You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hbase.apache.org by gi...@apache.org on 2017/06/28 15:01:22 UTC

[18/38] hbase-site git commit: Published site at 82d554e3783372cc6b05489452c815b57c06f6cd.

http://git-wip-us.apache.org/repos/asf/hbase-site/blob/2848f431/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html
----------------------------------------------------------------------
diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html
index e6ae33a..885e6f8 100644
--- a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html
+++ b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html
@@ -30,823 +30,874 @@
 <span class="sourceLineNo">022</span>import java.net.InetSocketAddress;<a name="line.22"></a>
 <span class="sourceLineNo">023</span>import java.net.URLDecoder;<a name="line.23"></a>
 <span class="sourceLineNo">024</span>import java.net.URLEncoder;<a name="line.24"></a>
-<span class="sourceLineNo">025</span>import java.util.ArrayList;<a name="line.25"></a>
-<span class="sourceLineNo">026</span>import java.util.Collection;<a name="line.26"></a>
-<span class="sourceLineNo">027</span>import java.util.List;<a name="line.27"></a>
-<span class="sourceLineNo">028</span>import java.util.Map;<a name="line.28"></a>
-<span class="sourceLineNo">029</span>import java.util.TreeMap;<a name="line.29"></a>
-<span class="sourceLineNo">030</span>import java.util.TreeSet;<a name="line.30"></a>
-<span class="sourceLineNo">031</span>import java.util.UUID;<a name="line.31"></a>
-<span class="sourceLineNo">032</span><a name="line.32"></a>
-<span class="sourceLineNo">033</span>import org.apache.commons.logging.Log;<a name="line.33"></a>
-<span class="sourceLineNo">034</span>import org.apache.commons.logging.LogFactory;<a name="line.34"></a>
-<span class="sourceLineNo">035</span>import org.apache.hadoop.conf.Configuration;<a name="line.35"></a>
-<span class="sourceLineNo">036</span>import org.apache.hadoop.fs.FileSystem;<a name="line.36"></a>
-<span class="sourceLineNo">037</span>import org.apache.hadoop.fs.Path;<a name="line.37"></a>
-<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.Cell;<a name="line.38"></a>
-<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.CellComparator;<a name="line.39"></a>
-<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.CellUtil;<a name="line.40"></a>
-<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.HColumnDescriptor;<a name="line.41"></a>
-<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.HConstants;<a name="line.42"></a>
-<span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.HRegionLocation;<a name="line.43"></a>
-<span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.HTableDescriptor;<a name="line.44"></a>
-<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.KeyValue;<a name="line.45"></a>
-<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.KeyValueUtil;<a name="line.46"></a>
-<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.TableName;<a name="line.47"></a>
-<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.classification.InterfaceAudience;<a name="line.48"></a>
-<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.49"></a>
-<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.50"></a>
-<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.client.Put;<a name="line.51"></a>
-<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.client.RegionLocator;<a name="line.52"></a>
-<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.client.Table;<a name="line.53"></a>
-<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.fs.HFileSystem;<a name="line.54"></a>
-<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.55"></a>
-<span class="sourceLineNo">056</span>import org.apache.hadoop.hbase.io.compress.Compression;<a name="line.56"></a>
-<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.io.compress.Compression.Algorithm;<a name="line.57"></a>
-<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.58"></a>
-<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.io.hfile.CacheConfig;<a name="line.59"></a>
-<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.io.hfile.HFile;<a name="line.60"></a>
-<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.io.hfile.HFileContext;<a name="line.61"></a>
-<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.io.hfile.HFileContextBuilder;<a name="line.62"></a>
-<span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.io.hfile.HFileWriterImpl;<a name="line.63"></a>
-<span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.64"></a>
-<span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.regionserver.HStore;<a name="line.65"></a>
-<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.regionserver.StoreFile;<a name="line.66"></a>
-<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.regionserver.StoreFileWriter;<a name="line.67"></a>
-<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.68"></a>
-<span class="sourceLineNo">069</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.69"></a>
-<span class="sourceLineNo">070</span>import org.apache.hadoop.hbase.util.FSUtils;<a name="line.70"></a>
-<span class="sourceLineNo">071</span>import org.apache.hadoop.io.NullWritable;<a name="line.71"></a>
-<span class="sourceLineNo">072</span>import org.apache.hadoop.io.SequenceFile;<a name="line.72"></a>
-<span class="sourceLineNo">073</span>import org.apache.hadoop.io.Text;<a name="line.73"></a>
-<span class="sourceLineNo">074</span>import org.apache.hadoop.mapreduce.Job;<a name="line.74"></a>
-<span class="sourceLineNo">075</span>import org.apache.hadoop.mapreduce.OutputFormat;<a name="line.75"></a>
-<span class="sourceLineNo">076</span>import org.apache.hadoop.mapreduce.RecordWriter;<a name="line.76"></a>
-<span class="sourceLineNo">077</span>import org.apache.hadoop.mapreduce.TaskAttemptContext;<a name="line.77"></a>
-<span class="sourceLineNo">078</span>import org.apache.hadoop.mapreduce.lib.output.FileOutputCommitter;<a name="line.78"></a>
-<span class="sourceLineNo">079</span>import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;<a name="line.79"></a>
-<span class="sourceLineNo">080</span>import org.apache.hadoop.mapreduce.lib.partition.TotalOrderPartitioner;<a name="line.80"></a>
-<span class="sourceLineNo">081</span><a name="line.81"></a>
-<span class="sourceLineNo">082</span>import com.google.common.annotations.VisibleForTesting;<a name="line.82"></a>
-<span class="sourceLineNo">083</span><a name="line.83"></a>
-<span class="sourceLineNo">084</span>/**<a name="line.84"></a>
-<span class="sourceLineNo">085</span> * Writes HFiles. Passed Cells must arrive in order.<a name="line.85"></a>
-<span class="sourceLineNo">086</span> * Writes current time as the sequence id for the file. Sets the major compacted<a name="line.86"></a>
-<span class="sourceLineNo">087</span> * attribute on created @{link {@link HFile}s. Calling write(null,null) will forcibly roll<a name="line.87"></a>
-<span class="sourceLineNo">088</span> * all HFiles being written.<a name="line.88"></a>
-<span class="sourceLineNo">089</span> * &lt;p&gt;<a name="line.89"></a>
-<span class="sourceLineNo">090</span> * Using this class as part of a MapReduce job is best done<a name="line.90"></a>
-<span class="sourceLineNo">091</span> * using {@link #configureIncrementalLoad(Job, HTableDescriptor, RegionLocator, Class)}.<a name="line.91"></a>
-<span class="sourceLineNo">092</span> */<a name="line.92"></a>
-<span class="sourceLineNo">093</span>@InterfaceAudience.Public<a name="line.93"></a>
-<span class="sourceLineNo">094</span>public class HFileOutputFormat2<a name="line.94"></a>
-<span class="sourceLineNo">095</span>    extends FileOutputFormat&lt;ImmutableBytesWritable, Cell&gt; {<a name="line.95"></a>
-<span class="sourceLineNo">096</span>  private static final Log LOG = LogFactory.getLog(HFileOutputFormat2.class);<a name="line.96"></a>
-<span class="sourceLineNo">097</span><a name="line.97"></a>
-<span class="sourceLineNo">098</span>  // The following constants are private since these are used by<a name="line.98"></a>
-<span class="sourceLineNo">099</span>  // HFileOutputFormat2 to internally transfer data between job setup and<a name="line.99"></a>
-<span class="sourceLineNo">100</span>  // reducer run using conf.<a name="line.100"></a>
-<span class="sourceLineNo">101</span>  // These should not be changed by the client.<a name="line.101"></a>
-<span class="sourceLineNo">102</span>  private static final String COMPRESSION_FAMILIES_CONF_KEY =<a name="line.102"></a>
-<span class="sourceLineNo">103</span>      "hbase.hfileoutputformat.families.compression";<a name="line.103"></a>
-<span class="sourceLineNo">104</span>  private static final String BLOOM_TYPE_FAMILIES_CONF_KEY =<a name="line.104"></a>
-<span class="sourceLineNo">105</span>      "hbase.hfileoutputformat.families.bloomtype";<a name="line.105"></a>
-<span class="sourceLineNo">106</span>  private static final String BLOCK_SIZE_FAMILIES_CONF_KEY =<a name="line.106"></a>
-<span class="sourceLineNo">107</span>      "hbase.mapreduce.hfileoutputformat.blocksize";<a name="line.107"></a>
-<span class="sourceLineNo">108</span>  private static final String DATABLOCK_ENCODING_FAMILIES_CONF_KEY =<a name="line.108"></a>
-<span class="sourceLineNo">109</span>      "hbase.mapreduce.hfileoutputformat.families.datablock.encoding";<a name="line.109"></a>
-<span class="sourceLineNo">110</span><a name="line.110"></a>
-<span class="sourceLineNo">111</span>  // This constant is public since the client can modify this when setting<a name="line.111"></a>
-<span class="sourceLineNo">112</span>  // up their conf object and thus refer to this symbol.<a name="line.112"></a>
-<span class="sourceLineNo">113</span>  // It is present for backwards compatibility reasons. Use it only to<a name="line.113"></a>
-<span class="sourceLineNo">114</span>  // override the auto-detection of datablock encoding.<a name="line.114"></a>
-<span class="sourceLineNo">115</span>  public static final String DATABLOCK_ENCODING_OVERRIDE_CONF_KEY =<a name="line.115"></a>
-<span class="sourceLineNo">116</span>      "hbase.mapreduce.hfileoutputformat.datablock.encoding";<a name="line.116"></a>
-<span class="sourceLineNo">117</span><a name="line.117"></a>
-<span class="sourceLineNo">118</span>  /**<a name="line.118"></a>
-<span class="sourceLineNo">119</span>   * Keep locality while generating HFiles for bulkload. See HBASE-12596<a name="line.119"></a>
-<span class="sourceLineNo">120</span>   */<a name="line.120"></a>
-<span class="sourceLineNo">121</span>  public static final String LOCALITY_SENSITIVE_CONF_KEY =<a name="line.121"></a>
-<span class="sourceLineNo">122</span>      "hbase.bulkload.locality.sensitive.enabled";<a name="line.122"></a>
-<span class="sourceLineNo">123</span>  private static final boolean DEFAULT_LOCALITY_SENSITIVE = true;<a name="line.123"></a>
-<span class="sourceLineNo">124</span>  private static final String OUTPUT_TABLE_NAME_CONF_KEY =<a name="line.124"></a>
-<span class="sourceLineNo">125</span>      "hbase.mapreduce.hfileoutputformat.table.name";<a name="line.125"></a>
-<span class="sourceLineNo">126</span><a name="line.126"></a>
-<span class="sourceLineNo">127</span>  public static final String STORAGE_POLICY_PROPERTY = "hbase.hstore.storagepolicy";<a name="line.127"></a>
-<span class="sourceLineNo">128</span>  public static final String STORAGE_POLICY_PROPERTY_CF_PREFIX = STORAGE_POLICY_PROPERTY + ".";<a name="line.128"></a>
-<span class="sourceLineNo">129</span><a name="line.129"></a>
-<span class="sourceLineNo">130</span>  @Override<a name="line.130"></a>
-<span class="sourceLineNo">131</span>  public RecordWriter&lt;ImmutableBytesWritable, Cell&gt; getRecordWriter(<a name="line.131"></a>
-<span class="sourceLineNo">132</span>      final TaskAttemptContext context) throws IOException, InterruptedException {<a name="line.132"></a>
-<span class="sourceLineNo">133</span>    return createRecordWriter(context);<a name="line.133"></a>
-<span class="sourceLineNo">134</span>  }<a name="line.134"></a>
-<span class="sourceLineNo">135</span><a name="line.135"></a>
-<span class="sourceLineNo">136</span>  static &lt;V extends Cell&gt; RecordWriter&lt;ImmutableBytesWritable, V&gt;<a name="line.136"></a>
-<span class="sourceLineNo">137</span>  createRecordWriter(final TaskAttemptContext context) throws IOException {<a name="line.137"></a>
-<span class="sourceLineNo">138</span>    return new HFileRecordWriter&lt;&gt;(context, null);<a name="line.138"></a>
-<span class="sourceLineNo">139</span>  }<a name="line.139"></a>
+<span class="sourceLineNo">025</span>import java.nio.charset.StandardCharsets;<a name="line.25"></a>
+<span class="sourceLineNo">026</span>import java.util.ArrayList;<a name="line.26"></a>
+<span class="sourceLineNo">027</span>import java.util.Arrays;<a name="line.27"></a>
+<span class="sourceLineNo">028</span>import java.util.Collection;<a name="line.28"></a>
+<span class="sourceLineNo">029</span>import java.util.List;<a name="line.29"></a>
+<span class="sourceLineNo">030</span>import java.util.Map;<a name="line.30"></a>
+<span class="sourceLineNo">031</span>import java.util.Set;<a name="line.31"></a>
+<span class="sourceLineNo">032</span>import java.util.TreeMap;<a name="line.32"></a>
+<span class="sourceLineNo">033</span>import java.util.TreeSet;<a name="line.33"></a>
+<span class="sourceLineNo">034</span>import java.util.UUID;<a name="line.34"></a>
+<span class="sourceLineNo">035</span>import java.util.function.Function;<a name="line.35"></a>
+<span class="sourceLineNo">036</span>import java.util.stream.Collectors;<a name="line.36"></a>
+<span class="sourceLineNo">037</span><a name="line.37"></a>
+<span class="sourceLineNo">038</span>import org.apache.commons.lang.StringUtils;<a name="line.38"></a>
+<span class="sourceLineNo">039</span>import org.apache.commons.logging.Log;<a name="line.39"></a>
+<span class="sourceLineNo">040</span>import org.apache.commons.logging.LogFactory;<a name="line.40"></a>
+<span class="sourceLineNo">041</span>import org.apache.hadoop.conf.Configuration;<a name="line.41"></a>
+<span class="sourceLineNo">042</span>import org.apache.hadoop.fs.FileSystem;<a name="line.42"></a>
+<span class="sourceLineNo">043</span>import org.apache.hadoop.fs.Path;<a name="line.43"></a>
+<span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.Cell;<a name="line.44"></a>
+<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.CellComparator;<a name="line.45"></a>
+<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.CellUtil;<a name="line.46"></a>
+<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.HColumnDescriptor;<a name="line.47"></a>
+<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.HConstants;<a name="line.48"></a>
+<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.HRegionLocation;<a name="line.49"></a>
+<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.HTableDescriptor;<a name="line.50"></a>
+<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.KeyValue;<a name="line.51"></a>
+<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.KeyValueUtil;<a name="line.52"></a>
+<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.TableName;<a name="line.53"></a>
+<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.classification.InterfaceAudience;<a name="line.54"></a>
+<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.55"></a>
+<span class="sourceLineNo">056</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.56"></a>
+<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.client.Put;<a name="line.57"></a>
+<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.client.RegionLocator;<a name="line.58"></a>
+<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.client.Table;<a name="line.59"></a>
+<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.fs.HFileSystem;<a name="line.60"></a>
+<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.61"></a>
+<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.io.compress.Compression;<a name="line.62"></a>
+<span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.io.compress.Compression.Algorithm;<a name="line.63"></a>
+<span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.64"></a>
+<span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.io.hfile.CacheConfig;<a name="line.65"></a>
+<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.io.hfile.HFile;<a name="line.66"></a>
+<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.io.hfile.HFileContext;<a name="line.67"></a>
+<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.io.hfile.HFileContextBuilder;<a name="line.68"></a>
+<span class="sourceLineNo">069</span>import org.apache.hadoop.hbase.io.hfile.HFileWriterImpl;<a name="line.69"></a>
+<span class="sourceLineNo">070</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.70"></a>
+<span class="sourceLineNo">071</span>import org.apache.hadoop.hbase.regionserver.HStore;<a name="line.71"></a>
+<span class="sourceLineNo">072</span>import org.apache.hadoop.hbase.regionserver.StoreFile;<a name="line.72"></a>
+<span class="sourceLineNo">073</span>import org.apache.hadoop.hbase.regionserver.StoreFileWriter;<a name="line.73"></a>
+<span class="sourceLineNo">074</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.74"></a>
+<span class="sourceLineNo">075</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.75"></a>
+<span class="sourceLineNo">076</span>import org.apache.hadoop.hbase.util.FSUtils;<a name="line.76"></a>
+<span class="sourceLineNo">077</span>import org.apache.hadoop.io.NullWritable;<a name="line.77"></a>
+<span class="sourceLineNo">078</span>import org.apache.hadoop.io.SequenceFile;<a name="line.78"></a>
+<span class="sourceLineNo">079</span>import org.apache.hadoop.io.Text;<a name="line.79"></a>
+<span class="sourceLineNo">080</span>import org.apache.hadoop.mapreduce.Job;<a name="line.80"></a>
+<span class="sourceLineNo">081</span>import org.apache.hadoop.mapreduce.OutputFormat;<a name="line.81"></a>
+<span class="sourceLineNo">082</span>import org.apache.hadoop.mapreduce.RecordWriter;<a name="line.82"></a>
+<span class="sourceLineNo">083</span>import org.apache.hadoop.mapreduce.TaskAttemptContext;<a name="line.83"></a>
+<span class="sourceLineNo">084</span>import org.apache.hadoop.mapreduce.lib.output.FileOutputCommitter;<a name="line.84"></a>
+<span class="sourceLineNo">085</span>import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;<a name="line.85"></a>
+<span class="sourceLineNo">086</span>import org.apache.hadoop.mapreduce.lib.partition.TotalOrderPartitioner;<a name="line.86"></a>
+<span class="sourceLineNo">087</span><a name="line.87"></a>
+<span class="sourceLineNo">088</span>import com.google.common.annotations.VisibleForTesting;<a name="line.88"></a>
+<span class="sourceLineNo">089</span><a name="line.89"></a>
+<span class="sourceLineNo">090</span>/**<a name="line.90"></a>
+<span class="sourceLineNo">091</span> * Writes HFiles. Passed Cells must arrive in order.<a name="line.91"></a>
+<span class="sourceLineNo">092</span> * Writes current time as the sequence id for the file. Sets the major compacted<a name="line.92"></a>
+<span class="sourceLineNo">093</span> * attribute on created @{link {@link HFile}s. Calling write(null,null) will forcibly roll<a name="line.93"></a>
+<span class="sourceLineNo">094</span> * all HFiles being written.<a name="line.94"></a>
+<span class="sourceLineNo">095</span> * &lt;p&gt;<a name="line.95"></a>
+<span class="sourceLineNo">096</span> * Using this class as part of a MapReduce job is best done<a name="line.96"></a>
+<span class="sourceLineNo">097</span> * using {@link #configureIncrementalLoad(Job, HTableDescriptor, RegionLocator)}.<a name="line.97"></a>
+<span class="sourceLineNo">098</span> */<a name="line.98"></a>
+<span class="sourceLineNo">099</span>@InterfaceAudience.Public<a name="line.99"></a>
+<span class="sourceLineNo">100</span>public class HFileOutputFormat2<a name="line.100"></a>
+<span class="sourceLineNo">101</span>    extends FileOutputFormat&lt;ImmutableBytesWritable, Cell&gt; {<a name="line.101"></a>
+<span class="sourceLineNo">102</span>  private static final Log LOG = LogFactory.getLog(HFileOutputFormat2.class);<a name="line.102"></a>
+<span class="sourceLineNo">103</span>  static class TableInfo {<a name="line.103"></a>
+<span class="sourceLineNo">104</span>    private HTableDescriptor hTableDescriptor;<a name="line.104"></a>
+<span class="sourceLineNo">105</span>    private RegionLocator regionLocator;<a name="line.105"></a>
+<span class="sourceLineNo">106</span><a name="line.106"></a>
+<span class="sourceLineNo">107</span>    public TableInfo(HTableDescriptor hTableDescriptor, RegionLocator regionLocator) {<a name="line.107"></a>
+<span class="sourceLineNo">108</span>      this.hTableDescriptor = hTableDescriptor;<a name="line.108"></a>
+<span class="sourceLineNo">109</span>      this.regionLocator = regionLocator;<a name="line.109"></a>
+<span class="sourceLineNo">110</span>    }<a name="line.110"></a>
+<span class="sourceLineNo">111</span><a name="line.111"></a>
+<span class="sourceLineNo">112</span>    public HTableDescriptor getHTableDescriptor() {<a name="line.112"></a>
+<span class="sourceLineNo">113</span>      return hTableDescriptor;<a name="line.113"></a>
+<span class="sourceLineNo">114</span>    }<a name="line.114"></a>
+<span class="sourceLineNo">115</span><a name="line.115"></a>
+<span class="sourceLineNo">116</span>    public RegionLocator getRegionLocator() {<a name="line.116"></a>
+<span class="sourceLineNo">117</span>      return regionLocator;<a name="line.117"></a>
+<span class="sourceLineNo">118</span>    }<a name="line.118"></a>
+<span class="sourceLineNo">119</span>  }<a name="line.119"></a>
+<span class="sourceLineNo">120</span><a name="line.120"></a>
+<span class="sourceLineNo">121</span>  protected static final byte[] tableSeparator = ";".getBytes(StandardCharsets.UTF_8);<a name="line.121"></a>
+<span class="sourceLineNo">122</span><a name="line.122"></a>
+<span class="sourceLineNo">123</span>  protected static byte[] combineTableNameSuffix(byte[] tableName,<a name="line.123"></a>
+<span class="sourceLineNo">124</span>                                       byte[] suffix ) {<a name="line.124"></a>
+<span class="sourceLineNo">125</span>    return Bytes.add(tableName, tableSeparator, suffix);<a name="line.125"></a>
+<span class="sourceLineNo">126</span>  }<a name="line.126"></a>
+<span class="sourceLineNo">127</span><a name="line.127"></a>
+<span class="sourceLineNo">128</span>  // The following constants are private since these are used by<a name="line.128"></a>
+<span class="sourceLineNo">129</span>  // HFileOutputFormat2 to internally transfer data between job setup and<a name="line.129"></a>
+<span class="sourceLineNo">130</span>  // reducer run using conf.<a name="line.130"></a>
+<span class="sourceLineNo">131</span>  // These should not be changed by the client.<a name="line.131"></a>
+<span class="sourceLineNo">132</span>  static final String COMPRESSION_FAMILIES_CONF_KEY =<a name="line.132"></a>
+<span class="sourceLineNo">133</span>      "hbase.hfileoutputformat.families.compression";<a name="line.133"></a>
+<span class="sourceLineNo">134</span>  static final String BLOOM_TYPE_FAMILIES_CONF_KEY =<a name="line.134"></a>
+<span class="sourceLineNo">135</span>      "hbase.hfileoutputformat.families.bloomtype";<a name="line.135"></a>
+<span class="sourceLineNo">136</span>  static final String BLOCK_SIZE_FAMILIES_CONF_KEY =<a name="line.136"></a>
+<span class="sourceLineNo">137</span>      "hbase.mapreduce.hfileoutputformat.blocksize";<a name="line.137"></a>
+<span class="sourceLineNo">138</span>  static final String DATABLOCK_ENCODING_FAMILIES_CONF_KEY =<a name="line.138"></a>
+<span class="sourceLineNo">139</span>      "hbase.mapreduce.hfileoutputformat.families.datablock.encoding";<a name="line.139"></a>
 <span class="sourceLineNo">140</span><a name="line.140"></a>
-<span class="sourceLineNo">141</span>  protected static class HFileRecordWriter&lt;V extends Cell&gt;<a name="line.141"></a>
-<span class="sourceLineNo">142</span>      extends RecordWriter&lt;ImmutableBytesWritable, V&gt; {<a name="line.142"></a>
-<span class="sourceLineNo">143</span>    private final TaskAttemptContext context;<a name="line.143"></a>
-<span class="sourceLineNo">144</span>    private final Path outputPath;<a name="line.144"></a>
-<span class="sourceLineNo">145</span>    private final Path outputDir;<a name="line.145"></a>
-<span class="sourceLineNo">146</span>    private final Configuration conf;<a name="line.146"></a>
-<span class="sourceLineNo">147</span>    private final FileSystem fs;<a name="line.147"></a>
-<span class="sourceLineNo">148</span><a name="line.148"></a>
-<span class="sourceLineNo">149</span>    private final long maxsize;<a name="line.149"></a>
-<span class="sourceLineNo">150</span><a name="line.150"></a>
-<span class="sourceLineNo">151</span>    private final Algorithm defaultCompression;<a name="line.151"></a>
-<span class="sourceLineNo">152</span>    private final boolean compactionExclude;<a name="line.152"></a>
-<span class="sourceLineNo">153</span><a name="line.153"></a>
-<span class="sourceLineNo">154</span>    private final Map&lt;byte[], Algorithm&gt; compressionMap;<a name="line.154"></a>
-<span class="sourceLineNo">155</span>    private final Map&lt;byte[], BloomType&gt; bloomTypeMap;<a name="line.155"></a>
-<span class="sourceLineNo">156</span>    private final Map&lt;byte[], Integer&gt; blockSizeMap;<a name="line.156"></a>
-<span class="sourceLineNo">157</span><a name="line.157"></a>
-<span class="sourceLineNo">158</span>    private final Map&lt;byte[], DataBlockEncoding&gt; datablockEncodingMap;<a name="line.158"></a>
-<span class="sourceLineNo">159</span>    private final DataBlockEncoding overriddenEncoding;<a name="line.159"></a>
-<span class="sourceLineNo">160</span><a name="line.160"></a>
-<span class="sourceLineNo">161</span>    private final Map&lt;byte[], WriterLength&gt; writers;<a name="line.161"></a>
-<span class="sourceLineNo">162</span>    private byte[] previousRow;<a name="line.162"></a>
-<span class="sourceLineNo">163</span>    private final byte[] now;<a name="line.163"></a>
-<span class="sourceLineNo">164</span>    private boolean rollRequested;<a name="line.164"></a>
-<span class="sourceLineNo">165</span><a name="line.165"></a>
-<span class="sourceLineNo">166</span>    /**<a name="line.166"></a>
-<span class="sourceLineNo">167</span>     * Mapredue job will create a temp path for outputting results. If out != null, it means that<a name="line.167"></a>
-<span class="sourceLineNo">168</span>     * the caller has set the temp working dir; If out == null, it means we need to set it here.<a name="line.168"></a>
-<span class="sourceLineNo">169</span>     * Used by HFileOutputFormat2 and MultiTableHFileOutputFormat. MultiTableHFileOutputFormat will give us<a name="line.169"></a>
-<span class="sourceLineNo">170</span>     * temp working dir at the table level and HFileOutputFormat2 has to set it here within this<a name="line.170"></a>
-<span class="sourceLineNo">171</span>     * constructor.<a name="line.171"></a>
-<span class="sourceLineNo">172</span>     */<a name="line.172"></a>
-<span class="sourceLineNo">173</span>    public HFileRecordWriter(final TaskAttemptContext taContext, final Path out)<a name="line.173"></a>
-<span class="sourceLineNo">174</span>        throws IOException {<a name="line.174"></a>
-<span class="sourceLineNo">175</span>      // Get the path of the temporary output file<a name="line.175"></a>
-<span class="sourceLineNo">176</span>      context = taContext;<a name="line.176"></a>
-<span class="sourceLineNo">177</span><a name="line.177"></a>
-<span class="sourceLineNo">178</span>      if (out == null) {<a name="line.178"></a>
-<span class="sourceLineNo">179</span>        outputPath = FileOutputFormat.getOutputPath(context);<a name="line.179"></a>
-<span class="sourceLineNo">180</span>        outputDir = new FileOutputCommitter(outputPath, context).getWorkPath();<a name="line.180"></a>
-<span class="sourceLineNo">181</span>      } else {<a name="line.181"></a>
-<span class="sourceLineNo">182</span>        outputPath = out;<a name="line.182"></a>
-<span class="sourceLineNo">183</span>        outputDir = outputPath;<a name="line.183"></a>
-<span class="sourceLineNo">184</span>      }<a name="line.184"></a>
-<span class="sourceLineNo">185</span><a name="line.185"></a>
-<span class="sourceLineNo">186</span>      conf = context.getConfiguration();<a name="line.186"></a>
-<span class="sourceLineNo">187</span>      fs = outputDir.getFileSystem(conf);<a name="line.187"></a>
-<span class="sourceLineNo">188</span><a name="line.188"></a>
-<span class="sourceLineNo">189</span>      // These configs. are from hbase-*.xml<a name="line.189"></a>
-<span class="sourceLineNo">190</span>      maxsize = conf.getLong(HConstants.HREGION_MAX_FILESIZE, HConstants.DEFAULT_MAX_FILE_SIZE);<a name="line.190"></a>
-<span class="sourceLineNo">191</span><a name="line.191"></a>
-<span class="sourceLineNo">192</span>      // Invented config. Add to hbase-*.xml if other than default compression.<a name="line.192"></a>
-<span class="sourceLineNo">193</span>      String defaultCompressionStr = conf.get("hfile.compression", Compression.Algorithm.NONE.getName());<a name="line.193"></a>
-<span class="sourceLineNo">194</span>      defaultCompression = HFileWriterImpl.compressionByName(defaultCompressionStr);<a name="line.194"></a>
-<span class="sourceLineNo">195</span>      compactionExclude =<a name="line.195"></a>
-<span class="sourceLineNo">196</span>          conf.getBoolean("hbase.mapreduce.hfileoutputformat.compaction.exclude", false);<a name="line.196"></a>
+<span class="sourceLineNo">141</span>  // This constant is public since the client can modify this when setting<a name="line.141"></a>
+<span class="sourceLineNo">142</span>  // up their conf object and thus refer to this symbol.<a name="line.142"></a>
+<span class="sourceLineNo">143</span>  // It is present for backwards compatibility reasons. Use it only to<a name="line.143"></a>
+<span class="sourceLineNo">144</span>  // override the auto-detection of datablock encoding.<a name="line.144"></a>
+<span class="sourceLineNo">145</span>  public static final String DATABLOCK_ENCODING_OVERRIDE_CONF_KEY =<a name="line.145"></a>
+<span class="sourceLineNo">146</span>      "hbase.mapreduce.hfileoutputformat.datablock.encoding";<a name="line.146"></a>
+<span class="sourceLineNo">147</span><a name="line.147"></a>
+<span class="sourceLineNo">148</span>  /**<a name="line.148"></a>
+<span class="sourceLineNo">149</span>   * Keep locality while generating HFiles for bulkload. See HBASE-12596<a name="line.149"></a>
+<span class="sourceLineNo">150</span>   */<a name="line.150"></a>
+<span class="sourceLineNo">151</span>  public static final String LOCALITY_SENSITIVE_CONF_KEY =<a name="line.151"></a>
+<span class="sourceLineNo">152</span>      "hbase.bulkload.locality.sensitive.enabled";<a name="line.152"></a>
+<span class="sourceLineNo">153</span>  private static final boolean DEFAULT_LOCALITY_SENSITIVE = true;<a name="line.153"></a>
+<span class="sourceLineNo">154</span>  static final String OUTPUT_TABLE_NAME_CONF_KEY =<a name="line.154"></a>
+<span class="sourceLineNo">155</span>      "hbase.mapreduce.hfileoutputformat.table.name";<a name="line.155"></a>
+<span class="sourceLineNo">156</span>  static final String MULTI_TABLE_HFILEOUTPUTFORMAT_CONF_KEY =<a name="line.156"></a>
+<span class="sourceLineNo">157</span>          "hbase.mapreduce.use.multi.table.hfileoutputformat";<a name="line.157"></a>
+<span class="sourceLineNo">158</span><a name="line.158"></a>
+<span class="sourceLineNo">159</span>  public static final String STORAGE_POLICY_PROPERTY = "hbase.hstore.storagepolicy";<a name="line.159"></a>
+<span class="sourceLineNo">160</span>  public static final String STORAGE_POLICY_PROPERTY_CF_PREFIX = STORAGE_POLICY_PROPERTY + ".";<a name="line.160"></a>
+<span class="sourceLineNo">161</span><a name="line.161"></a>
+<span class="sourceLineNo">162</span>  @Override<a name="line.162"></a>
+<span class="sourceLineNo">163</span>  public RecordWriter&lt;ImmutableBytesWritable, Cell&gt; getRecordWriter(<a name="line.163"></a>
+<span class="sourceLineNo">164</span>      final TaskAttemptContext context) throws IOException, InterruptedException {<a name="line.164"></a>
+<span class="sourceLineNo">165</span>    return createRecordWriter(context);<a name="line.165"></a>
+<span class="sourceLineNo">166</span>  }<a name="line.166"></a>
+<span class="sourceLineNo">167</span><a name="line.167"></a>
+<span class="sourceLineNo">168</span>  protected static byte[] getTableNameSuffixedWithFamily(byte[] tableName, byte[] family) {<a name="line.168"></a>
+<span class="sourceLineNo">169</span>    return combineTableNameSuffix(tableName, family);<a name="line.169"></a>
+<span class="sourceLineNo">170</span>  }<a name="line.170"></a>
+<span class="sourceLineNo">171</span><a name="line.171"></a>
+<span class="sourceLineNo">172</span>  static &lt;V extends Cell&gt; RecordWriter&lt;ImmutableBytesWritable, V&gt;<a name="line.172"></a>
+<span class="sourceLineNo">173</span>      createRecordWriter(final TaskAttemptContext context)<a name="line.173"></a>
+<span class="sourceLineNo">174</span>          throws IOException {<a name="line.174"></a>
+<span class="sourceLineNo">175</span><a name="line.175"></a>
+<span class="sourceLineNo">176</span>    // Get the path of the temporary output file<a name="line.176"></a>
+<span class="sourceLineNo">177</span>    final Path outputPath = FileOutputFormat.getOutputPath(context);<a name="line.177"></a>
+<span class="sourceLineNo">178</span>    final Path outputDir = new FileOutputCommitter(outputPath, context).getWorkPath();<a name="line.178"></a>
+<span class="sourceLineNo">179</span>    final Configuration conf = context.getConfiguration();<a name="line.179"></a>
+<span class="sourceLineNo">180</span>    final boolean writeMultipleTables = conf.getBoolean(MULTI_TABLE_HFILEOUTPUTFORMAT_CONF_KEY, false) ;<a name="line.180"></a>
+<span class="sourceLineNo">181</span>    final String writeTableNames = conf.get(OUTPUT_TABLE_NAME_CONF_KEY);<a name="line.181"></a>
+<span class="sourceLineNo">182</span>    if (writeTableNames==null || writeTableNames.isEmpty()) {<a name="line.182"></a>
+<span class="sourceLineNo">183</span>      throw new IllegalArgumentException("Configuration parameter " + OUTPUT_TABLE_NAME_CONF_KEY<a name="line.183"></a>
+<span class="sourceLineNo">184</span>              + " cannot be empty");<a name="line.184"></a>
+<span class="sourceLineNo">185</span>    }<a name="line.185"></a>
+<span class="sourceLineNo">186</span>    final FileSystem fs = outputDir.getFileSystem(conf);<a name="line.186"></a>
+<span class="sourceLineNo">187</span>    // These configs. are from hbase-*.xml<a name="line.187"></a>
+<span class="sourceLineNo">188</span>    final long maxsize = conf.getLong(HConstants.HREGION_MAX_FILESIZE,<a name="line.188"></a>
+<span class="sourceLineNo">189</span>        HConstants.DEFAULT_MAX_FILE_SIZE);<a name="line.189"></a>
+<span class="sourceLineNo">190</span>    // Invented config.  Add to hbase-*.xml if other than default compression.<a name="line.190"></a>
+<span class="sourceLineNo">191</span>    final String defaultCompressionStr = conf.get("hfile.compression",<a name="line.191"></a>
+<span class="sourceLineNo">192</span>        Compression.Algorithm.NONE.getName());<a name="line.192"></a>
+<span class="sourceLineNo">193</span>    final Algorithm defaultCompression = HFileWriterImpl<a name="line.193"></a>
+<span class="sourceLineNo">194</span>        .compressionByName(defaultCompressionStr);<a name="line.194"></a>
+<span class="sourceLineNo">195</span>    final boolean compactionExclude = conf.getBoolean(<a name="line.195"></a>
+<span class="sourceLineNo">196</span>        "hbase.mapreduce.hfileoutputformat.compaction.exclude", false);<a name="line.196"></a>
 <span class="sourceLineNo">197</span><a name="line.197"></a>
-<span class="sourceLineNo">198</span>      // create a map from column family to the compression algorithm<a name="line.198"></a>
-<span class="sourceLineNo">199</span>      compressionMap = createFamilyCompressionMap(conf);<a name="line.199"></a>
-<span class="sourceLineNo">200</span>      bloomTypeMap = createFamilyBloomTypeMap(conf);<a name="line.200"></a>
-<span class="sourceLineNo">201</span>      blockSizeMap = createFamilyBlockSizeMap(conf);<a name="line.201"></a>
-<span class="sourceLineNo">202</span><a name="line.202"></a>
-<span class="sourceLineNo">203</span>      // Config for data block encoding<a name="line.203"></a>
-<span class="sourceLineNo">204</span>      String dataBlockEncodingStr = conf.get(DATABLOCK_ENCODING_OVERRIDE_CONF_KEY);<a name="line.204"></a>
-<span class="sourceLineNo">205</span>      datablockEncodingMap = createFamilyDataBlockEncodingMap(conf);<a name="line.205"></a>
-<span class="sourceLineNo">206</span>      if (dataBlockEncodingStr != null) {<a name="line.206"></a>
-<span class="sourceLineNo">207</span>        overriddenEncoding = DataBlockEncoding.valueOf(dataBlockEncodingStr);<a name="line.207"></a>
-<span class="sourceLineNo">208</span>      } else {<a name="line.208"></a>
-<span class="sourceLineNo">209</span>        overriddenEncoding = null;<a name="line.209"></a>
-<span class="sourceLineNo">210</span>      }<a name="line.210"></a>
-<span class="sourceLineNo">211</span><a name="line.211"></a>
-<span class="sourceLineNo">212</span>      writers = new TreeMap&lt;&gt;(Bytes.BYTES_COMPARATOR);<a name="line.212"></a>
-<span class="sourceLineNo">213</span>      previousRow = HConstants.EMPTY_BYTE_ARRAY;<a name="line.213"></a>
-<span class="sourceLineNo">214</span>      now = Bytes.toBytes(EnvironmentEdgeManager.currentTime());<a name="line.214"></a>
-<span class="sourceLineNo">215</span>      rollRequested = false;<a name="line.215"></a>
-<span class="sourceLineNo">216</span>    }<a name="line.216"></a>
-<span class="sourceLineNo">217</span><a name="line.217"></a>
-<span class="sourceLineNo">218</span>    @Override<a name="line.218"></a>
-<span class="sourceLineNo">219</span>    public void write(ImmutableBytesWritable row, V cell) throws IOException {<a name="line.219"></a>
-<span class="sourceLineNo">220</span>      KeyValue kv = KeyValueUtil.ensureKeyValue(cell);<a name="line.220"></a>
-<span class="sourceLineNo">221</span><a name="line.221"></a>
-<span class="sourceLineNo">222</span>      // null input == user explicitly wants to flush<a name="line.222"></a>
-<span class="sourceLineNo">223</span>      if (row == null &amp;&amp; kv == null) {<a name="line.223"></a>
-<span class="sourceLineNo">224</span>        rollWriters();<a name="line.224"></a>
-<span class="sourceLineNo">225</span>        return;<a name="line.225"></a>
-<span class="sourceLineNo">226</span>      }<a name="line.226"></a>
-<span class="sourceLineNo">227</span><a name="line.227"></a>
-<span class="sourceLineNo">228</span>      byte[] rowKey = CellUtil.cloneRow(kv);<a name="line.228"></a>
-<span class="sourceLineNo">229</span>      long length = kv.getLength();<a name="line.229"></a>
-<span class="sourceLineNo">230</span>      byte[] family = CellUtil.cloneFamily(kv);<a name="line.230"></a>
-<span class="sourceLineNo">231</span>      WriterLength wl = this.writers.get(family);<a name="line.231"></a>
-<span class="sourceLineNo">232</span><a name="line.232"></a>
-<span class="sourceLineNo">233</span>      // If this is a new column family, verify that the directory exists<a name="line.233"></a>
-<span class="sourceLineNo">234</span>      if (wl == null) {<a name="line.234"></a>
-<span class="sourceLineNo">235</span>        Path cfPath = new Path(outputDir, Bytes.toString(family));<a name="line.235"></a>
-<span class="sourceLineNo">236</span>        fs.mkdirs(cfPath);<a name="line.236"></a>
-<span class="sourceLineNo">237</span>        configureStoragePolicy(conf, fs, family, cfPath);<a name="line.237"></a>
-<span class="sourceLineNo">238</span>      }<a name="line.238"></a>
-<span class="sourceLineNo">239</span><a name="line.239"></a>
-<span class="sourceLineNo">240</span>      // If any of the HFiles for the column families has reached<a name="line.240"></a>
-<span class="sourceLineNo">241</span>      // maxsize, we need to roll all the writers<a name="line.241"></a>
-<span class="sourceLineNo">242</span>      if (wl != null &amp;&amp; wl.written + length &gt;= maxsize) {<a name="line.242"></a>
-<span class="sourceLineNo">243</span>        this.rollRequested = true;<a name="line.243"></a>
-<span class="sourceLineNo">244</span>      }<a name="line.244"></a>
-<span class="sourceLineNo">245</span><a name="line.245"></a>
-<span class="sourceLineNo">246</span>      // This can only happen once a row is finished though<a name="line.246"></a>
-<span class="sourceLineNo">247</span>      if (rollRequested &amp;&amp; Bytes.compareTo(this.previousRow, rowKey) != 0) {<a name="line.247"></a>
-<span class="sourceLineNo">248</span>        rollWriters();<a name="line.248"></a>
-<span class="sourceLineNo">249</span>      }<a name="line.249"></a>
+<span class="sourceLineNo">198</span>    final Set&lt;String&gt; allTableNames = Arrays.stream(writeTableNames.split(<a name="line.198"></a>
+<span class="sourceLineNo">199</span>            Bytes.toString(tableSeparator))).collect(Collectors.toSet());<a name="line.199"></a>
+<span class="sourceLineNo">200</span><a name="line.200"></a>
+<span class="sourceLineNo">201</span>    // create a map from column family to the compression algorithm<a name="line.201"></a>
+<span class="sourceLineNo">202</span>    final Map&lt;byte[], Algorithm&gt; compressionMap = createFamilyCompressionMap(conf);<a name="line.202"></a>
+<span class="sourceLineNo">203</span>    final Map&lt;byte[], BloomType&gt; bloomTypeMap = createFamilyBloomTypeMap(conf);<a name="line.203"></a>
+<span class="sourceLineNo">204</span>    final Map&lt;byte[], Integer&gt; blockSizeMap = createFamilyBlockSizeMap(conf);<a name="line.204"></a>
+<span class="sourceLineNo">205</span><a name="line.205"></a>
+<span class="sourceLineNo">206</span>    String dataBlockEncodingStr = conf.get(DATABLOCK_ENCODING_OVERRIDE_CONF_KEY);<a name="line.206"></a>
+<span class="sourceLineNo">207</span>    final Map&lt;byte[], DataBlockEncoding&gt; datablockEncodingMap<a name="line.207"></a>
+<span class="sourceLineNo">208</span>        = createFamilyDataBlockEncodingMap(conf);<a name="line.208"></a>
+<span class="sourceLineNo">209</span>    final DataBlockEncoding overriddenEncoding;<a name="line.209"></a>
+<span class="sourceLineNo">210</span>    if (dataBlockEncodingStr != null) {<a name="line.210"></a>
+<span class="sourceLineNo">211</span>      overriddenEncoding = DataBlockEncoding.valueOf(dataBlockEncodingStr);<a name="line.211"></a>
+<span class="sourceLineNo">212</span>    } else {<a name="line.212"></a>
+<span class="sourceLineNo">213</span>      overriddenEncoding = null;<a name="line.213"></a>
+<span class="sourceLineNo">214</span>    }<a name="line.214"></a>
+<span class="sourceLineNo">215</span><a name="line.215"></a>
+<span class="sourceLineNo">216</span>    return new RecordWriter&lt;ImmutableBytesWritable, V&gt;() {<a name="line.216"></a>
+<span class="sourceLineNo">217</span>      // Map of families to writers and how much has been output on the writer.<a name="line.217"></a>
+<span class="sourceLineNo">218</span>      private final Map&lt;byte[], WriterLength&gt; writers =<a name="line.218"></a>
+<span class="sourceLineNo">219</span>              new TreeMap&lt;&gt;(Bytes.BYTES_COMPARATOR);<a name="line.219"></a>
+<span class="sourceLineNo">220</span>      private byte[] previousRow = HConstants.EMPTY_BYTE_ARRAY;<a name="line.220"></a>
+<span class="sourceLineNo">221</span>      private final byte[] now = Bytes.toBytes(EnvironmentEdgeManager.currentTime());<a name="line.221"></a>
+<span class="sourceLineNo">222</span>      private boolean rollRequested = false;<a name="line.222"></a>
+<span class="sourceLineNo">223</span><a name="line.223"></a>
+<span class="sourceLineNo">224</span>      @Override<a name="line.224"></a>
+<span class="sourceLineNo">225</span>      public void write(ImmutableBytesWritable row, V cell)<a name="line.225"></a>
+<span class="sourceLineNo">226</span>          throws IOException {<a name="line.226"></a>
+<span class="sourceLineNo">227</span>        KeyValue kv = KeyValueUtil.ensureKeyValue(cell);<a name="line.227"></a>
+<span class="sourceLineNo">228</span><a name="line.228"></a>
+<span class="sourceLineNo">229</span>        // null input == user explicitly wants to flush<a name="line.229"></a>
+<span class="sourceLineNo">230</span>        if (row == null &amp;&amp; kv == null) {<a name="line.230"></a>
+<span class="sourceLineNo">231</span>          rollWriters();<a name="line.231"></a>
+<span class="sourceLineNo">232</span>          return;<a name="line.232"></a>
+<span class="sourceLineNo">233</span>        }<a name="line.233"></a>
+<span class="sourceLineNo">234</span><a name="line.234"></a>
+<span class="sourceLineNo">235</span>        byte[] rowKey = CellUtil.cloneRow(kv);<a name="line.235"></a>
+<span class="sourceLineNo">236</span>        long length = kv.getLength();<a name="line.236"></a>
+<span class="sourceLineNo">237</span>        byte[] family = CellUtil.cloneFamily(kv);<a name="line.237"></a>
+<span class="sourceLineNo">238</span>        byte[] tableNameBytes = null;<a name="line.238"></a>
+<span class="sourceLineNo">239</span>        if (writeMultipleTables) {<a name="line.239"></a>
+<span class="sourceLineNo">240</span>          tableNameBytes = MultiTableHFileOutputFormat.getTableName(row.get());<a name="line.240"></a>
+<span class="sourceLineNo">241</span>          if (!allTableNames.contains(Bytes.toString(tableNameBytes))) {<a name="line.241"></a>
+<span class="sourceLineNo">242</span>            throw new IllegalArgumentException("TableName '" + Bytes.toString(tableNameBytes) +<a name="line.242"></a>
+<span class="sourceLineNo">243</span>                    "' not" + " expected");<a name="line.243"></a>
+<span class="sourceLineNo">244</span>          }<a name="line.244"></a>
+<span class="sourceLineNo">245</span>        } else {<a name="line.245"></a>
+<span class="sourceLineNo">246</span>          tableNameBytes = writeTableNames.getBytes(StandardCharsets.UTF_8);<a name="line.246"></a>
+<span class="sourceLineNo">247</span>        }<a name="line.247"></a>
+<span class="sourceLineNo">248</span>        byte[] tableAndFamily = getTableNameSuffixedWithFamily(tableNameBytes, family);<a name="line.248"></a>
+<span class="sourceLineNo">249</span>        WriterLength wl = this.writers.get(tableAndFamily);<a name="line.249"></a>
 <span class="sourceLineNo">250</span><a name="line.250"></a>
-<span class="sourceLineNo">251</span>      // create a new WAL writer, if necessary<a name="line.251"></a>
-<span class="sourceLineNo">252</span>      if (wl == null || wl.writer == null) {<a name="line.252"></a>
-<span class="sourceLineNo">253</span>        if (conf.getBoolean(LOCALITY_SENSITIVE_CONF_KEY, DEFAULT_LOCALITY_SENSITIVE)) {<a name="line.253"></a>
-<span class="sourceLineNo">254</span>          HRegionLocation loc = null;<a name="line.254"></a>
-<span class="sourceLineNo">255</span>          String tableName = conf.get(OUTPUT_TABLE_NAME_CONF_KEY);<a name="line.255"></a>
-<span class="sourceLineNo">256</span>          if (tableName != null) {<a name="line.256"></a>
-<span class="sourceLineNo">257</span>            try (Connection connection = ConnectionFactory.createConnection(conf);<a name="line.257"></a>
-<span class="sourceLineNo">258</span>                RegionLocator locator = connection.getRegionLocator(TableName.valueOf(tableName))) {<a name="line.258"></a>
-<span class="sourceLineNo">259</span>              loc = locator.getRegionLocation(rowKey);<a name="line.259"></a>
-<span class="sourceLineNo">260</span>            } catch (Throwable e) {<a name="line.260"></a>
-<span class="sourceLineNo">261</span>              LOG.warn("there's something wrong when locating rowkey: " + Bytes.toString(rowKey),<a name="line.261"></a>
-<span class="sourceLineNo">262</span>                  e);<a name="line.262"></a>
-<span class="sourceLineNo">263</span>              loc = null;<a name="line.263"></a>
-<span class="sourceLineNo">264</span>            }<a name="line.264"></a>
-<span class="sourceLineNo">265</span>          }<a name="line.265"></a>
-<span class="sourceLineNo">266</span><a name="line.266"></a>
-<span class="sourceLineNo">267</span>          if (null == loc) {<a name="line.267"></a>
-<span class="sourceLineNo">268</span>            if (LOG.isTraceEnabled()) {<a name="line.268"></a>
-<span class="sourceLineNo">269</span>              LOG.trace(<a name="line.269"></a>
-<span class="sourceLineNo">270</span>                  "failed to get region location, so use default writer: " + Bytes.toString(rowKey));<a name="line.270"></a>
-<span class="sourceLineNo">271</span>            }<a name="line.271"></a>
-<span class="sourceLineNo">272</span>            wl = getNewWriter(family, conf, null);<a name="line.272"></a>
-<span class="sourceLineNo">273</span>          } else {<a name="line.273"></a>
-<span class="sourceLineNo">274</span>            if (LOG.isDebugEnabled()) {<a name="line.274"></a>
-<span class="sourceLineNo">275</span>              LOG.debug("first rowkey: [" + Bytes.toString(rowKey) + "]");<a name="line.275"></a>
-<span class="sourceLineNo">276</span>            }<a name="line.276"></a>
-<span class="sourceLineNo">277</span>            InetSocketAddress initialIsa = new InetSocketAddress(loc.getHostname(), loc.getPort());<a name="line.277"></a>
-<span class="sourceLineNo">278</span>            if (initialIsa.isUnresolved()) {<a name="line.278"></a>
-<span class="sourceLineNo">279</span>              if (LOG.isTraceEnabled()) {<a name="line.279"></a>
-<span class="sourceLineNo">280</span>                LOG.trace("failed to resolve bind address: " + loc.getHostname() + ":"<a name="line.280"></a>
-<span class="sourceLineNo">281</span>                    + loc.getPort() + ", so use default writer");<a name="line.281"></a>
-<span class="sourceLineNo">282</span>              }<a name="line.282"></a>
-<span class="sourceLineNo">283</span>              wl = getNewWriter(family, conf, null);<a name="line.283"></a>
-<span class="sourceLineNo">284</span>            } else {<a name="line.284"></a>
-<span class="sourceLineNo">285</span>              if (LOG.isDebugEnabled()) {<a name="line.285"></a>
-<span class="sourceLineNo">286</span>                LOG.debug("use favored nodes writer: " + initialIsa.getHostString());<a name="line.286"></a>
-<span class="sourceLineNo">287</span>              }<a name="line.287"></a>
-<span class="sourceLineNo">288</span>              wl = getNewWriter(family, conf, new InetSocketAddress[] { initialIsa });<a name="line.288"></a>
-<span class="sourceLineNo">289</span>            }<a name="line.289"></a>
-<span class="sourceLineNo">290</span>          }<a name="line.290"></a>
-<span class="sourceLineNo">291</span>        } else {<a name="line.291"></a>
-<span class="sourceLineNo">292</span>          wl = getNewWriter(family, conf, null);<a name="line.292"></a>
-<span class="sourceLineNo">293</span>        }<a name="line.293"></a>
-<span class="sourceLineNo">294</span>      }<a name="line.294"></a>
-<span class="sourceLineNo">295</span><a name="line.295"></a>
-<span class="sourceLineNo">296</span>      // we now have the proper WAL writer. full steam ahead<a name="line.296"></a>
-<span class="sourceLineNo">297</span>      kv.updateLatestStamp(this.now);<a name="line.297"></a>
-<span class="sourceLineNo">298</span>      wl.writer.append(kv);<a name="line.298"></a>
-<span class="sourceLineNo">299</span>      wl.written += length;<a name="line.299"></a>
-<span class="sourceLineNo">300</span><a name="line.300"></a>
-<span class="sourceLineNo">301</span>      // Copy the row so we know when a row transition.<a name="line.301"></a>
-<span class="sourceLineNo">302</span>      this.previousRow = rowKey;<a name="line.302"></a>
-<span class="sourceLineNo">303</span>    }<a name="line.303"></a>
-<span class="sourceLineNo">304</span><a name="line.304"></a>
-<span class="sourceLineNo">305</span>    private void rollWriters() throws IOException {<a name="line.305"></a>
-<span class="sourceLineNo">306</span>      for (WriterLength wl : this.writers.values()) {<a name="line.306"></a>
-<span class="sourceLineNo">307</span>        if (wl.writer != null) {<a name="line.307"></a>
-<span class="sourceLineNo">308</span>          LOG.info(<a name="line.308"></a>
-<span class="sourceLineNo">309</span>              "Writer=" + wl.writer.getPath() + ((wl.written == 0) ? "" : ", wrote=" + wl.written));<a name="line.309"></a>
-<span class="sourceLineNo">310</span>          close(wl.writer);<a name="line.310"></a>
-<span class="sourceLineNo">311</span>        }<a name="line.311"></a>
-<span class="sourceLineNo">312</span>        wl.writer = null;<a name="line.312"></a>
-<span class="sourceLineNo">313</span>        wl.written = 0;<a name="line.313"></a>
-<span class="sourceLineNo">314</span>      }<a name="line.314"></a>
-<span class="sourceLineNo">315</span>      this.rollRequested = false;<a name="line.315"></a>
-<span class="sourceLineNo">316</span>    }<a name="line.316"></a>
-<span class="sourceLineNo">317</span><a name="line.317"></a>
-<span class="sourceLineNo">318</span>    /*<a name="line.318"></a>
-<span class="sourceLineNo">319</span>     * Create a new StoreFile.Writer.<a name="line.319"></a>
-<span class="sourceLineNo">320</span>     * @param family<a name="line.320"></a>
-<span class="sourceLineNo">321</span>     * @return A WriterLength, containing a new StoreFile.Writer.<a name="line.321"></a>
-<span class="sourceLineNo">322</span>     * @throws IOException<a name="line.322"></a>
-<span class="sourceLineNo">323</span>     */<a name="line.323"></a>
-<span class="sourceLineNo">324</span>    @edu.umd.cs.findbugs.annotations.SuppressWarnings(value = "BX_UNBOXING_IMMEDIATELY_REBOXED",<a name="line.324"></a>
-<span class="sourceLineNo">325</span>        justification = "Not important")<a name="line.325"></a>
-<span class="sourceLineNo">326</span>    private WriterLength getNewWriter(byte[] family, Configuration conf,<a name="line.326"></a>
-<span class="sourceLineNo">327</span>        InetSocketAddress[] favoredNodes) throws IOException {<a name="line.327"></a>
-<span class="sourceLineNo">328</span>      WriterLength wl = new WriterLength();<a name="line.328"></a>
-<span class="sourceLineNo">329</span>      Path familyDir = new Path(outputDir, Bytes.toString(family));<a name="line.329"></a>
-<span class="sourceLineNo">330</span>      Algorithm compression = compressionMap.get(family);<a name="line.330"></a>
-<span class="sourceLineNo">331</span>      compression = compression == null ? defaultCompression : compression;<a name="line.331"></a>
-<span class="sourceLineNo">332</span>      BloomType bloomType = bloomTypeMap.get(family);<a name="line.332"></a>
-<span class="sourceLineNo">333</span>      bloomType = bloomType == null ? BloomType.NONE : bloomType;<a name="line.333"></a>
-<span class="sourceLineNo">334</span>      Integer blockSize = blockSizeMap.get(family);<a name="line.334"></a>
-<span class="sourceLineNo">335</span>      blockSize = blockSize == null ? HConstants.DEFAULT_BLOCKSIZE : blockSize;<a name="line.335"></a>
-<span class="sourceLineNo">336</span>      DataBlockEncoding encoding = overriddenEncoding;<a name="line.336"></a>
-<span class="sourceLineNo">337</span>      encoding = encoding == null ? datablockEncodingMap.get(family) : encoding;<a name="line.337"></a>
-<span class="sourceLineNo">338</span>      encoding = encoding == null ? DataBlockEncoding.NONE : encoding;<a name="line.338"></a>
-<span class="sourceLineNo">339</span>      Configuration tempConf = new Configuration(conf);<a name="line.339"></a>
-<span class="sourceLineNo">340</span>      tempConf.setFloat(HConstants.HFILE_BLOCK_CACHE_SIZE_KEY, 0.0f);<a name="line.340"></a>
-<span class="sourceLineNo">341</span>      HFileContextBuilder contextBuilder = new HFileContextBuilder().withCompression(compression)<a name="line.341"></a>
-<span class="sourceLineNo">342</span>          .withChecksumType(HStore.getChecksumType(conf))<a name="line.342"></a>
-<span class="sourceLineNo">343</span>          .withBytesPerCheckSum(HStore.getBytesPerChecksum(conf)).withBlockSize(blockSize);<a name="line.343"></a>
-<span class="sourceLineNo">344</span><a name="line.344"></a>
-<span class="sourceLineNo">345</span>      if (HFile.getFormatVersion(conf) &gt;= HFile.MIN_FORMAT_VERSION_WITH_TAGS) {<a name="line.345"></a>
-<span class="sourceLineNo">346</span>        contextBuilder.withIncludesTags(true);<a name="line.346"></a>
-<span class="sourceLineNo">347</span>      }<a name="line.347"></a>
-<span class="sourceLineNo">348</span><a name="line.348"></a>
-<span class="sourceLineNo">349</span>      contextBuilder.withDataBlockEncoding(encoding);<a name="line.349"></a>
-<span class="sourceLineNo">350</span>      HFileContext hFileContext = contextBuilder.build();<a name="line.350"></a>
-<span class="sourceLineNo">351</span><a name="line.351"></a>
-<span class="sourceLineNo">352</span>      if (null == favoredNodes) {<a name="line.352"></a>
-<span class="sourceLineNo">353</span>        wl.writer = new StoreFileWriter.Builder(conf, new CacheConfig(tempConf), fs)<a name="line.353"></a>
-<span class="sourceLineNo">354</span>            .withOutputDir(familyDir).withBloomType(bloomType)<a name="line.354"></a>
-<span class="sourceLineNo">355</span>            .withComparator(CellComparator.COMPARATOR).withFileContext(hFileContext).build();<a name="line.355"></a>
-<span class="sourceLineNo">356</span>      } else {<a name="line.356"></a>
-<span class="sourceLineNo">357</span>        wl.writer =<a name="line.357"></a>
-<span class="sourceLineNo">358</span>            new StoreFileWriter.Builder(conf, new CacheConfig(tempConf), new HFileSystem(fs))<a name="line.358"></a>
-<span class="sourceLineNo">359</span>                .withOutputDir(familyDir).withBloomType(bloomType)<a name="line.359"></a>
-<span class="sourceLineNo">360</span>                .withComparator(CellComparator.COMPARATOR).withFileContext(hFileContext)<a name="line.360"></a>
-<span class="sourceLineNo">361</span>                .withFavoredNodes(favoredNodes).build();<a name="line.361"></a>
-<span class="sourceLineNo">362</span>      }<a name="line.362"></a>
-<span class="sourceLineNo">363</span><a name="line.363"></a>
-<span class="sourceLineNo">364</span>      this.writers.put(family, wl);<a name="line.364"></a>
-<span class="sourceLineNo">365</span>      return wl;<a name="line.365"></a>
-<span class="sourceLineNo">366</span>    }<a name="line.366"></a>
-<span class="sourceLineNo">367</span><a name="line.367"></a>
-<span class="sourceLineNo">368</span>    private void close(final StoreFileWriter w) throws IOException {<a name="line.368"></a>
-<span class="sourceLineNo">369</span>      if (w != null) {<a name="line.369"></a>
-<span class="sourceLineNo">370</span>        w.appendFileInfo(StoreFile.BULKLOAD_TIME_KEY, Bytes.toBytes(EnvironmentEdgeManager.currentTime()));<a name="line.370"></a>
-<span class="sourceLineNo">371</span>        w.appendFileInfo(StoreFile.BULKLOAD_TASK_KEY,<a name="line.371"></a>
-<span class="sourceLineNo">372</span>            Bytes.toBytes(context.getTaskAttemptID().toString()));<a name="line.372"></a>
-<span class="sourceLineNo">373</span>        w.appendFileInfo(StoreFile.MAJOR_COMPACTION_KEY, Bytes.toBytes(true));<a name="line.373"></a>
-<span class="sourceLineNo">374</span>        w.appendFileInfo(StoreFile.EXCLUDE_FROM_MINOR_COMPACTION_KEY,<a name="line.374"></a>
-<span class="sourceLineNo">375</span>            Bytes.toBytes(compactionExclude));<a name="line.375"></a>
-<span class="sourceLineNo">376</span>        w.appendTrackedTimestampsToMetadata();<a name="line.376"></a>
-<span class="sourceLineNo">377</span>        w.close();<a name="line.377"></a>
-<span class="sourceLineNo">378</span>      }<a name="line.378"></a>
-<span class="sourceLineNo">379</span>    }<a name="line.379"></a>
-<span class="sourceLineNo">380</span><a name="line.380"></a>
-<span class="sourceLineNo">381</span>    @Override<a name="line.381"></a>
-<span class="sourceLineNo">382</span>    public void close(TaskAttemptContext c) throws IOException, InterruptedException {<a name="line.382"></a>
-<span class="sourceLineNo">383</span>      for (WriterLength wl : this.writers.values()) {<a name="line.383"></a>
-<span class="sourceLineNo">384</span>        close(wl.writer);<a name="line.384"></a>
-<span class="sourceLineNo">385</span>      }<a name="line.385"></a>
-<span class="sourceLineNo">386</span>    }<a name="line.386"></a>
-<span class="sourceLineNo">387</span>  }<a name="line.387"></a>
-<span class="sourceLineNo">388</span><a name="line.388"></a>
-<span class="sourceLineNo">389</span>  /**<a name="line.389"></a>
-<span class="sourceLineNo">390</span>   * Configure block storage policy for CF after the directory is created.<a name="line.390"></a>
-<span class="sourceLineNo">391</span>   */<a name="line.391"></a>
-<span class="sourceLineNo">392</span>  static void configureStoragePolicy(final Configuration conf, final FileSystem fs,<a name="line.392"></a>
-<span class="sourceLineNo">393</span>      byte[] family, Path cfPath) {<a name="line.393"></a>
-<span class="sourceLineNo">394</span>    if (null == conf || null == fs || null == family || null == cfPath) {<a name="line.394"></a>
-<span class="sourceLineNo">395</span>      return;<a name="line.395"></a>
-<span class="sourceLineNo">396</span>    }<a name="line.396"></a>
-<span class="sourceLineNo">397</span><a name="line.397"></a>
-<span class="sourceLineNo">398</span>    String policy =<a name="line.398"></a>
-<span class="sourceLineNo">399</span>        conf.get(STORAGE_POLICY_PROPERTY_CF_PREFIX + Bytes.toString(family),<a name="line.399"></a>
-<span class="sourceLineNo">400</span>          conf.get(STORAGE_POLICY_PROPERTY));<a name="line.400"></a>
-<span class="sourceLineNo">401</span>    FSUtils.setStoragePolicy(fs, cfPath, policy);<a name="line.401"></a>
-<span class="sourceLineNo">402</span>  }<a name="line.402"></a>
-<span class="sourceLineNo">403</span><a name="line.403"></a>
-<span class="sourceLineNo">404</span>  /*<a name="line.404"></a>
-<span class="sourceLineNo">405</span>   * Data structure to hold a Writer and amount of data written on it.<a name="line.405"></a>
-<span class="sourceLineNo">406</span>   */<a name="line.406"></a>
-<span class="sourceLineNo">407</span>  static class WriterLength {<a name="line.407"></a>
-<span class="sourceLineNo">408</span>    long written = 0;<a name="line.408"></a>
-<span class="sourceLineNo">409</span>    StoreFileWriter writer = null;<a name="line.409"></a>
-<span class="sourceLineNo">410</span>  }<a name="line.410"></a>
-<span class="sourceLineNo">411</span><a name="line.411"></a>
-<span class="sourceLineNo">412</span>  /**<a name="line.412"></a>
-<span class="sourceLineNo">413</span>   * Return the start keys of all of the regions in this table,<a name="line.413"></a>
-<span class="sourceLineNo">414</span>   * as a list of ImmutableBytesWritable.<a name="line.414"></a>
-<span class="sourceLineNo">415</span>   */<a name="line.415"></a>
-<span class="sourceLineNo">416</span>  private static List&lt;ImmutableBytesWritable&gt; getRegionStartKeys(RegionLocator table)<a name="line.416"></a>
-<span class="sourceLineNo">417</span>  throws IOException {<a name="line.417"></a>
-<span class="sourceLineNo">418</span>    byte[][] byteKeys = table.getStartKeys();<a name="line.418"></a>
-<span class="sourceLineNo">419</span>    ArrayList&lt;ImmutableBytesWritable&gt; ret = new ArrayList&lt;&gt;(byteKeys.length);<a name="line.419"></a>
-<span class="sourceLineNo">420</span>    for (byte[] byteKey : byteKeys) {<a name="line.420"></a>
-<span class="sourceLineNo">421</span>      ret.add(new ImmutableBytesWritable(byteKey));<a name="line.421"></a>
-<span class="sourceLineNo">422</span>    }<a name="line.422"></a>
-<span class="sourceLineNo">423</span>    return ret;<a name="line.423"></a>
-<span class="sourceLineNo">424</span>  }<a name="line.424"></a>
-<span class="sourceLineNo">425</span><a name="line.425"></a>
-<span class="sourceLineNo">426</span>  /**<a name="line.426"></a>
-<span class="sourceLineNo">427</span>   * Write out a {@link SequenceFile} that can be read by<a name="line.427"></a>
-<span class="sourceLineNo">428</span>   * {@link TotalOrderPartitioner} that contains the split points in startKeys.<a name="line.428"></a>
-<span class="sourceLineNo">429</span>   */<a name="line.429"></a>
-<span class="sourceLineNo">430</span>  @SuppressWarnings("deprecation")<a name="line.430"></a>
-<span class="sourceLineNo">431</span>  private static void writePartitions(Configuration conf, Path partitionsPath,<a name="line.431"></a>
-<span class="sourceLineNo">432</span>      List&lt;ImmutableBytesWritable&gt; startKeys) throws IOException {<a name="line.432"></a>
-<span class="sourceLineNo">433</span>    LOG.info("Writing partition information to " + partitionsPath);<a name="line.433"></a>
-<span class="sourceLineNo">434</span>    if (startKeys.isEmpty()) {<a name="line.434"></a>
-<span class="sourceLineNo">435</span>      throw new IllegalArgumentException("No regions passed");<a name="line.435"></a>
-<span class="sourceLineNo">436</span>    }<a name="line.436"></a>
-<span class="sourceLineNo">437</span><a name="line.437"></a>
-<span class="sourceLineNo">438</span>    // We're generating a list of split points, and we don't ever<a name="line.438"></a>
-<span class="sourceLineNo">439</span>    // have keys &lt; the first region (which has an empty start key)<a name="line.439"></a>
-<span class="sourceLineNo">440</span>    // so we need to remove it. Otherwise we would end up with an<a name="line.440"></a>
-<span class="sourceLineNo">441</span>    // empty reducer with index 0<a name="line.441"></a>
-<span class="sourceLineNo">442</span>    TreeSet&lt;ImmutableBytesWritable&gt; sorted = new TreeSet&lt;&gt;(startKeys);<a name="line.442"></a>
-<span class="sourceLineNo">443</span><a name="line.443"></a>
-<span class="sourceLineNo">444</span>    ImmutableBytesWritable first = sorted.first();<a name="line.444"></a>
-<span class="sourceLineNo">445</span>    if (!first.equals(HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.445"></a>
-<span class="sourceLineNo">446</span>      throw new IllegalArgumentException(<a name="line.446"></a>
-<span class="sourceLineNo">447</span>          "First region of table should have empty start key. Instead has: "<a name="line.447"></a>
-<span class="sourceLineNo">448</span>          + Bytes.toStringBinary(first.get()));<a name="line.448"></a>
-<span class="sourceLineNo">449</span>    }<a name="line.449"></a>
-<span class="sourceLineNo">450</span>    sorted.remove(first);<a name="line.450"></a>
-<span class="sourceLineNo">451</span><a name="line.451"></a>
-<span class="sourceLineNo">452</span>    // Write the actual file<a name="line.452"></a>
-<span class="sourceLineNo">453</span>    FileSystem fs = partitionsPath.getFileSystem(conf);<a name="line.453"></a>
-<span class="sourceLineNo">454</span>    SequenceFile.Writer writer = SequenceFile.createWriter(<a name="line.454"></a>
-<span class="sourceLineNo">455</span>      fs, conf, partitionsPath, ImmutableBytesWritable.class,<a name="line.455"></a>
-<span class="sourceLineNo">456</span>      NullWritable.class);<a name="line.456"></a>
-<span class="sourceLineNo">457</span><a name="line.457"></a>
-<span class="sourceLineNo">458</span>    try {<a name="line.458"></a>
-<span class="sourceLineNo">459</span>      for (ImmutableBytesWritable startKey : sorted) {<a name="line.459"></a>
-<span class="sourceLineNo">460</span>        writer.append(startKey, NullWritable.get());<a name="line.460"></a>
-<span class="sourceLineNo">461</span>      }<a name="line.461"></a>
-<span class="sourceLineNo">462</span>    } finally {<a name="line.462"></a>
-<span class="sourceLineNo">463</span>      writer.close();<a name="line.463"></a>
-<span class="sourceLineNo">464</span>    }<a name="line.464"></a>
-<span class="sourceLineNo">465</span>  }<a name="line.465"></a>
-<span class="sourceLineNo">466</span><a name="line.466"></a>
-<span class="sourceLineNo">467</span>  /**<a name="line.467"></a>
-<span class="sourceLineNo">468</span>   * Configure a MapReduce Job to perform an incremental load into the given<a name="line.468"></a>
-<span class="sourceLineNo">469</span>   * table. This<a name="line.469"></a>
-<span class="sourceLineNo">470</span>   * &lt;ul&gt;<a name="line.470"></a>
-<span class="sourceLineNo">471</span>   *   &lt;li&gt;Inspects the table to configure a total order partitioner&lt;/li&gt;<a name="line.471"></a>
-<span class="sourceLineNo">472</span>   *   &lt;li&gt;Uploads the partitions file to the cluster and adds it to the DistributedCache&lt;/li&gt;<a name="line.472"></a>
-<span class="sourceLineNo">473</span>   *   &lt;li&gt;Sets the number of reduce tasks to match the current number of regions&lt;/li&gt;<a name="line.473"></a>
-<span class="sourceLineNo">474</span>   *   &lt;li&gt;Sets the output key/value class to match HFileOutputFormat2's requirements&lt;/li&gt;<a name="line.474"></a>
-<span class="sourceLineNo">475</span>   *   &lt;li&gt;Sets the reducer up to perform the appropriate sorting (either KeyValueSortReducer or<a name="line.475"></a>
-<span class="sourceLineNo">476</span>   *     PutSortReducer)&lt;/li&gt;<a name="line.476"></a>
-<span class="sourceLineNo">477</span>   * &lt;/ul&gt;<a name="line.477"></a>
-<span class="sourceLineNo">478</span>   * The user should be sure to set the map output value class to either KeyValue or Put before<a name="line.478"></a>
-<span class="sourceLineNo">479</span>   * running this function.<a name="line.479"></a>
-<span class="sourceLineNo">480</span>   */<a name="line.480"></a>
-<span class="sourceLineNo">481</span>  public static void configureIncrementalLoad(Job job, Table table, RegionLocator regionLocator)<a name="line.481"></a>
-<span class="sourceLineNo">482</span>      throws IOException {<a name="line.482"></a>
-<span class="sourceLineNo">483</span>    configureIncrementalLoad(job, table.getTableDescriptor(), regionLocator);<a name="line.483"></a>
-<span class="sourceLineNo">484</span>  }<a name="line.484"></a>
-<span class="sourceLineNo">485</span><a name="line.485"></a>
-<span class="sourceLineNo">486</span>  /**<a name="line.486"></a>
-<span class="sourceLineNo">487</span>   * Configure a MapReduce Job to perform an incremental load into the given<a name="line.487"></a>
-<span class="sourceLineNo">488</span>   * table. This<a name="line.488"></a>
-<span class="sourceLineNo">489</span>   * &lt;ul&gt;<a name="line.489"></a>
-<span class="sourceLineNo">490</span>   *   &lt;li&gt;Inspects the table to configure a total order partitioner&lt;/li&gt;<a name="line.490"></a>
-<span class="sourceLineNo">491</span>   *   &lt;li&gt;Uploads the partitions file to the cluster and adds it to the DistributedCache&lt;/li&gt;<a name="line.491"></a>
-<span class="sourceLineNo">492</span>   *   &lt;li&gt;Sets the number of reduce tasks to match the current number of regions&lt;/li&gt;<a name="line.492"></a>
-<span class="sourceLineNo">493</span>   *   &lt;li&gt;Sets the output key/value class to match HFileOutputFormat2's requirements&lt;/li&gt;<a name="line.493"></a>
-<span class="sourceLineNo">494</span>   *   &lt;li&gt;Sets the reducer up to perform the appropriate sorting (either KeyValueSortReducer or<a name="line.494"></a>
-<span class="sourceLineNo">495</span>   *     PutSortReducer)&lt;/li&gt;<a name="line.495"></a>
-<span class="sourceLineNo">496</span>   * &lt;/ul&gt;<a name="line.496"></a>
-<span class="sourceLineNo">497</span>   * The user should be sure to set the map output value class to either KeyValue or Put before<a name="line.497"></a>
-<span class="sourceLineNo">498</span>   * running this function.<a name="line.498"></a>
-<span class="sourceLineNo">499</span>   */<a name="line.499"></a>
-<span class="sourceLineNo">500</span>  public static void configureIncrementalLoad(Job job, HTableDescriptor tableDescriptor,<a name="line.500"></a>
-<span class="sourceLineNo">501</span>      RegionLocator regionLocator) throws IOException {<a name="line.501"></a>
-<span class="sourceLineNo">502</span>    configureIncrementalLoad(job, tableDescriptor, regionLocator, HFileOutputFormat2.class);<a name="line.502"></a>
-<span class="sourceLineNo">503</span>  }<a name="line.503"></a>
-<span class="sourceLineNo">504</span><a name="line.504"></a>
-<span class="sourceLineNo">505</span>  static void configureIncrementalLoad(Job job, HTableDescriptor tableDescriptor,<a name="line.505"></a>
-<span class="sourceLineNo">506</span>      RegionLocator regionLocator, Class&lt;? extends OutputFormat&lt;?, ?&gt;&gt; cls) throws IOException,<a name="line.506"></a>
-<span class="sourceLineNo">507</span>      UnsupportedEncodingException {<a name="line.507"></a>
-<span class="sourceLineNo">508</span>    Configuration conf = job.getConfiguration();<a name="line.508"></a>
-<span class="sourceLineNo">509</span>    job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.509"></a>
-<span class="sourceLineNo">510</span>    job.setOutputValueClass(KeyValue.class);<a name="line.510"></a>
-<span class="sourceLineNo">511</span>    job.setOutputFormatClass(cls);<a name="line.511"></a>
-<span class="sourceLineNo">512</span><a name="line.512"></a>
-<span class="sourceLineNo">513</span>    // Based on the configured map output class, set the correct reducer to properly<a name="line.513"></a>
-<span class="sourceLineNo">514</span>    // sort the incoming values.<a name="line.514"></a>
-<span class="sourceLineNo">515</span>    // TODO it would be nice to pick one or the other of these formats.<a name="line.515"></a>
-<span class="sourceLineNo">516</span>    if (KeyValue.class.equals(job.getMapOutputValueClass())) {<a name="line.516"></a>
-<span class="sourceLineNo">517</span>      job.setReducerClass(KeyValueSortReducer.class);<a name="line.517"></a>
-<span class="sourceLineNo">518</span>    } else if (Put.class.equals(job.getMapOutputValueClass())) {<a name="line.518"></a>
-<span class="sourceLineNo">519</span>      job.setReducerClass(PutSortReducer.class);<a name="line.519"></a>
-<span class="sourceLineNo">520</span>    } else if (Text.class.equals(job.getMapOutputValueClass())) {<a name="line.520"></a>
-<span class="sourceLineNo">521</span>      job.setReducerClass(TextSortReducer.class);<a name="line.521"></a>
-<span class="sourceLineNo">522</span>    } else {<a name="line.522"></a>
-<span class="sourceLineNo">523</span>      LOG.warn("Unknown map output value type:" + job.getMapOutputValueClass());<a name="line.523"></a>
-<span class="sourceLineNo">524</span>    }<a name="line.524"></a>
+<span class="sourceLineNo">251</span>        // If this is a new column family, verify that the directory exists<a name="line.251"></a>
+<span class="sourceLineNo">252</span>        if (wl == null) {<a name="line.252"></a>
+<span class="sourceLineNo">253</span>          Path writerPath = null;<a name="line.253"></a>
+<span class="sourceLineNo">254</span>          if (writeMultipleTables) {<a name="line.254"></a>
+<span class="sourceLineNo">255</span>            writerPath = new Path(outputDir, new Path(Bytes.toString(tableNameBytes), Bytes<a name="line.255"></a>
+<span class="sourceLineNo">256</span>                    .toString(family)));<a name="line.256"></a>
+<span class="sourceLineNo">257</span>          }<a name="line.257"></a>
+<span class="sourceLineNo">258</span>          else {<a name="line.258"></a>
+<span class="sourceLineNo">259</span>            writerPath = new Path(outputDir, Bytes.toString(family));<a name="line.259"></a>
+<span class="sourceLineNo">260</span>          }<a name="line.260"></a>
+<span class="sourceLineNo">261</span>          fs.mkdirs(writerPath);<a name="line.261"></a>
+<span class="sourceLineNo">262</span>          configureStoragePolicy(conf, fs, tableAndFamily, writerPath);<a name="line.262"></a>
+<span class="sourceLineNo">263</span>        }<a name="line.263"></a>
+<span class="sourceLineNo">264</span><a name="line.264"></a>
+<span class="sourceLineNo">265</span>        // If any of the HFiles for the column families has reached<a name="line.265"></a>
+<span class="sourceLineNo">266</span>        // maxsize, we need to roll all the writers<a name="line.266"></a>
+<span class="sourceLineNo">267</span>        if (wl != null &amp;&amp; wl.written + length &gt;= maxsize) {<a name="line.267"></a>
+<span class="sourceLineNo">268</span>          this.rollRequested = true;<a name="line.268"></a>
+<span class="sourceLineNo">269</span>        }<a name="line.269"></a>
+<span class="sourceLineNo">270</span><a name="line.270"></a>
+<span class="sourceLineNo">271</span>        // This can only happen once a row is finished though<a name="line.271"></a>
+<span class="sourceLineNo">272</span>        if (rollRequested &amp;&amp; Bytes.compareTo(this.previousRow, rowKey) != 0) {<a name="line.272"></a>
+<span class="sourceLineNo">273</span>          rollWriters();<a name="line.273"></a>
+<span class="sourceLineNo">274</span>        }<a name="line.274"></a>
+<span class="sourceLineNo">275</span><a name="line.275"></a>
+<span class="sourceLineNo">276</span>        // create a new WAL writer, if necessary<a name="line.276"></a>
+<span class="sourceLineNo">277</span>        if (wl == null || wl.writer == null) {<a name="line.277"></a>
+<span class="sourceLineNo">278</span>          if (conf.getBoolean(LOCALITY_SENSITIVE_CONF_KEY, DEFAULT_LOCALITY_SENSITIVE)) {<a name="line.278"></a>
+<span class="sourceLineNo">279</span>            HRegionLocation loc = null;<a name="line.279"></a>
+<span class="sourceLineNo">280</span><a name="line.280"></a>
+<span class="sourceLineNo">281</span>            String tableName = Bytes.toString(tableNameBytes);<a name="line.281"></a>
+<span class="sourceLineNo">282</span>            if (tableName != null) {<a name="line.282"></a>
+<span class="sourceLineNo">283</span>              try (Connection connection = ConnectionFactory.createConnection(conf);<a name="line.283"></a>
+<span class="sourceLineNo">284</span>                     RegionLocator locator =<a name="line.284"></a>
+<span class="sourceLineNo">285</span>                       connection.getRegionLocator(TableName.valueOf(tableName))) {<a name="line.285"></a>
+<span class="sourceLineNo">286</span>                loc = locator.getRegionLocation(rowKey);<a name="line.286"></a>
+<span class="sourceLineNo">287</span>              } catch (Throwable e) {<a name="line.287"></a>
+<span class="sourceLineNo">288</span>                LOG.warn("There's something wrong when locating rowkey: " +<a name="line.288"></a>
+<span class="sourceLineNo">289</span>                  Bytes.toString(rowKey) + " for tablename: " + tableName, e);<a name="line.289"></a>
+<span class="sourceLineNo">290</span>                loc = null;<a name="line.290"></a>
+<span class="sourceLineNo">291</span>              } }<a name="line.291"></a>
+<span class="sourceLineNo">292</span><a name="line.292"></a>
+<span class="sourceLineNo">293</span>            if (null == loc) {<a name="line.293"></a>
+<span class="sourceLineNo">294</span>              if (LOG.isTraceEnabled()) {<a name="line.294"></a>
+<span class="sourceLineNo">295</span>                LOG.trace("failed to get region location, so use default writer for rowkey: " +<a name="line.295"></a>
+<span class="sourceLineNo">296</span>                  Bytes.toString(rowKey));<a name="line.296"></a>
+<span class="sourceLineNo">297</span>              }<a name="line.297"></a>
+<span class="sourceLineNo">298</span>              wl = getNewWriter(tableNameBytes, family, conf, null);<a name="line.298"></a>
+<span class="sourceLineNo">299</span>            } else {<a name="line.299"></a>
+<span class="sourceLineNo">300</span>              if (LOG.isDebugEnabled()) {<a name="line.300"></a>
+<span class="sourceLineNo">301</span>                LOG.debug("first rowkey: [" + Bytes.toString(rowKey) + "]");<a name="line.301"></a>
+<span class="sourceLineNo">302</span>              }<a name="line.302"></a>
+<span class="sourceLineNo">303</span>              InetSocketAddress initialIsa =<a name="line.303"></a>
+<span class="sourceLineNo">304</span>                  new InetSocketAddress(loc.getHostname(), loc.getPort());<a name="line.304"></a>
+<span class="sourceLineNo">305</span>              if (initialIsa.isUnresolved()) {<a name="line.305"></a>
+<span class="sourceLineNo">306</span>                if (LOG.isTraceEnabled()) {<a name="line.306"></a>
+<span class="sourceLineNo">307</span>                  LOG.trace("failed to resolve bind address: " + loc.getHostname() + ":"<a name="line.307"></a>
+<span class="sourceLineNo">308</span>                      + loc.getPort() + ", so use default writer");<a name="line.308"></a>
+<span class="sourceLineNo">309</span>                }<a name="line.309"></a>
+<span class="sourceLineNo">310</span>                wl = getNewWriter(tableNameBytes, family, conf, null);<a name="line.310"></a>
+<span class="sourceLineNo">311</span>              } else {<a name="line.311"></a>
+<span class="sourceLineNo">312</span>                if (LOG.isDebugEnabled()) {<a name="line.312"></a>
+<span class="sourceLineNo">313</span>                  LOG.debug("use favored nodes writer: " + initialIsa.getHostString());<a name="line.313"></a>
+<span class="sourceLineNo">314</span>                }<a name="line.314"></a>
+<span class="sourceLineNo">315</span>                wl = getNewWriter(tableNameBytes, family, conf, new InetSocketAddress[] { initialIsa<a name="line.315"></a>
+<span class="sourceLineNo">316</span>                });<a name="line.316"></a>
+<span class="sourceLineNo">317</span>              }<a name="line.317"></a>
+<span class="sourceLineNo">318</span>            }<a name="line.318"></a>
+<span class="sourceLineNo">319</span>          } else {<a name="line.319"></a>
+<span class="sourceLineNo">320</span>            wl = getNewWriter(tableNameBytes, family, conf, null);<a name="line.320"></a>
+<span class="sourceLineNo">321</span>          }<a name="line.321"></a>
+<span class="sourceLineNo">322</span>        }<a name="line.322"></a>
+<span class="sourceLineNo">323</span><a name="line.323"></a>
+<span class="sourceLineNo">324</span>        // we now have the proper WAL writer. full steam ahead<a name="line.324"></a>
+<span class="sourceLineNo">325</span>        kv.updateLatestStamp(this.now);<a name="line.325"></a>
+<span class="sourceLineNo">326</span>        wl.writer.append(kv);<a name="line.326"></a>
+<span class="sourceLineNo">327</span>        wl.written += length;<a name="line.327"></a>
+<span class="sourceLineNo">328</span><a name="line.328"></a>
+<span class="sourceLineNo">329</span>        // Copy the row so we know when a row transition.<a name="line.329"></a>
+<span class="sourceLineNo">330</span>        this.previousRow = rowKey;<a name="line.330"></a>
+<span class="sourceLineNo">331</span>      }<a name="line.331"></a>
+<span class="sourceLineNo">332</span><a name="line.332"></a>
+<span class="sourceLineNo">333</span>      private void rollWriters() throws IOException {<a name="line.333"></a>
+<span class="sourceLineNo">334</span>        for (WriterLength wl : this.writers.values()) {<a name="line.334"></a>
+<span class="sourceLineNo">335</span>          if (wl.writer != null) {<a name="line.335"></a>
+<span class="sourceLineNo">336</span>            LOG.info(<a name="line.336"></a>
+<span class="sourceLineNo">337</span>                "Writer=" + wl.writer.getPath() + ((wl.written == 0)? "": ", wrote=" + wl.written));<a name="line.337"></a>
+<span class="sourceLineNo">338</span>            close(wl.writer);<a name="line.338"></a>
+<span class="sourceLineNo">339</span>          }<a name="line.339"></a>
+<span class="sourceLineNo">340</span>          wl.writer = null;<a name="line.340"></a>
+<span class="sourceLineNo">341</span>          wl.written = 0;<a name="line.341"></a>
+<span class="sourceLineNo">342</span>        }<a name="line.342"></a>
+<span class="sourceLineNo">343</span>        this.rollRequested = false;<a name="line.343"></a>
+<span class="sourceLineNo">344</span>      }<a name="line.344"></a>
+<span class="sourceLineNo">345</span><a name="line.345"></a>
+<span class="sourceLineNo">346</span>      /*<a name="line.346"></a>
+<span class="sourceLineNo">347</span>       * Create a new StoreFile.Writer.<a name="line.347"></a>
+<span class="sourceLineNo">348</span>       * @param family<a name="line.348"></a>
+<span class="sourceLineNo">349</span>       * @return A WriterLength, containing a new StoreFile.Writer.<a name="line.349"></a>
+<span class="sourceLineNo">350</span>       * @throws IOException<a name="line.350"></a>
+<span class="sourceLineNo">351</span>       */<a name="line.351"></a>
+<span class="sourceLineNo">352</span>      @edu.umd.cs.findbugs.annotations.SuppressWarnings(value="BX_UNBOXING_IMMEDIATELY_REBOXED",<a name="line.352"></a>
+<span class="sourceLineNo">353</span>          justification="Not important")<a name="line.353"></a>
+<span class="sourceLineNo">354</span>      private WriterLength getNewWriter(byte[] tableName, byte[] family, Configuration<a name="line.354"></a>
+<span class="sourceLineNo">355</span>              conf, InetSocketAddress[] favoredNodes) throws IOException {<a name="line.355"></a>
+<span class="sourceLineNo">356</span>        byte[] tableAndFamily = getTableNameSuffixedWithFamily(tableName, family);<a name="line.356"></a>
+<span class="sourceLineNo">357</span>        Path familydir = new Path(outputDir, Bytes.toString(family));<a name="line.357"></a>
+<span class="sourceLineNo">358</span>        if (writeMultipleTables) {<a name="line.358"></a>
+<span class="sourceLineNo">359</span>          familydir = new Path(outputDir,<a name="line.359"></a>
+<span class="sourceLineNo">360</span>                  new Path(Bytes.toString(tableName), Bytes.toString(family)));<a name="line.360"></a>
+<span class="sourceLineNo">361</span>        }<a name="line.361"></a>
+<span class="sourceLineNo">362</span>        WriterLength wl = new WriterLength();<a name="line.362"></a>
+<span class="sourceLineNo">363</span>        Algorithm compression = compressionMap.get(tableAndFamily);<a name="line.363"></a>
+<span class="sourceLineNo">364</span>        compression = compression == null ? defaultCompression : compression;<a name="line.364"></a>
+<span class="sourceLineNo">365</span>        BloomType bloomType = bloomTypeMap.get(tableAndFamily);<a name="line.365"></a>
+<span class="sourceLineNo">366</span>        bloomType = bloomType == null ? BloomType.NONE : bloomType;<a name="line.366"></a>
+<span class="sourceLineNo">367</span>        Integer blockSize = blockSizeMap.get(tableAndFamily);<a name="line.367"></a>
+<span class="sourceLineNo">368</span>        blockSize = blockSize == null ? HConstants.DEFAULT_BLOCKSIZE : blockSize;<a name="line.368"></a>
+<span class="sourceLineNo">369</span>        DataBlockEncoding encoding = overriddenEncoding;<a name="line.369"></a>
+<span class="sourceLineNo">370</span>        encoding = encoding == null ? datablockEncodingMap.get(tableAndFamily) : encoding;<a name="line.370"></a>
+<span class="sourceLineNo">371</span>        encoding = encoding == null ? DataBlockEncoding.NONE : encoding;<a name="line.371"></a>
+<span class="sourceLineNo">372</span>        Configuration tempConf = new Configuration(conf);<a name="line.372"></a>
+<span class="sourceLineNo">373</span>        tempConf.setFloat(HConstants.HFILE_BLOCK_CACHE_SIZE_KEY, 0.0f);<a name="line.373"></a>
+<span class="sourceLineNo">374</span>        HFileContextBuilder contextBuilder = new HFileContextBuilder()<a name="line.374

<TRUNCATED>