You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hbase.apache.org by mi...@apache.org on 2016/06/27 17:20:04 UTC
[37/52] [partial] hbase-site git commit: Published site at
424b789e0394fdc074b1725275e30b04e601b486.
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/a0cdd2a9/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableOutputFormat.html
----------------------------------------------------------------------
diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableOutputFormat.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableOutputFormat.html
index 6d4afc9..e104f60 100644
--- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableOutputFormat.html
+++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableOutputFormat.html
@@ -37,196 +37,210 @@
<span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.HConstants;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hadoop.hbase.TableName;<a name="line.31"></a>
-<span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.client.BufferedMutator;<a name="line.32"></a>
-<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.33"></a>
-<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.34"></a>
-<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.client.Delete;<a name="line.35"></a>
-<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.client.Mutation;<a name="line.36"></a>
-<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.client.Put;<a name="line.37"></a>
-<span class="sourceLineNo">038</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.38"></a>
-<span class="sourceLineNo">039</span>import org.apache.hadoop.mapreduce.OutputCommitter;<a name="line.39"></a>
-<span class="sourceLineNo">040</span>import org.apache.hadoop.mapreduce.OutputFormat;<a name="line.40"></a>
-<span class="sourceLineNo">041</span>import org.apache.hadoop.mapreduce.RecordWriter;<a name="line.41"></a>
-<span class="sourceLineNo">042</span>import org.apache.hadoop.mapreduce.TaskAttemptContext;<a name="line.42"></a>
-<span class="sourceLineNo">043</span><a name="line.43"></a>
-<span class="sourceLineNo">044</span>/**<a name="line.44"></a>
-<span class="sourceLineNo">045</span> * Convert Map/Reduce output and write it to an HBase table. The KEY is ignored<a name="line.45"></a>
-<span class="sourceLineNo">046</span> * while the output value <u>must</u> be either a {@link Put} or a<a name="line.46"></a>
-<span class="sourceLineNo">047</span> * {@link Delete} instance.<a name="line.47"></a>
-<span class="sourceLineNo">048</span> */<a name="line.48"></a>
-<span class="sourceLineNo">049</span>@InterfaceAudience.Public<a name="line.49"></a>
-<span class="sourceLineNo">050</span>@InterfaceStability.Stable<a name="line.50"></a>
-<span class="sourceLineNo">051</span>public class TableOutputFormat<KEY> extends OutputFormat<KEY, Mutation><a name="line.51"></a>
-<span class="sourceLineNo">052</span>implements Configurable {<a name="line.52"></a>
-<span class="sourceLineNo">053</span><a name="line.53"></a>
-<span class="sourceLineNo">054</span> private static final Log LOG = LogFactory.getLog(TableOutputFormat.class);<a name="line.54"></a>
-<span class="sourceLineNo">055</span><a name="line.55"></a>
-<span class="sourceLineNo">056</span> /** Job parameter that specifies the output table. */<a name="line.56"></a>
-<span class="sourceLineNo">057</span> public static final String OUTPUT_TABLE = "hbase.mapred.outputtable";<a name="line.57"></a>
+<span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.TableNotEnabledException;<a name="line.32"></a>
+<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.TableNotFoundException;<a name="line.33"></a>
+<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.client.Admin;<a name="line.34"></a>
+<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.client.BufferedMutator;<a name="line.35"></a>
+<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.36"></a>
+<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.37"></a>
+<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.client.Delete;<a name="line.38"></a>
+<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.client.Mutation;<a name="line.39"></a>
+<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.client.Put;<a name="line.40"></a>
+<span class="sourceLineNo">041</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.41"></a>
+<span class="sourceLineNo">042</span>import org.apache.hadoop.mapreduce.OutputCommitter;<a name="line.42"></a>
+<span class="sourceLineNo">043</span>import org.apache.hadoop.mapreduce.OutputFormat;<a name="line.43"></a>
+<span class="sourceLineNo">044</span>import org.apache.hadoop.mapreduce.RecordWriter;<a name="line.44"></a>
+<span class="sourceLineNo">045</span>import org.apache.hadoop.mapreduce.TaskAttemptContext;<a name="line.45"></a>
+<span class="sourceLineNo">046</span><a name="line.46"></a>
+<span class="sourceLineNo">047</span>/**<a name="line.47"></a>
+<span class="sourceLineNo">048</span> * Convert Map/Reduce output and write it to an HBase table. The KEY is ignored<a name="line.48"></a>
+<span class="sourceLineNo">049</span> * while the output value <u>must</u> be either a {@link Put} or a<a name="line.49"></a>
+<span class="sourceLineNo">050</span> * {@link Delete} instance.<a name="line.50"></a>
+<span class="sourceLineNo">051</span> */<a name="line.51"></a>
+<span class="sourceLineNo">052</span>@InterfaceAudience.Public<a name="line.52"></a>
+<span class="sourceLineNo">053</span>@InterfaceStability.Stable<a name="line.53"></a>
+<span class="sourceLineNo">054</span>public class TableOutputFormat<KEY> extends OutputFormat<KEY, Mutation><a name="line.54"></a>
+<span class="sourceLineNo">055</span>implements Configurable {<a name="line.55"></a>
+<span class="sourceLineNo">056</span><a name="line.56"></a>
+<span class="sourceLineNo">057</span> private static final Log LOG = LogFactory.getLog(TableOutputFormat.class);<a name="line.57"></a>
<span class="sourceLineNo">058</span><a name="line.58"></a>
-<span class="sourceLineNo">059</span> /**<a name="line.59"></a>
-<span class="sourceLineNo">060</span> * Prefix for configuration property overrides to apply in {@link #setConf(Configuration)}.<a name="line.60"></a>
-<span class="sourceLineNo">061</span> * For keys matching this prefix, the prefix is stripped, and the value is set in the<a name="line.61"></a>
-<span class="sourceLineNo">062</span> * configuration with the resulting key, ie. the entry "hbase.mapred.output.key1 = value1"<a name="line.62"></a>
-<span class="sourceLineNo">063</span> * would be set in the configuration as "key1 = value1". Use this to set properties<a name="line.63"></a>
-<span class="sourceLineNo">064</span> * which should only be applied to the {@code TableOutputFormat} configuration and not the<a name="line.64"></a>
-<span class="sourceLineNo">065</span> * input configuration.<a name="line.65"></a>
-<span class="sourceLineNo">066</span> */<a name="line.66"></a>
-<span class="sourceLineNo">067</span> public static final String OUTPUT_CONF_PREFIX = "hbase.mapred.output.";<a name="line.67"></a>
-<span class="sourceLineNo">068</span><a name="line.68"></a>
-<span class="sourceLineNo">069</span> /**<a name="line.69"></a>
-<span class="sourceLineNo">070</span> * Optional job parameter to specify a peer cluster.<a name="line.70"></a>
-<span class="sourceLineNo">071</span> * Used specifying remote cluster when copying between hbase clusters (the<a name="line.71"></a>
-<span class="sourceLineNo">072</span> * source is picked up from <code>hbase-site.xml</code>).<a name="line.72"></a>
-<span class="sourceLineNo">073</span> * @see TableMapReduceUtil#initTableReducerJob(String, Class, org.apache.hadoop.mapreduce.Job, Class, String, String, String)<a name="line.73"></a>
-<span class="sourceLineNo">074</span> */<a name="line.74"></a>
-<span class="sourceLineNo">075</span> public static final String QUORUM_ADDRESS = OUTPUT_CONF_PREFIX + "quorum";<a name="line.75"></a>
-<span class="sourceLineNo">076</span><a name="line.76"></a>
-<span class="sourceLineNo">077</span> /** Optional job parameter to specify peer cluster's ZK client port */<a name="line.77"></a>
-<span class="sourceLineNo">078</span> public static final String QUORUM_PORT = OUTPUT_CONF_PREFIX + "quorum.port";<a name="line.78"></a>
+<span class="sourceLineNo">059</span> /** Job parameter that specifies the output table. */<a name="line.59"></a>
+<span class="sourceLineNo">060</span> public static final String OUTPUT_TABLE = "hbase.mapred.outputtable";<a name="line.60"></a>
+<span class="sourceLineNo">061</span><a name="line.61"></a>
+<span class="sourceLineNo">062</span> /**<a name="line.62"></a>
+<span class="sourceLineNo">063</span> * Prefix for configuration property overrides to apply in {@link #setConf(Configuration)}.<a name="line.63"></a>
+<span class="sourceLineNo">064</span> * For keys matching this prefix, the prefix is stripped, and the value is set in the<a name="line.64"></a>
+<span class="sourceLineNo">065</span> * configuration with the resulting key, ie. the entry "hbase.mapred.output.key1 = value1"<a name="line.65"></a>
+<span class="sourceLineNo">066</span> * would be set in the configuration as "key1 = value1". Use this to set properties<a name="line.66"></a>
+<span class="sourceLineNo">067</span> * which should only be applied to the {@code TableOutputFormat} configuration and not the<a name="line.67"></a>
+<span class="sourceLineNo">068</span> * input configuration.<a name="line.68"></a>
+<span class="sourceLineNo">069</span> */<a name="line.69"></a>
+<span class="sourceLineNo">070</span> public static final String OUTPUT_CONF_PREFIX = "hbase.mapred.output.";<a name="line.70"></a>
+<span class="sourceLineNo">071</span><a name="line.71"></a>
+<span class="sourceLineNo">072</span> /**<a name="line.72"></a>
+<span class="sourceLineNo">073</span> * Optional job parameter to specify a peer cluster.<a name="line.73"></a>
+<span class="sourceLineNo">074</span> * Used specifying remote cluster when copying between hbase clusters (the<a name="line.74"></a>
+<span class="sourceLineNo">075</span> * source is picked up from <code>hbase-site.xml</code>).<a name="line.75"></a>
+<span class="sourceLineNo">076</span> * @see TableMapReduceUtil#initTableReducerJob(String, Class, org.apache.hadoop.mapreduce.Job, Class, String, String, String)<a name="line.76"></a>
+<span class="sourceLineNo">077</span> */<a name="line.77"></a>
+<span class="sourceLineNo">078</span> public static final String QUORUM_ADDRESS = OUTPUT_CONF_PREFIX + "quorum";<a name="line.78"></a>
<span class="sourceLineNo">079</span><a name="line.79"></a>
-<span class="sourceLineNo">080</span> /** Optional specification of the rs class name of the peer cluster */<a name="line.80"></a>
-<span class="sourceLineNo">081</span> public static final String<a name="line.81"></a>
-<span class="sourceLineNo">082</span> REGION_SERVER_CLASS = OUTPUT_CONF_PREFIX + "rs.class";<a name="line.82"></a>
-<span class="sourceLineNo">083</span> /** Optional specification of the rs impl name of the peer cluster */<a name="line.83"></a>
+<span class="sourceLineNo">080</span> /** Optional job parameter to specify peer cluster's ZK client port */<a name="line.80"></a>
+<span class="sourceLineNo">081</span> public static final String QUORUM_PORT = OUTPUT_CONF_PREFIX + "quorum.port";<a name="line.81"></a>
+<span class="sourceLineNo">082</span><a name="line.82"></a>
+<span class="sourceLineNo">083</span> /** Optional specification of the rs class name of the peer cluster */<a name="line.83"></a>
<span class="sourceLineNo">084</span> public static final String<a name="line.84"></a>
-<span class="sourceLineNo">085</span> REGION_SERVER_IMPL = OUTPUT_CONF_PREFIX + "rs.impl";<a name="line.85"></a>
-<span class="sourceLineNo">086</span><a name="line.86"></a>
-<span class="sourceLineNo">087</span> /** The configuration. */<a name="line.87"></a>
-<span class="sourceLineNo">088</span> private Configuration conf = null;<a name="line.88"></a>
+<span class="sourceLineNo">085</span> REGION_SERVER_CLASS = OUTPUT_CONF_PREFIX + "rs.class";<a name="line.85"></a>
+<span class="sourceLineNo">086</span> /** Optional specification of the rs impl name of the peer cluster */<a name="line.86"></a>
+<span class="sourceLineNo">087</span> public static final String<a name="line.87"></a>
+<span class="sourceLineNo">088</span> REGION_SERVER_IMPL = OUTPUT_CONF_PREFIX + "rs.impl";<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 the reducer output to an HBase table.<a name="line.91"></a>
-<span class="sourceLineNo">092</span> */<a name="line.92"></a>
-<span class="sourceLineNo">093</span> protected class TableRecordWriter<a name="line.93"></a>
-<span class="sourceLineNo">094</span> extends RecordWriter<KEY, Mutation> {<a name="line.94"></a>
-<span class="sourceLineNo">095</span><a name="line.95"></a>
-<span class="sourceLineNo">096</span> private Connection connection;<a name="line.96"></a>
-<span class="sourceLineNo">097</span> private BufferedMutator mutator;<a name="line.97"></a>
+<span class="sourceLineNo">090</span> /** The configuration. */<a name="line.90"></a>
+<span class="sourceLineNo">091</span> private Configuration conf = null;<a name="line.91"></a>
+<span class="sourceLineNo">092</span><a name="line.92"></a>
+<span class="sourceLineNo">093</span> /**<a name="line.93"></a>
+<span class="sourceLineNo">094</span> * Writes the reducer output to an HBase table.<a name="line.94"></a>
+<span class="sourceLineNo">095</span> */<a name="line.95"></a>
+<span class="sourceLineNo">096</span> protected class TableRecordWriter<a name="line.96"></a>
+<span class="sourceLineNo">097</span> extends RecordWriter<KEY, Mutation> {<a name="line.97"></a>
<span class="sourceLineNo">098</span><a name="line.98"></a>
-<span class="sourceLineNo">099</span> /**<a name="line.99"></a>
-<span class="sourceLineNo">100</span> * @throws IOException <a name="line.100"></a>
-<span class="sourceLineNo">101</span> * <a name="line.101"></a>
-<span class="sourceLineNo">102</span> */<a name="line.102"></a>
-<span class="sourceLineNo">103</span> public TableRecordWriter() throws IOException {<a name="line.103"></a>
-<span class="sourceLineNo">104</span> String tableName = conf.get(OUTPUT_TABLE);<a name="line.104"></a>
-<span class="sourceLineNo">105</span> this.connection = ConnectionFactory.createConnection(conf);<a name="line.105"></a>
-<span class="sourceLineNo">106</span> this.mutator = connection.getBufferedMutator(TableName.valueOf(tableName));<a name="line.106"></a>
-<span class="sourceLineNo">107</span> LOG.info("Created table instance for " + tableName);<a name="line.107"></a>
-<span class="sourceLineNo">108</span> }<a name="line.108"></a>
-<span class="sourceLineNo">109</span> /**<a name="line.109"></a>
-<span class="sourceLineNo">110</span> * Closes the writer, in this case flush table commits.<a name="line.110"></a>
-<span class="sourceLineNo">111</span> *<a name="line.111"></a>
-<span class="sourceLineNo">112</span> * @param context The context.<a name="line.112"></a>
-<span class="sourceLineNo">113</span> * @throws IOException When closing the writer fails.<a name="line.113"></a>
-<span class="sourceLineNo">114</span> * @see RecordWriter#close(TaskAttemptContext)<a name="line.114"></a>
-<span class="sourceLineNo">115</span> */<a name="line.115"></a>
-<span class="sourceLineNo">116</span> @Override<a name="line.116"></a>
-<span class="sourceLineNo">117</span> public void close(TaskAttemptContext context)<a name="line.117"></a>
-<span class="sourceLineNo">118</span> throws IOException {<a name="line.118"></a>
-<span class="sourceLineNo">119</span> mutator.close();<a name="line.119"></a>
-<span class="sourceLineNo">120</span> connection.close();<a name="line.120"></a>
-<span class="sourceLineNo">121</span> }<a name="line.121"></a>
-<span class="sourceLineNo">122</span><a name="line.122"></a>
-<span class="sourceLineNo">123</span> /**<a name="line.123"></a>
-<span class="sourceLineNo">124</span> * Writes a key/value pair into the table.<a name="line.124"></a>
-<span class="sourceLineNo">125</span> *<a name="line.125"></a>
-<span class="sourceLineNo">126</span> * @param key The key.<a name="line.126"></a>
-<span class="sourceLineNo">127</span> * @param value The value.<a name="line.127"></a>
-<span class="sourceLineNo">128</span> * @throws IOException When writing fails.<a name="line.128"></a>
-<span class="sourceLineNo">129</span> * @see RecordWriter#write(Object, Object)<a name="line.129"></a>
-<span class="sourceLineNo">130</span> */<a name="line.130"></a>
-<span class="sourceLineNo">131</span> @Override<a name="line.131"></a>
-<span class="sourceLineNo">132</span> public void write(KEY key, Mutation value)<a name="line.132"></a>
-<span class="sourceLineNo">133</span> throws IOException {<a name="line.133"></a>
-<span class="sourceLineNo">134</span> if (!(value instanceof Put) && !(value instanceof Delete)) {<a name="line.134"></a>
-<span class="sourceLineNo">135</span> throw new IOException("Pass a Delete or a Put");<a name="line.135"></a>
-<span class="sourceLineNo">136</span> }<a name="line.136"></a>
-<span class="sourceLineNo">137</span> mutator.mutate(value);<a name="line.137"></a>
-<span class="sourceLineNo">138</span> }<a name="line.138"></a>
-<span class="sourceLineNo">139</span> }<a name="line.139"></a>
-<span class="sourceLineNo">140</span><a name="line.140"></a>
-<span class="sourceLineNo">141</span> /**<a name="line.141"></a>
-<span class="sourceLineNo">142</span> * Creates a new record writer.<a name="line.142"></a>
-<span class="sourceLineNo">143</span> * <a name="line.143"></a>
-<span class="sourceLineNo">144</span> * Be aware that the baseline javadoc gives the impression that there is a single<a name="line.144"></a>
-<span class="sourceLineNo">145</span> * {@link RecordWriter} per job but in HBase, it is more natural if we give you a new<a name="line.145"></a>
-<span class="sourceLineNo">146</span> * RecordWriter per call of this method. You must close the returned RecordWriter when done.<a name="line.146"></a>
-<span class="sourceLineNo">147</span> * Failure to do so will drop writes.<a name="line.147"></a>
-<span class="sourceLineNo">148</span> *<a name="line.148"></a>
-<span class="sourceLineNo">149</span> * @param context The current task context.<a name="line.149"></a>
-<span class="sourceLineNo">150</span> * @return The newly created writer instance.<a name="line.150"></a>
-<span class="sourceLineNo">151</span> * @throws IOException When creating the writer fails.<a name="line.151"></a>
-<span class="sourceLineNo">152</span> * @throws InterruptedException When the jobs is cancelled.<a name="line.152"></a>
-<span class="sourceLineNo">153</span> */<a name="line.153"></a>
-<span class="sourceLineNo">154</span> @Override<a name="line.154"></a>
-<span class="sourceLineNo">155</span> public RecordWriter<KEY, Mutation> getRecordWriter(TaskAttemptContext context)<a name="line.155"></a>
-<span class="sourceLineNo">156</span> throws IOException, InterruptedException {<a name="line.156"></a>
-<span class="sourceLineNo">157</span> return new TableRecordWriter();<a name="line.157"></a>
-<span class="sourceLineNo">158</span> }<a name="line.158"></a>
-<span class="sourceLineNo">159</span><a name="line.159"></a>
-<span class="sourceLineNo">160</span> /**<a name="line.160"></a>
-<span class="sourceLineNo">161</span> * Checks if the output target exists.<a name="line.161"></a>
-<span class="sourceLineNo">162</span> *<a name="line.162"></a>
-<span class="sourceLineNo">163</span> * @param context The current context.<a name="line.163"></a>
-<span class="sourceLineNo">164</span> * @throws IOException When the check fails.<a name="line.164"></a>
-<span class="sourceLineNo">165</span> * @throws InterruptedException When the job is aborted.<a name="line.165"></a>
-<span class="sourceLineNo">166</span> * @see OutputFormat#checkOutputSpecs(JobContext)<a name="line.166"></a>
-<span class="sourceLineNo">167</span> */<a name="line.167"></a>
-<span class="sourceLineNo">168</span> @Override<a name="line.168"></a>
-<span class="sourceLineNo">169</span> public void checkOutputSpecs(JobContext context) throws IOException,<a name="line.169"></a>
-<span class="sourceLineNo">170</span> InterruptedException {<a name="line.170"></a>
-<span class="sourceLineNo">171</span> // TODO Check if the table exists?<a name="line.171"></a>
-<span class="sourceLineNo">172</span><a name="line.172"></a>
-<span class="sourceLineNo">173</span> }<a name="line.173"></a>
+<span class="sourceLineNo">099</span> private Connection connection;<a name="line.99"></a>
+<span class="sourceLineNo">100</span> private BufferedMutator mutator;<a name="line.100"></a>
+<span class="sourceLineNo">101</span><a name="line.101"></a>
+<span class="sourceLineNo">102</span> /**<a name="line.102"></a>
+<span class="sourceLineNo">103</span> * @throws IOException <a name="line.103"></a>
+<span class="sourceLineNo">104</span> * <a name="line.104"></a>
+<span class="sourceLineNo">105</span> */<a name="line.105"></a>
+<span class="sourceLineNo">106</span> public TableRecordWriter() throws IOException {<a name="line.106"></a>
+<span class="sourceLineNo">107</span> String tableName = conf.get(OUTPUT_TABLE);<a name="line.107"></a>
+<span class="sourceLineNo">108</span> this.connection = ConnectionFactory.createConnection(conf);<a name="line.108"></a>
+<span class="sourceLineNo">109</span> this.mutator = connection.getBufferedMutator(TableName.valueOf(tableName));<a name="line.109"></a>
+<span class="sourceLineNo">110</span> LOG.info("Created table instance for " + tableName);<a name="line.110"></a>
+<span class="sourceLineNo">111</span> }<a name="line.111"></a>
+<span class="sourceLineNo">112</span> /**<a name="line.112"></a>
+<span class="sourceLineNo">113</span> * Closes the writer, in this case flush table commits.<a name="line.113"></a>
+<span class="sourceLineNo">114</span> *<a name="line.114"></a>
+<span class="sourceLineNo">115</span> * @param context The context.<a name="line.115"></a>
+<span class="sourceLineNo">116</span> * @throws IOException When closing the writer fails.<a name="line.116"></a>
+<span class="sourceLineNo">117</span> * @see RecordWriter#close(TaskAttemptContext)<a name="line.117"></a>
+<span class="sourceLineNo">118</span> */<a name="line.118"></a>
+<span class="sourceLineNo">119</span> @Override<a name="line.119"></a>
+<span class="sourceLineNo">120</span> public void close(TaskAttemptContext context)<a name="line.120"></a>
+<span class="sourceLineNo">121</span> throws IOException {<a name="line.121"></a>
+<span class="sourceLineNo">122</span> mutator.close();<a name="line.122"></a>
+<span class="sourceLineNo">123</span> connection.close();<a name="line.123"></a>
+<span class="sourceLineNo">124</span> }<a name="line.124"></a>
+<span class="sourceLineNo">125</span><a name="line.125"></a>
+<span class="sourceLineNo">126</span> /**<a name="line.126"></a>
+<span class="sourceLineNo">127</span> * Writes a key/value pair into the table.<a name="line.127"></a>
+<span class="sourceLineNo">128</span> *<a name="line.128"></a>
+<span class="sourceLineNo">129</span> * @param key The key.<a name="line.129"></a>
+<span class="sourceLineNo">130</span> * @param value The value.<a name="line.130"></a>
+<span class="sourceLineNo">131</span> * @throws IOException When writing fails.<a name="line.131"></a>
+<span class="sourceLineNo">132</span> * @see RecordWriter#write(Object, Object)<a name="line.132"></a>
+<span class="sourceLineNo">133</span> */<a name="line.133"></a>
+<span class="sourceLineNo">134</span> @Override<a name="line.134"></a>
+<span class="sourceLineNo">135</span> public void write(KEY key, Mutation value)<a name="line.135"></a>
+<span class="sourceLineNo">136</span> throws IOException {<a name="line.136"></a>
+<span class="sourceLineNo">137</span> if (!(value instanceof Put) && !(value instanceof Delete)) {<a name="line.137"></a>
+<span class="sourceLineNo">138</span> throw new IOException("Pass a Delete or a Put");<a name="line.138"></a>
+<span class="sourceLineNo">139</span> }<a name="line.139"></a>
+<span class="sourceLineNo">140</span> mutator.mutate(value);<a name="line.140"></a>
+<span class="sourceLineNo">141</span> }<a name="line.141"></a>
+<span class="sourceLineNo">142</span> }<a name="line.142"></a>
+<span class="sourceLineNo">143</span><a name="line.143"></a>
+<span class="sourceLineNo">144</span> /**<a name="line.144"></a>
+<span class="sourceLineNo">145</span> * Creates a new record writer.<a name="line.145"></a>
+<span class="sourceLineNo">146</span> * <a name="line.146"></a>
+<span class="sourceLineNo">147</span> * Be aware that the baseline javadoc gives the impression that there is a single<a name="line.147"></a>
+<span class="sourceLineNo">148</span> * {@link RecordWriter} per job but in HBase, it is more natural if we give you a new<a name="line.148"></a>
+<span class="sourceLineNo">149</span> * RecordWriter per call of this method. You must close the returned RecordWriter when done.<a name="line.149"></a>
+<span class="sourceLineNo">150</span> * Failure to do so will drop writes.<a name="line.150"></a>
+<span class="sourceLineNo">151</span> *<a name="line.151"></a>
+<span class="sourceLineNo">152</span> * @param context The current task context.<a name="line.152"></a>
+<span class="sourceLineNo">153</span> * @return The newly created writer instance.<a name="line.153"></a>
+<span class="sourceLineNo">154</span> * @throws IOException When creating the writer fails.<a name="line.154"></a>
+<span class="sourceLineNo">155</span> * @throws InterruptedException When the jobs is cancelled.<a name="line.155"></a>
+<span class="sourceLineNo">156</span> */<a name="line.156"></a>
+<span class="sourceLineNo">157</span> @Override<a name="line.157"></a>
+<span class="sourceLineNo">158</span> public RecordWriter<KEY, Mutation> getRecordWriter(TaskAttemptContext context)<a name="line.158"></a>
+<span class="sourceLineNo">159</span> throws IOException, InterruptedException {<a name="line.159"></a>
+<span class="sourceLineNo">160</span> return new TableRecordWriter();<a name="line.160"></a>
+<span class="sourceLineNo">161</span> }<a name="line.161"></a>
+<span class="sourceLineNo">162</span><a name="line.162"></a>
+<span class="sourceLineNo">163</span> /**<a name="line.163"></a>
+<span class="sourceLineNo">164</span> * Checks if the output table exists and is enabled.<a name="line.164"></a>
+<span class="sourceLineNo">165</span> *<a name="line.165"></a>
+<span class="sourceLineNo">166</span> * @param context The current context.<a name="line.166"></a>
+<span class="sourceLineNo">167</span> * @throws IOException When the check fails.<a name="line.167"></a>
+<span class="sourceLineNo">168</span> * @throws InterruptedException When the job is aborted.<a name="line.168"></a>
+<span class="sourceLineNo">169</span> * @see OutputFormat#checkOutputSpecs(JobContext)<a name="line.169"></a>
+<span class="sourceLineNo">170</span> */<a name="line.170"></a>
+<span class="sourceLineNo">171</span> @Override<a name="line.171"></a>
+<span class="sourceLineNo">172</span> public void checkOutputSpecs(JobContext context) throws IOException,<a name="line.172"></a>
+<span class="sourceLineNo">173</span> InterruptedException {<a name="line.173"></a>
<span class="sourceLineNo">174</span><a name="line.174"></a>
-<span class="sourceLineNo">175</span> /**<a name="line.175"></a>
-<span class="sourceLineNo">176</span> * Returns the output committer.<a name="line.176"></a>
-<span class="sourceLineNo">177</span> *<a name="line.177"></a>
-<span class="sourceLineNo">178</span> * @param context The current context.<a name="line.178"></a>
-<span class="sourceLineNo">179</span> * @return The committer.<a name="line.179"></a>
-<span class="sourceLineNo">180</span> * @throws IOException When creating the committer fails.<a name="line.180"></a>
-<span class="sourceLineNo">181</span> * @throws InterruptedException When the job is aborted.<a name="line.181"></a>
-<span class="sourceLineNo">182</span> * @see OutputFormat#getOutputCommitter(TaskAttemptContext)<a name="line.182"></a>
-<span class="sourceLineNo">183</span> */<a name="line.183"></a>
-<span class="sourceLineNo">184</span> @Override<a name="line.184"></a>
-<span class="sourceLineNo">185</span> public OutputCommitter getOutputCommitter(TaskAttemptContext context)<a name="line.185"></a>
-<span class="sourceLineNo">186</span> throws IOException, InterruptedException {<a name="line.186"></a>
-<span class="sourceLineNo">187</span> return new TableOutputCommitter();<a name="line.187"></a>
-<span class="sourceLineNo">188</span> }<a name="line.188"></a>
-<span class="sourceLineNo">189</span><a name="line.189"></a>
-<span class="sourceLineNo">190</span> @Override<a name="line.190"></a>
-<span class="sourceLineNo">191</span> public Configuration getConf() {<a name="line.191"></a>
-<span class="sourceLineNo">192</span> return conf;<a name="line.192"></a>
-<span class="sourceLineNo">193</span> }<a name="line.193"></a>
-<span class="sourceLineNo">194</span><a name="line.194"></a>
-<span class="sourceLineNo">195</span> @Override<a name="line.195"></a>
-<span class="sourceLineNo">196</span> public void setConf(Configuration otherConf) {<a name="line.196"></a>
-<span class="sourceLineNo">197</span> String tableName = otherConf.get(OUTPUT_TABLE);<a name="line.197"></a>
-<span class="sourceLineNo">198</span> if(tableName == null || tableName.length() <= 0) {<a name="line.198"></a>
-<span class="sourceLineNo">199</span> throw new IllegalArgumentException("Must specify table name");<a name="line.199"></a>
-<span class="sourceLineNo">200</span> }<a name="line.200"></a>
-<span class="sourceLineNo">201</span><a name="line.201"></a>
-<span class="sourceLineNo">202</span> String address = otherConf.get(QUORUM_ADDRESS);<a name="line.202"></a>
-<span class="sourceLineNo">203</span> int zkClientPort = otherConf.getInt(QUORUM_PORT, 0);<a name="line.203"></a>
-<span class="sourceLineNo">204</span> String serverClass = otherConf.get(REGION_SERVER_CLASS);<a name="line.204"></a>
-<span class="sourceLineNo">205</span> String serverImpl = otherConf.get(REGION_SERVER_IMPL);<a name="line.205"></a>
-<span class="sourceLineNo">206</span><a name="line.206"></a>
-<span class="sourceLineNo">207</span> try {<a name="line.207"></a>
-<span class="sourceLineNo">208</span> this.conf = HBaseConfiguration.createClusterConf(otherConf, address, OUTPUT_CONF_PREFIX);<a name="line.208"></a>
-<span class="sourceLineNo">209</span><a name="line.209"></a>
-<span class="sourceLineNo">210</span> if (serverClass != null) {<a name="line.210"></a>
-<span class="sourceLineNo">211</span> this.conf.set(HConstants.REGION_SERVER_IMPL, serverImpl);<a name="line.211"></a>
-<span class="sourceLineNo">212</span> }<a name="line.212"></a>
-<span class="sourceLineNo">213</span> if (zkClientPort != 0) {<a name="line.213"></a>
-<span class="sourceLineNo">214</span> this.conf.setInt(HConstants.ZOOKEEPER_CLIENT_PORT, zkClientPort);<a name="line.214"></a>
-<span class="sourceLineNo">215</span> }<a name="line.215"></a>
-<span class="sourceLineNo">216</span> } catch(IOException e) {<a name="line.216"></a>
-<span class="sourceLineNo">217</span> LOG.error(e);<a name="line.217"></a>
-<span class="sourceLineNo">218</span> throw new RuntimeException(e);<a name="line.218"></a>
-<span class="sourceLineNo">219</span> }<a name="line.219"></a>
-<span class="sourceLineNo">220</span> }<a name="line.220"></a>
-<span class="sourceLineNo">221</span>}<a name="line.221"></a>
+<span class="sourceLineNo">175</span> try (Admin admin = ConnectionFactory.createConnection(getConf()).getAdmin()) {<a name="line.175"></a>
+<span class="sourceLineNo">176</span> TableName tableName = TableName.valueOf(this.conf.get(OUTPUT_TABLE));<a name="line.176"></a>
+<span class="sourceLineNo">177</span> if (!admin.tableExists(tableName)) {<a name="line.177"></a>
+<span class="sourceLineNo">178</span> throw new TableNotFoundException("Can't write, table does not exist:" +<a name="line.178"></a>
+<span class="sourceLineNo">179</span> tableName.getNameAsString());<a name="line.179"></a>
+<span class="sourceLineNo">180</span> }<a name="line.180"></a>
+<span class="sourceLineNo">181</span><a name="line.181"></a>
+<span class="sourceLineNo">182</span> if (!admin.isTableEnabled(tableName)) {<a name="line.182"></a>
+<span class="sourceLineNo">183</span> throw new TableNotEnabledException("Can't write, table is not enabled: " +<a name="line.183"></a>
+<span class="sourceLineNo">184</span> tableName.getNameAsString());<a name="line.184"></a>
+<span class="sourceLineNo">185</span> }<a name="line.185"></a>
+<span class="sourceLineNo">186</span> }<a name="line.186"></a>
+<span class="sourceLineNo">187</span> }<a name="line.187"></a>
+<span class="sourceLineNo">188</span><a name="line.188"></a>
+<span class="sourceLineNo">189</span> /**<a name="line.189"></a>
+<span class="sourceLineNo">190</span> * Returns the output committer.<a name="line.190"></a>
+<span class="sourceLineNo">191</span> *<a name="line.191"></a>
+<span class="sourceLineNo">192</span> * @param context The current context.<a name="line.192"></a>
+<span class="sourceLineNo">193</span> * @return The committer.<a name="line.193"></a>
+<span class="sourceLineNo">194</span> * @throws IOException When creating the committer fails.<a name="line.194"></a>
+<span class="sourceLineNo">195</span> * @throws InterruptedException When the job is aborted.<a name="line.195"></a>
+<span class="sourceLineNo">196</span> * @see OutputFormat#getOutputCommitter(TaskAttemptContext)<a name="line.196"></a>
+<span class="sourceLineNo">197</span> */<a name="line.197"></a>
+<span class="sourceLineNo">198</span> @Override<a name="line.198"></a>
+<span class="sourceLineNo">199</span> public OutputCommitter getOutputCommitter(TaskAttemptContext context)<a name="line.199"></a>
+<span class="sourceLineNo">200</span> throws IOException, InterruptedException {<a name="line.200"></a>
+<span class="sourceLineNo">201</span> return new TableOutputCommitter();<a name="line.201"></a>
+<span class="sourceLineNo">202</span> }<a name="line.202"></a>
+<span class="sourceLineNo">203</span><a name="line.203"></a>
+<span class="sourceLineNo">204</span> @Override<a name="line.204"></a>
+<span class="sourceLineNo">205</span> public Configuration getConf() {<a name="line.205"></a>
+<span class="sourceLineNo">206</span> return conf;<a name="line.206"></a>
+<span class="sourceLineNo">207</span> }<a name="line.207"></a>
+<span class="sourceLineNo">208</span><a name="line.208"></a>
+<span class="sourceLineNo">209</span> @Override<a name="line.209"></a>
+<span class="sourceLineNo">210</span> public void setConf(Configuration otherConf) {<a name="line.210"></a>
+<span class="sourceLineNo">211</span> String tableName = otherConf.get(OUTPUT_TABLE);<a name="line.211"></a>
+<span class="sourceLineNo">212</span> if(tableName == null || tableName.length() <= 0) {<a name="line.212"></a>
+<span class="sourceLineNo">213</span> throw new IllegalArgumentException("Must specify table name");<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> String address = otherConf.get(QUORUM_ADDRESS);<a name="line.216"></a>
+<span class="sourceLineNo">217</span> int zkClientPort = otherConf.getInt(QUORUM_PORT, 0);<a name="line.217"></a>
+<span class="sourceLineNo">218</span> String serverClass = otherConf.get(REGION_SERVER_CLASS);<a name="line.218"></a>
+<span class="sourceLineNo">219</span> String serverImpl = otherConf.get(REGION_SERVER_IMPL);<a name="line.219"></a>
+<span class="sourceLineNo">220</span><a name="line.220"></a>
+<span class="sourceLineNo">221</span> try {<a name="line.221"></a>
+<span class="sourceLineNo">222</span> this.conf = HBaseConfiguration.createClusterConf(otherConf, address, OUTPUT_CONF_PREFIX);<a name="line.222"></a>
+<span class="sourceLineNo">223</span><a name="line.223"></a>
+<span class="sourceLineNo">224</span> if (serverClass != null) {<a name="line.224"></a>
+<span class="sourceLineNo">225</span> this.conf.set(HConstants.REGION_SERVER_IMPL, serverImpl);<a name="line.225"></a>
+<span class="sourceLineNo">226</span> }<a name="line.226"></a>
+<span class="sourceLineNo">227</span> if (zkClientPort != 0) {<a name="line.227"></a>
+<span class="sourceLineNo">228</span> this.conf.setInt(HConstants.ZOOKEEPER_CLIENT_PORT, zkClientPort);<a name="line.228"></a>
+<span class="sourceLineNo">229</span> }<a name="line.229"></a>
+<span class="sourceLineNo">230</span> } catch(IOException e) {<a name="line.230"></a>
+<span class="sourceLineNo">231</span> LOG.error(e);<a name="line.231"></a>
+<span class="sourceLineNo">232</span> throw new RuntimeException(e);<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>}<a name="line.235"></a>
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/a0cdd2a9/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TextSortReducer.html
----------------------------------------------------------------------
diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TextSortReducer.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TextSortReducer.html
index f3aa9a6..6ab4f9e 100644
--- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TextSortReducer.html
+++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TextSortReducer.html
@@ -109,120 +109,118 @@
<span class="sourceLineNo">101</span> */<a name="line.101"></a>
<span class="sourceLineNo">102</span> @Override<a name="line.102"></a>
<span class="sourceLineNo">103</span> protected void setup(Context context) {<a name="line.103"></a>
-<span class="sourceLineNo">104</span> doSetup(context);<a name="line.104"></a>
-<span class="sourceLineNo">105</span><a name="line.105"></a>
-<span class="sourceLineNo">106</span> Configuration conf = context.getConfiguration();<a name="line.106"></a>
-<span class="sourceLineNo">107</span><a name="line.107"></a>
-<span class="sourceLineNo">108</span> parser = new ImportTsv.TsvParser(conf.get(ImportTsv.COLUMNS_CONF_KEY), separator);<a name="line.108"></a>
-<span class="sourceLineNo">109</span> if (parser.getRowKeyColumnIndex() == -1) {<a name="line.109"></a>
-<span class="sourceLineNo">110</span> throw new RuntimeException("No row key column specified");<a name="line.110"></a>
-<span class="sourceLineNo">111</span> }<a name="line.111"></a>
-<span class="sourceLineNo">112</span> this.kvCreator = new CellCreator(conf);<a name="line.112"></a>
-<span class="sourceLineNo">113</span> }<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> * Handles common parameter initialization that a subclass might want to leverage.<a name="line.116"></a>
-<span class="sourceLineNo">117</span> * @param context<a name="line.117"></a>
+<span class="sourceLineNo">104</span> Configuration conf = context.getConfiguration();<a name="line.104"></a>
+<span class="sourceLineNo">105</span> doSetup(context, conf);<a name="line.105"></a>
+<span class="sourceLineNo">106</span><a name="line.106"></a>
+<span class="sourceLineNo">107</span> parser = new ImportTsv.TsvParser(conf.get(ImportTsv.COLUMNS_CONF_KEY), separator);<a name="line.107"></a>
+<span class="sourceLineNo">108</span> if (parser.getRowKeyColumnIndex() == -1) {<a name="line.108"></a>
+<span class="sourceLineNo">109</span> throw new RuntimeException("No row key column specified");<a name="line.109"></a>
+<span class="sourceLineNo">110</span> }<a name="line.110"></a>
+<span class="sourceLineNo">111</span> this.kvCreator = new CellCreator(conf);<a name="line.111"></a>
+<span class="sourceLineNo">112</span> }<a name="line.112"></a>
+<span class="sourceLineNo">113</span><a name="line.113"></a>
+<span class="sourceLineNo">114</span> /**<a name="line.114"></a>
+<span class="sourceLineNo">115</span> * Handles common parameter initialization that a subclass might want to leverage.<a name="line.115"></a>
+<span class="sourceLineNo">116</span> * @param context<a name="line.116"></a>
+<span class="sourceLineNo">117</span> * @param conf<a name="line.117"></a>
<span class="sourceLineNo">118</span> */<a name="line.118"></a>
-<span class="sourceLineNo">119</span> protected void doSetup(Context context) {<a name="line.119"></a>
-<span class="sourceLineNo">120</span> Configuration conf = context.getConfiguration();<a name="line.120"></a>
-<span class="sourceLineNo">121</span><a name="line.121"></a>
-<span class="sourceLineNo">122</span> // If a custom separator has been used,<a name="line.122"></a>
-<span class="sourceLineNo">123</span> // decode it back from Base64 encoding.<a name="line.123"></a>
-<span class="sourceLineNo">124</span> separator = conf.get(ImportTsv.SEPARATOR_CONF_KEY);<a name="line.124"></a>
-<span class="sourceLineNo">125</span> if (separator == null) {<a name="line.125"></a>
-<span class="sourceLineNo">126</span> separator = ImportTsv.DEFAULT_SEPARATOR;<a name="line.126"></a>
-<span class="sourceLineNo">127</span> } else {<a name="line.127"></a>
-<span class="sourceLineNo">128</span> separator = new String(Base64.decode(separator));<a name="line.128"></a>
-<span class="sourceLineNo">129</span> }<a name="line.129"></a>
-<span class="sourceLineNo">130</span><a name="line.130"></a>
-<span class="sourceLineNo">131</span> // Should never get 0 as we are setting this to a valid value in job configuration.<a name="line.131"></a>
-<span class="sourceLineNo">132</span> ts = conf.getLong(ImportTsv.TIMESTAMP_CONF_KEY, 0);<a name="line.132"></a>
-<span class="sourceLineNo">133</span><a name="line.133"></a>
-<span class="sourceLineNo">134</span> skipBadLines = context.getConfiguration().getBoolean(ImportTsv.SKIP_LINES_CONF_KEY, true);<a name="line.134"></a>
-<span class="sourceLineNo">135</span> badLineCount = context.getCounter("ImportTsv", "Bad Lines");<a name="line.135"></a>
-<span class="sourceLineNo">136</span> }<a name="line.136"></a>
-<span class="sourceLineNo">137</span> <a name="line.137"></a>
-<span class="sourceLineNo">138</span> @Override<a name="line.138"></a>
-<span class="sourceLineNo">139</span> protected void reduce(<a name="line.139"></a>
-<span class="sourceLineNo">140</span> ImmutableBytesWritable rowKey,<a name="line.140"></a>
-<span class="sourceLineNo">141</span> java.lang.Iterable<Text> lines,<a name="line.141"></a>
-<span class="sourceLineNo">142</span> Reducer<ImmutableBytesWritable, Text,<a name="line.142"></a>
-<span class="sourceLineNo">143</span> ImmutableBytesWritable, KeyValue>.Context context)<a name="line.143"></a>
-<span class="sourceLineNo">144</span> throws java.io.IOException, InterruptedException<a name="line.144"></a>
-<span class="sourceLineNo">145</span> {<a name="line.145"></a>
-<span class="sourceLineNo">146</span> // although reduce() is called per-row, handle pathological case<a name="line.146"></a>
-<span class="sourceLineNo">147</span> long threshold = context.getConfiguration().getLong(<a name="line.147"></a>
-<span class="sourceLineNo">148</span> "reducer.row.threshold", 1L * (1<<30));<a name="line.148"></a>
-<span class="sourceLineNo">149</span> Iterator<Text> iter = lines.iterator();<a name="line.149"></a>
-<span class="sourceLineNo">150</span> while (iter.hasNext()) {<a name="line.150"></a>
-<span class="sourceLineNo">151</span> Set<KeyValue> kvs = new TreeSet<KeyValue>(CellComparator.COMPARATOR);<a name="line.151"></a>
-<span class="sourceLineNo">152</span> long curSize = 0;<a name="line.152"></a>
-<span class="sourceLineNo">153</span> // stop at the end or the RAM threshold<a name="line.153"></a>
-<span class="sourceLineNo">154</span> while (iter.hasNext() && curSize < threshold) {<a name="line.154"></a>
-<span class="sourceLineNo">155</span> Text line = iter.next();<a name="line.155"></a>
-<span class="sourceLineNo">156</span> byte[] lineBytes = line.getBytes();<a name="line.156"></a>
-<span class="sourceLineNo">157</span> try {<a name="line.157"></a>
-<span class="sourceLineNo">158</span> ImportTsv.TsvParser.ParsedLine parsed = parser.parse(lineBytes, line.getLength());<a name="line.158"></a>
-<span class="sourceLineNo">159</span> // Retrieve timestamp if exists<a name="line.159"></a>
-<span class="sourceLineNo">160</span> ts = parsed.getTimestamp(ts);<a name="line.160"></a>
-<span class="sourceLineNo">161</span> cellVisibilityExpr = parsed.getCellVisibility();<a name="line.161"></a>
-<span class="sourceLineNo">162</span> ttl = parsed.getCellTTL();<a name="line.162"></a>
-<span class="sourceLineNo">163</span> <a name="line.163"></a>
-<span class="sourceLineNo">164</span> // create tags for the parsed line<a name="line.164"></a>
-<span class="sourceLineNo">165</span> List<Tag> tags = new ArrayList<Tag>();<a name="line.165"></a>
-<span class="sourceLineNo">166</span> if (cellVisibilityExpr != null) {<a name="line.166"></a>
-<span class="sourceLineNo">167</span> tags.addAll(kvCreator.getVisibilityExpressionResolver().createVisibilityExpTags(<a name="line.167"></a>
-<span class="sourceLineNo">168</span> cellVisibilityExpr));<a name="line.168"></a>
-<span class="sourceLineNo">169</span> }<a name="line.169"></a>
-<span class="sourceLineNo">170</span> // Add TTL directly to the KV so we can vary them when packing more than one KV<a name="line.170"></a>
-<span class="sourceLineNo">171</span> // into puts<a name="line.171"></a>
-<span class="sourceLineNo">172</span> if (ttl > 0) {<a name="line.172"></a>
-<span class="sourceLineNo">173</span> tags.add(new ArrayBackedTag(TagType.TTL_TAG_TYPE, Bytes.toBytes(ttl)));<a name="line.173"></a>
-<span class="sourceLineNo">174</span> }<a name="line.174"></a>
-<span class="sourceLineNo">175</span> for (int i = 0; i < parsed.getColumnCount(); i++) {<a name="line.175"></a>
-<span class="sourceLineNo">176</span> if (i == parser.getRowKeyColumnIndex() || i == parser.getTimestampKeyColumnIndex()<a name="line.176"></a>
-<span class="sourceLineNo">177</span> || i == parser.getAttributesKeyColumnIndex() || i == parser.getCellVisibilityColumnIndex()<a name="line.177"></a>
-<span class="sourceLineNo">178</span> || i == parser.getCellTTLColumnIndex()) {<a name="line.178"></a>
-<span class="sourceLineNo">179</span> continue;<a name="line.179"></a>
-<span class="sourceLineNo">180</span> }<a name="line.180"></a>
-<span class="sourceLineNo">181</span> // Creating the KV which needs to be directly written to HFiles. Using the Facade<a name="line.181"></a>
-<span class="sourceLineNo">182</span> // KVCreator for creation of kvs.<a name="line.182"></a>
-<span class="sourceLineNo">183</span> Cell cell = this.kvCreator.create(lineBytes, parsed.getRowKeyOffset(),<a name="line.183"></a>
-<span class="sourceLineNo">184</span> parsed.getRowKeyLength(), parser.getFamily(i), 0, parser.getFamily(i).length,<a name="line.184"></a>
-<span class="sourceLineNo">185</span> parser.getQualifier(i), 0, parser.getQualifier(i).length, ts, lineBytes,<a name="line.185"></a>
-<span class="sourceLineNo">186</span> parsed.getColumnOffset(i), parsed.getColumnLength(i), tags);<a name="line.186"></a>
-<span class="sourceLineNo">187</span> KeyValue kv = KeyValueUtil.ensureKeyValue(cell);<a name="line.187"></a>
-<span class="sourceLineNo">188</span> kvs.add(kv);<a name="line.188"></a>
-<span class="sourceLineNo">189</span> curSize += kv.heapSize();<a name="line.189"></a>
-<span class="sourceLineNo">190</span> }<a name="line.190"></a>
-<span class="sourceLineNo">191</span> } catch (ImportTsv.TsvParser.BadTsvLineException | IllegalArgumentException<a name="line.191"></a>
-<span class="sourceLineNo">192</span> | InvalidLabelException badLine) {<a name="line.192"></a>
-<span class="sourceLineNo">193</span> if (skipBadLines) {<a name="line.193"></a>
-<span class="sourceLineNo">194</span> System.err.println("Bad line." + badLine.getMessage());<a name="line.194"></a>
-<span class="sourceLineNo">195</span> incrementBadLineCount(1);<a name="line.195"></a>
-<span class="sourceLineNo">196</span> continue;<a name="line.196"></a>
-<span class="sourceLineNo">197</span> }<a name="line.197"></a>
-<span class="sourceLineNo">198</span> throw new IOException(badLine);<a name="line.198"></a>
-<span class="sourceLineNo">199</span> }<a name="line.199"></a>
-<span class="sourceLineNo">200</span> }<a name="line.200"></a>
-<span class="sourceLineNo">201</span> context.setStatus("Read " + kvs.size() + " entries of " + kvs.getClass()<a name="line.201"></a>
-<span class="sourceLineNo">202</span> + "(" + StringUtils.humanReadableInt(curSize) + ")");<a name="line.202"></a>
-<span class="sourceLineNo">203</span> int index = 0;<a name="line.203"></a>
-<span class="sourceLineNo">204</span> for (KeyValue kv : kvs) {<a name="line.204"></a>
-<span class="sourceLineNo">205</span> context.write(rowKey, kv);<a name="line.205"></a>
-<span class="sourceLineNo">206</span> if (++index > 0 && index % 100 == 0)<a name="line.206"></a>
-<span class="sourceLineNo">207</span> context.setStatus("Wrote " + index + " key values.");<a name="line.207"></a>
-<span class="sourceLineNo">208</span> }<a name="line.208"></a>
-<span class="sourceLineNo">209</span><a name="line.209"></a>
-<span class="sourceLineNo">210</span> // if we have more entries to process<a name="line.210"></a>
-<span class="sourceLineNo">211</span> if (iter.hasNext()) {<a name="line.211"></a>
-<span class="sourceLineNo">212</span> // force flush because we cannot guarantee intra-row sorted order<a name="line.212"></a>
-<span class="sourceLineNo">213</span> context.write(null, 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> }<a name="line.216"></a>
-<span class="sourceLineNo">217</span>}<a name="line.217"></a>
+<span class="sourceLineNo">119</span> protected void doSetup(Context context, Configuration conf) {<a name="line.119"></a>
+<span class="sourceLineNo">120</span> // If a custom separator has been used,<a name="line.120"></a>
+<span class="sourceLineNo">121</span> // decode it back from Base64 encoding.<a name="line.121"></a>
+<span class="sourceLineNo">122</span> separator = conf.get(ImportTsv.SEPARATOR_CONF_KEY);<a name="line.122"></a>
+<span class="sourceLineNo">123</span> if (separator == null) {<a name="line.123"></a>
+<span class="sourceLineNo">124</span> separator = ImportTsv.DEFAULT_SEPARATOR;<a name="line.124"></a>
+<span class="sourceLineNo">125</span> } else {<a name="line.125"></a>
+<span class="sourceLineNo">126</span> separator = new String(Base64.decode(separator));<a name="line.126"></a>
+<span class="sourceLineNo">127</span> }<a name="line.127"></a>
+<span class="sourceLineNo">128</span><a name="line.128"></a>
+<span class="sourceLineNo">129</span> // Should never get 0 as we are setting this to a valid value in job configuration.<a name="line.129"></a>
+<span class="sourceLineNo">130</span> ts = conf.getLong(ImportTsv.TIMESTAMP_CONF_KEY, 0);<a name="line.130"></a>
+<span class="sourceLineNo">131</span><a name="line.131"></a>
+<span class="sourceLineNo">132</span> skipBadLines = context.getConfiguration().getBoolean(ImportTsv.SKIP_LINES_CONF_KEY, true);<a name="line.132"></a>
+<span class="sourceLineNo">133</span> badLineCount = context.getCounter("ImportTsv", "Bad Lines");<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> @Override<a name="line.136"></a>
+<span class="sourceLineNo">137</span> protected void reduce(<a name="line.137"></a>
+<span class="sourceLineNo">138</span> ImmutableBytesWritable rowKey,<a name="line.138"></a>
+<span class="sourceLineNo">139</span> java.lang.Iterable<Text> lines,<a name="line.139"></a>
+<span class="sourceLineNo">140</span> Reducer<ImmutableBytesWritable, Text,<a name="line.140"></a>
+<span class="sourceLineNo">141</span> ImmutableBytesWritable, KeyValue>.Context context)<a name="line.141"></a>
+<span class="sourceLineNo">142</span> throws java.io.IOException, InterruptedException<a name="line.142"></a>
+<span class="sourceLineNo">143</span> {<a name="line.143"></a>
+<span class="sourceLineNo">144</span> // although reduce() is called per-row, handle pathological case<a name="line.144"></a>
+<span class="sourceLineNo">145</span> long threshold = context.getConfiguration().getLong(<a name="line.145"></a>
+<span class="sourceLineNo">146</span> "reducer.row.threshold", 1L * (1<<30));<a name="line.146"></a>
+<span class="sourceLineNo">147</span> Iterator<Text> iter = lines.iterator();<a name="line.147"></a>
+<span class="sourceLineNo">148</span> while (iter.hasNext()) {<a name="line.148"></a>
+<span class="sourceLineNo">149</span> Set<KeyValue> kvs = new TreeSet<KeyValue>(CellComparator.COMPARATOR);<a name="line.149"></a>
+<span class="sourceLineNo">150</span> long curSize = 0;<a name="line.150"></a>
+<span class="sourceLineNo">151</span> // stop at the end or the RAM threshold<a name="line.151"></a>
+<span class="sourceLineNo">152</span> while (iter.hasNext() && curSize < threshold) {<a name="line.152"></a>
+<span class="sourceLineNo">153</span> Text line = iter.next();<a name="line.153"></a>
+<span class="sourceLineNo">154</span> byte[] lineBytes = line.getBytes();<a name="line.154"></a>
+<span class="sourceLineNo">155</span> try {<a name="line.155"></a>
+<span class="sourceLineNo">156</span> ImportTsv.TsvParser.ParsedLine parsed = parser.parse(lineBytes, line.getLength());<a name="line.156"></a>
+<span class="sourceLineNo">157</span> // Retrieve timestamp if exists<a name="line.157"></a>
+<span class="sourceLineNo">158</span> ts = parsed.getTimestamp(ts);<a name="line.158"></a>
+<span class="sourceLineNo">159</span> cellVisibilityExpr = parsed.getCellVisibility();<a name="line.159"></a>
+<span class="sourceLineNo">160</span> ttl = parsed.getCellTTL();<a name="line.160"></a>
+<span class="sourceLineNo">161</span> <a name="line.161"></a>
+<span class="sourceLineNo">162</span> // create tags for the parsed line<a name="line.162"></a>
+<span class="sourceLineNo">163</span> List<Tag> tags = new ArrayList<Tag>();<a name="line.163"></a>
+<span class="sourceLineNo">164</span> if (cellVisibilityExpr != null) {<a name="line.164"></a>
+<span class="sourceLineNo">165</span> tags.addAll(kvCreator.getVisibilityExpressionResolver().createVisibilityExpTags(<a name="line.165"></a>
+<span class="sourceLineNo">166</span> cellVisibilityExpr));<a name="line.166"></a>
+<span class="sourceLineNo">167</span> }<a name="line.167"></a>
+<span class="sourceLineNo">168</span> // Add TTL directly to the KV so we can vary them when packing more than one KV<a name="line.168"></a>
+<span class="sourceLineNo">169</span> // into puts<a name="line.169"></a>
+<span class="sourceLineNo">170</span> if (ttl > 0) {<a name="line.170"></a>
+<span class="sourceLineNo">171</span> tags.add(new ArrayBackedTag(TagType.TTL_TAG_TYPE, Bytes.toBytes(ttl)));<a name="line.171"></a>
+<span class="sourceLineNo">172</span> }<a name="line.172"></a>
+<span class="sourceLineNo">173</span> for (int i = 0; i < parsed.getColumnCount(); i++) {<a name="line.173"></a>
+<span class="sourceLineNo">174</span> if (i == parser.getRowKeyColumnIndex() || i == parser.getTimestampKeyColumnIndex()<a name="line.174"></a>
+<span class="sourceLineNo">175</span> || i == parser.getAttributesKeyColumnIndex() || i == parser.getCellVisibilityColumnIndex()<a name="line.175"></a>
+<span class="sourceLineNo">176</span> || i == parser.getCellTTLColumnIndex()) {<a name="line.176"></a>
+<span class="sourceLineNo">177</span> continue;<a name="line.177"></a>
+<span class="sourceLineNo">178</span> }<a name="line.178"></a>
+<span class="sourceLineNo">179</span> // Creating the KV which needs to be directly written to HFiles. Using the Facade<a name="line.179"></a>
+<span class="sourceLineNo">180</span> // KVCreator for creation of kvs.<a name="line.180"></a>
+<span class="sourceLineNo">181</span> Cell cell = this.kvCreator.create(lineBytes, parsed.getRowKeyOffset(),<a name="line.181"></a>
+<span class="sourceLineNo">182</span> parsed.getRowKeyLength(), parser.getFamily(i), 0, parser.getFamily(i).length,<a name="line.182"></a>
+<span class="sourceLineNo">183</span> parser.getQualifier(i), 0, parser.getQualifier(i).length, ts, lineBytes,<a name="line.183"></a>
+<span class="sourceLineNo">184</span> parsed.getColumnOffset(i), parsed.getColumnLength(i), tags);<a name="line.184"></a>
+<span class="sourceLineNo">185</span> KeyValue kv = KeyValueUtil.ensureKeyValue(cell);<a name="line.185"></a>
+<span class="sourceLineNo">186</span> kvs.add(kv);<a name="line.186"></a>
+<span class="sourceLineNo">187</span> curSize += kv.heapSize();<a name="line.187"></a>
+<span class="sourceLineNo">188</span> }<a name="line.188"></a>
+<span class="sourceLineNo">189</span> } catch (ImportTsv.TsvParser.BadTsvLineException | IllegalArgumentException<a name="line.189"></a>
+<span class="sourceLineNo">190</span> | InvalidLabelException badLine) {<a name="line.190"></a>
+<span class="sourceLineNo">191</span> if (skipBadLines) {<a name="line.191"></a>
+<span class="sourceLineNo">192</span> System.err.println("Bad line." + badLine.getMessage());<a name="line.192"></a>
+<span class="sourceLineNo">193</span> incrementBadLineCount(1);<a name="line.193"></a>
+<span class="sourceLineNo">194</span> continue;<a name="line.194"></a>
+<span class="sourceLineNo">195</span> }<a name="line.195"></a>
+<span class="sourceLineNo">196</span> throw new IOException(badLine);<a name="line.196"></a>
+<span class="sourceLineNo">197</span> }<a name="line.197"></a>
+<span class="sourceLineNo">198</span> }<a name="line.198"></a>
+<span class="sourceLineNo">199</span> context.setStatus("Read " + kvs.size() + " entries of " + kvs.getClass()<a name="line.199"></a>
+<span class="sourceLineNo">200</span> + "(" + StringUtils.humanReadableInt(curSize) + ")");<a name="line.200"></a>
+<span class="sourceLineNo">201</span> int index = 0;<a name="line.201"></a>
+<span class="sourceLineNo">202</span> for (KeyValue kv : kvs) {<a name="line.202"></a>
+<span class="sourceLineNo">203</span> context.write(rowKey, kv);<a name="line.203"></a>
+<span class="sourceLineNo">204</span> if (++index > 0 && index % 100 == 0)<a name="line.204"></a>
+<span class="sourceLineNo">205</span> context.setStatus("Wrote " + index + " key values.");<a name="line.205"></a>
+<span class="sourceLineNo">206</span> }<a name="line.206"></a>
+<span class="sourceLineNo">207</span><a name="line.207"></a>
+<span class="sourceLineNo">208</span> // if we have more entries to process<a name="line.208"></a>
+<span class="sourceLineNo">209</span> if (iter.hasNext()) {<a name="line.209"></a>
+<span class="sourceLineNo">210</span> // force flush because we cannot guarantee intra-row sorted order<a name="line.210"></a>
+<span class="sourceLineNo">211</span> context.write(null, null);<a name="line.211"></a>
+<span class="sourceLineNo">212</span> }<a name="line.212"></a>
+<span class="sourceLineNo">213</span> }<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>
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/a0cdd2a9/book.html
----------------------------------------------------------------------
diff --git a/book.html b/book.html
index 2a6a527..8244382 100644
--- a/book.html
+++ b/book.html
@@ -4743,7 +4743,7 @@ Configuration that it is thought rare anyone would change can exist only in code
<dd>
<div class="paragraph">
<div class="title">Description</div>
-<p>Whether asynchronous WAL replication to the secondary region replicas is enabled or not. If this is enabled, a replication peer named "region_replica_replication" will be created which will tail the logs and replicate the mutations to region replicas for tables that have region replication > 1. If this is enabled once, disabling this replication also requires disabling the replication peer using shell or ReplicationAdmin java class. Replication to secondary region replicas works over standard inter-cluster replication. So replication, if disabled explicitly, also has to be enabled by setting "hbase.replication" to true for this feature to work.</p>
+<p>Whether asynchronous WAL replication to the secondary region replicas is enabled or not. If this is enabled, a replication peer named "region_replica_replication" will be created which will tail the logs and replicate the mutations to region replicas for tables that have region replication > 1. If this is enabled once, disabling this replication also requires disabling the replication peer using shell or ReplicationAdmin java class. Replication to secondary region replicas works over standard inter-cluster replication.</p>
</div>
<div class="paragraph">
<div class="title">Default</div>
@@ -16437,7 +16437,7 @@ Instead you can change the number of region replicas per table to increase or de
<span class="tag"><name></span>hbase.region.replica.replication.enabled<span class="tag"></name></span>
<span class="tag"><value></span>true<span class="tag"></value></span>
<span class="tag"><description></span>
- Whether asynchronous WAL replication to the secondary region replicas is enabled or not. If this is enabled, a replication peer named "region_replica_replication" will be created which will tail the logs and replicate the mutations to region replicas for tables that have region replication <span class="error">></span> 1. If this is enabled once, disabling this replication also requires disabling the replication peer using shell or ReplicationAdmin java class. Replication to secondary region replicas works over standard inter-cluster replication. So replication, if disabled explicitly, also has to be enabled by setting "hbase.replication"� to true for this feature to work.
+ Whether asynchronous WAL replication to the secondary region replicas is enabled or not. If this is enabled, a replication peer named "region_replica_replication" will be created which will tail the logs and replicate the mutations to region replicas for tables that have region replication <span class="error">></span> 1. If this is enabled once, disabling this replication also requires disabling the replication peer using shell or ReplicationAdmin java class. Replication to secondary region replicas works over standard inter-cluster replication.
<span class="tag"></description></span>
<span class="tag"></property></span>
<span class="tag"><property></span>
@@ -25669,12 +25669,6 @@ The new layout will be:</p>
<td class="tableblock halign-left valign-top"><p class="tableblock">rs</p></td>
</tr>
<tr>
-<td class="tableblock halign-left valign-top"><p class="tableblock">hbase.replication</p></td>
-<td class="tableblock halign-left valign-top"><p class="tableblock">Whether replication is enabled or disabled on a given
- cluster</p></td>
-<td class="tableblock halign-left valign-top"><p class="tableblock">true</p></td>
-</tr>
-<tr>
<td class="tableblock halign-left valign-top"><p class="tableblock">replication.sleep.before.failover</p></td>
<td class="tableblock halign-left valign-top"><p class="tableblock">How many milliseconds a worker should sleep before attempting to replicate
a dead region server’s WAL queues.</p></td>
@@ -33754,7 +33748,7 @@ The server will return cellblocks compressed using this same compressor as long
<div id="footer">
<div id="footer-text">
Version 2.0.0-SNAPSHOT<br>
-Last updated 2016-04-08 14:30:12 UTC
+Last updated 2016-04-11 14:30:43 UTC
</div>
</div>
</body>
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/a0cdd2a9/bulk-loads.html
----------------------------------------------------------------------
diff --git a/bulk-loads.html b/bulk-loads.html
index 6a5ea54..e7f5511 100644
--- a/bulk-loads.html
+++ b/bulk-loads.html
@@ -7,7 +7,7 @@
<head>
<meta charset="UTF-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
- <meta name="Date-Revision-yyyymmdd" content="20160622" />
+ <meta name="Date-Revision-yyyymmdd" content="20160627" />
<meta http-equiv="Content-Language" content="en" />
<title>Apache HBase –
Bulk Loads in Apache HBase (TM)
@@ -305,7 +305,7 @@ under the License. -->
<a href="http://www.apache.org/">The Apache Software Foundation</a>.
All rights reserved.
- <li id="publishDate" class="pull-right">Last Published: 2016-06-22</li>
+ <li id="publishDate" class="pull-right">Last Published: 2016-06-27</li>
</p>
</div>