
      Gets the ranges to scan over from a job.
      Parameters:
      context - the Hadoop context for the configured job
      Returns:
      the ranges
      Throws:
      IOException
      Since:
      1.5.0
      See Also:
    • fetchColumns

      public static void fetchColumns(org.apache.hadoop.mapreduce.Job job, Collection<org.apache.accumulo.core.util.Pair<org.apache.hadoop.io.Text,org.apache.hadoop.io.Text>> columnFamilyColumnQualifierPairs)
      Restricts the columns that will be mapped over for this job for the default input table.
      Parameters:
      job - the Hadoop job instance to be configured
      columnFamilyColumnQualifierPairs - a pair of Text objects corresponding to column family and column qualifier. If the column qualifier is null, the entire column family is selected. An empty set is the default and is equivalent to scanning the all columns.
      Since:
      1.5.0
    • getFetchedColumns

      protected static Set<org.apache.accumulo.core.util.Pair<org.apache.hadoop.io.Text,org.apache.hadoop.io.Text>> getFetchedColumns(org.apache.hadoop.mapreduce.JobContext context)
      Gets the columns to be mapped over from this job.
      Parameters:
      context - the Hadoop context for the configured job
      Returns:
      a set of columns
      Since:
      1.5.0
      See Also:
    • addIterator

      public static void addIterator(org.apache.hadoop.mapreduce.Job job, IteratorSetting cfg)
      Encode an iterator on the single input table for this job.
      Parameters:
      job - the Hadoop job instance to be configured
      cfg - the configuration of the iterator
      Since:
      1.5.0
    • getIterators

      protected static List<IteratorSetting> getIterators(org.apache.hadoop.mapreduce.JobContext context)
      Gets a list of the iterator settings (for iterators to apply to a scanner) from this configuration.
      Parameters:
      context - the Hadoop context for the configured job
      Returns:
      a list of iterators
      Since:
      1.5.0
      See Also:
    • setAutoAdjustRanges

      public static void setAutoAdjustRanges(org.apache.hadoop.mapreduce.Job job, boolean enableFeature)
      Controls the automatic adjustment of ranges for this job. This feature merges overlapping ranges, then splits them to align with tablet boundaries. Disabling this feature will cause exactly one Map task to be created for each specified range. The default setting is enabled. *

      By default, this feature is enabled.

      Parameters:
      job - the Hadoop job instance to be configured
      enableFeature - the feature is enabled if true, disabled otherwise
      Since:
      1.5.0
      See Also:
    • getAutoAdjustRanges

      protected static boolean getAutoAdjustRanges(org.apache.hadoop.mapreduce.JobContext context)
      Determines whether a configuration has auto-adjust ranges enabled. Must be enabled when setBatchScan(Job, boolean) is true.
      Parameters:
      context - the Hadoop context for the configured job
      Returns:
      false if the feature is disabled, true otherwise
      Since:
      1.5.0
      See Also:
    • setScanIsolation

      public static void setScanIsolation(org.apache.hadoop.mapreduce.Job job, boolean enableFeature)
      Controls the use of the IsolatedScanner in this job.

      By default, this feature is disabled.

      Parameters:
      job - the Hadoop job instance to be configured
      enableFeature - the feature is enabled if true, disabled otherwise
      Since:
      1.5.0
    • isIsolated

      protected static boolean isIsolated(org.apache.hadoop.mapreduce.JobContext context)
      Determines whether a configuration has isolation enabled.
      Parameters:
      context - the Hadoop context for the configured job
      Returns:
      true if the feature is enabled, false otherwise
      Since:
      1.5.0
      See Also:
    • setLocalIterators

      public static void setLocalIterators(org.apache.hadoop.mapreduce.Job job, boolean enableFeature)
      Controls the use of the ClientSideIteratorScanner in this job. Enabling this feature will cause the iterator stack to be constructed within the Map task, rather than within the Accumulo TServer. To use this feature, all classes needed for those iterators must be available on the classpath for the task.

      By default, this feature is disabled.

      Parameters:
      job - the Hadoop job instance to be configured
      enableFeature - the feature is enabled if true, disabled otherwise
      Since:
      1.5.0
    • usesLocalIterators

      protected static boolean usesLocalIterators(org.apache.hadoop.mapreduce.JobContext context)
      Determines whether a configuration uses local iterators.
      Parameters:
      context - the Hadoop context for the configured job
      Returns:
      true if the feature is enabled, false otherwise
      Since:
      1.5.0
      See Also:
    • setOfflineTableScan

      public static void setOfflineTableScan(org.apache.hadoop.mapreduce.Job job, boolean enableFeature)
      Enable reading offline tables. By default, this feature is disabled and only online tables are scanned. This will make the map reduce job directly read the table's files. If the table is not offline, then the job will fail. If the table comes online during the map reduce job, it is likely that the job will fail.

      To use this option, the map reduce user will need access to read the Accumulo directory in HDFS.

      Reading the offline table will create the scan time iterator stack in the map process. So any iterators that are configured for the table will need to be on the mapper's classpath.

      One way to use this feature is to clone a table, take the clone offline, and use the clone as the input table for a map reduce job. If you plan to map reduce over the data many times, it may be better to the compact the table, clone it, take it offline, and use the clone for all map reduce jobs. The reason to do this is that compaction will reduce each tablet in the table to one file, and it is faster to read from one file.

      There are two possible advantages to reading a tables file directly out of HDFS. First, you may see better read performance. Second, it will support speculative execution better. When reading an online table speculative execution can put more load on an already slow tablet server.

      By default, this feature is disabled.

      Parameters:
      job - the Hadoop job instance to be configured
      enableFeature - the feature is enabled if true, disabled otherwise
      Since:
      1.5.0
    • isOfflineScan

      protected static boolean isOfflineScan(org.apache.hadoop.mapreduce.JobContext context)
      Determines whether a configuration has the offline table scan feature enabled.
      Parameters:
      context - the Hadoop context for the configured job
      Returns:
      true if the feature is enabled, false otherwise
      Since:
      1.5.0
      See Also:
    • setBatchScan

      public static void setBatchScan(org.apache.hadoop.mapreduce.Job job, boolean enableFeature)
      Controls the use of the 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)

      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.

      Note: that the value of setAutoAdjustRanges(Job, boolean) is ignored and is assumed to be true when BatchScan option is enabled.

      This configuration is incompatible with:

      By default, this feature is disabled.

      Parameters:
      job - the Hadoop job instance to be configured
      enableFeature - the feature is enabled if true, disabled otherwise
      Since:
      1.7.0
    • isBatchScan

      public static boolean isBatchScan(org.apache.hadoop.mapreduce.JobContext context)
      Determines whether a configuration has the BatchScanner feature enabled.
      Parameters:
      context - the Hadoop context for the configured job
      Since:
      1.7.0
      See Also:
    • setSamplerConfiguration

      public static void setSamplerConfiguration(org.apache.hadoop.mapreduce.Job job, SamplerConfiguration samplerConfig)
      Causes input format to read sample data. If sample data was created using a different configuration or a tables sampler configuration changes while reading data, then the input format will throw an error.
      Parameters:
      job - the Hadoop job instance to be configured
      samplerConfig - The sampler configuration that sample must have been created with inorder for reading sample data to succeed.
      Since:
      1.8.0
      See Also:
    • getTabletLocator

      @Deprecated protected static org.apache.accumulo.core.client.impl.TabletLocator getTabletLocator(org.apache.hadoop.mapreduce.JobContext context) throws TableNotFoundException
      Deprecated.
      since 1.6.0
      Initializes an Accumulo TabletLocator based on the configuration.
      Parameters:
      context - the Hadoop context for the configured job
      Returns:
      an Accumulo tablet locator
      Throws:
      TableNotFoundException - if the table name set on the configuration doesn't exist
      Since:
      1.5.0