http://git-wip-us.apache.org/repos/asf/hbase-site/blob/22cff34f/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiHFileOutputFormat.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiHFileOutputFormat.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiHFileOutputFormat.html index 39ebcf3..9340f9d 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiHFileOutputFormat.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiHFileOutputFormat.html @@ -73,41 +73,40 @@ <span class="sourceLineNo">065</span> final FileSystem fs = outputDir.getFileSystem(conf);<a name="line.65"></a> <span class="sourceLineNo">066</span><a name="line.66"></a> <span class="sourceLineNo">067</span> // Map of tables to writers<a name="line.67"></a> -<span class="sourceLineNo">068</span> final Map<ImmutableBytesWritable, RecordWriter<ImmutableBytesWritable, V>> tableWriters =<a name="line.68"></a> -<span class="sourceLineNo">069</span> new HashMap<ImmutableBytesWritable, RecordWriter<ImmutableBytesWritable, V>>();<a name="line.69"></a> -<span class="sourceLineNo">070</span><a name="line.70"></a> -<span class="sourceLineNo">071</span> return new RecordWriter<ImmutableBytesWritable, V>() {<a name="line.71"></a> -<span class="sourceLineNo">072</span> @Override<a name="line.72"></a> -<span class="sourceLineNo">073</span> public void write(ImmutableBytesWritable tableName, V cell)<a name="line.73"></a> -<span class="sourceLineNo">074</span> throws IOException, InterruptedException {<a name="line.74"></a> -<span class="sourceLineNo">075</span> RecordWriter<ImmutableBytesWritable, V> tableWriter = tableWriters.get(tableName);<a name="line.75"></a> -<span class="sourceLineNo">076</span> // if there is new table, verify that table directory exists<a name="line.76"></a> -<span class="sourceLineNo">077</span> if (tableWriter == null) {<a name="line.77"></a> -<span class="sourceLineNo">078</span> // using table name as directory name<a name="line.78"></a> -<span class="sourceLineNo">079</span> final Path tableOutputDir = new Path(outputDir, Bytes.toString(tableName.copyBytes()));<a name="line.79"></a> -<span class="sourceLineNo">080</span> fs.mkdirs(tableOutputDir);<a name="line.80"></a> -<span class="sourceLineNo">081</span> LOG.info("Writing Table '" + tableName.toString() + "' data into following directory"<a name="line.81"></a> -<span class="sourceLineNo">082</span> + tableOutputDir.toString());<a name="line.82"></a> -<span class="sourceLineNo">083</span><a name="line.83"></a> -<span class="sourceLineNo">084</span> // Create writer for one specific table<a name="line.84"></a> -<span class="sourceLineNo">085</span> tableWriter = new HFileOutputFormat2.HFileRecordWriter<V>(context, tableOutputDir);<a name="line.85"></a> -<span class="sourceLineNo">086</span> // Put table into map<a name="line.86"></a> -<span class="sourceLineNo">087</span> tableWriters.put(tableName, tableWriter);<a name="line.87"></a> -<span class="sourceLineNo">088</span> }<a name="line.88"></a> -<span class="sourceLineNo">089</span> // Write <Row, Cell> into tableWriter<a name="line.89"></a> -<span class="sourceLineNo">090</span> // in the original code, it does not use Row<a name="line.90"></a> -<span class="sourceLineNo">091</span> tableWriter.write(null, cell);<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> @Override<a name="line.94"></a> -<span class="sourceLineNo">095</span> public void close(TaskAttemptContext c) throws IOException, InterruptedException {<a name="line.95"></a> -<span class="sourceLineNo">096</span> for (RecordWriter<ImmutableBytesWritable, V> writer : tableWriters.values()) {<a name="line.96"></a> -<span class="sourceLineNo">097</span> writer.close(c);<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> };<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">068</span> final Map<ImmutableBytesWritable, RecordWriter<ImmutableBytesWritable, V>> tableWriters = new HashMap<>();<a name="line.68"></a> +<span class="sourceLineNo">069</span><a name="line.69"></a> +<span class="sourceLineNo">070</span> return new RecordWriter<ImmutableBytesWritable, V>() {<a name="line.70"></a> +<span class="sourceLineNo">071</span> @Override<a name="line.71"></a> +<span class="sourceLineNo">072</span> public void write(ImmutableBytesWritable tableName, V cell)<a name="line.72"></a> +<span class="sourceLineNo">073</span> throws IOException, InterruptedException {<a name="line.73"></a> +<span class="sourceLineNo">074</span> RecordWriter<ImmutableBytesWritable, V> tableWriter = tableWriters.get(tableName);<a name="line.74"></a> +<span class="sourceLineNo">075</span> // if there is new table, verify that table directory exists<a name="line.75"></a> +<span class="sourceLineNo">076</span> if (tableWriter == null) {<a name="line.76"></a> +<span class="sourceLineNo">077</span> // using table name as directory name<a name="line.77"></a> +<span class="sourceLineNo">078</span> final Path tableOutputDir = new Path(outputDir, Bytes.toString(tableName.copyBytes()));<a name="line.78"></a> +<span class="sourceLineNo">079</span> fs.mkdirs(tableOutputDir);<a name="line.79"></a> +<span class="sourceLineNo">080</span> LOG.info("Writing Table '" + tableName.toString() + "' data into following directory"<a name="line.80"></a> +<span class="sourceLineNo">081</span> + tableOutputDir.toString());<a name="line.81"></a> +<span class="sourceLineNo">082</span><a name="line.82"></a> +<span class="sourceLineNo">083</span> // Create writer for one specific table<a name="line.83"></a> +<span class="sourceLineNo">084</span> tableWriter = new HFileOutputFormat2.HFileRecordWriter<>(context, tableOutputDir);<a name="line.84"></a> +<span class="sourceLineNo">085</span> // Put table into map<a name="line.85"></a> +<span class="sourceLineNo">086</span> tableWriters.put(tableName, tableWriter);<a name="line.86"></a> +<span class="sourceLineNo">087</span> }<a name="line.87"></a> +<span class="sourceLineNo">088</span> // Write <Row, Cell> into tableWriter<a name="line.88"></a> +<span class="sourceLineNo">089</span> // in the original code, it does not use Row<a name="line.89"></a> +<span class="sourceLineNo">090</span> tableWriter.write(null, cell);<a name="line.90"></a> +<span class="sourceLineNo">091</span> }<a name="line.91"></a> +<span class="sourceLineNo">092</span><a name="line.92"></a> +<span class="sourceLineNo">093</span> @Override<a name="line.93"></a> +<span class="sourceLineNo">094</span> public void close(TaskAttemptContext c) throws IOException, InterruptedException {<a name="line.94"></a> +<span class="sourceLineNo">095</span> for (RecordWriter<ImmutableBytesWritable, V> writer : tableWriters.values()) {<a name="line.95"></a> +<span class="sourceLineNo">096</span> writer.close(c);<a name="line.96"></a> +<span class="sourceLineNo">097</span> }<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> }<a name="line.100"></a> +<span class="sourceLineNo">101</span>}<a name="line.101"></a>
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/22cff34f/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiTableInputFormat.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiTableInputFormat.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiTableInputFormat.html index f4e33a0..bb2a823 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiTableInputFormat.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiTableInputFormat.html @@ -100,7 +100,7 @@ <span class="sourceLineNo">092</span> throw new IllegalArgumentException("There must be at least 1 scan configuration set to : "<a name="line.92"></a> <span class="sourceLineNo">093</span> + SCANS);<a name="line.93"></a> <span class="sourceLineNo">094</span> }<a name="line.94"></a> -<span class="sourceLineNo">095</span> List<Scan> scans = new ArrayList<Scan>();<a name="line.95"></a> +<span class="sourceLineNo">095</span> List<Scan> scans = new ArrayList<>();<a name="line.95"></a> <span class="sourceLineNo">096</span><a name="line.96"></a> <span class="sourceLineNo">097</span> for (int i = 0; i < rawScans.length; i++) {<a name="line.97"></a> <span class="sourceLineNo">098</span> try {<a name="line.98"></a> http://git-wip-us.apache.org/repos/asf/hbase-site/blob/22cff34f/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiTableInputFormatBase.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiTableInputFormatBase.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiTableInputFormatBase.html index f32d89d..97b325f 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiTableInputFormatBase.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiTableInputFormatBase.html @@ -171,7 +171,7 @@ <span class="sourceLineNo">163</span> throw new IOException("No scans were provided.");<a name="line.163"></a> <span class="sourceLineNo">164</span> }<a name="line.164"></a> <span class="sourceLineNo">165</span><a name="line.165"></a> -<span class="sourceLineNo">166</span> Map<TableName, List<Scan>> tableMaps = new HashMap<TableName, List<Scan>>();<a name="line.166"></a> +<span class="sourceLineNo">166</span> Map<TableName, List<Scan>> tableMaps = new HashMap<>();<a name="line.166"></a> <span class="sourceLineNo">167</span> for (Scan scan : scans) {<a name="line.167"></a> <span class="sourceLineNo">168</span> byte[] tableNameBytes = scan.getAttribute(Scan.SCAN_ATTRIBUTES_TABLE_NAME);<a name="line.168"></a> <span class="sourceLineNo">169</span> if (tableNameBytes == null)<a name="line.169"></a> @@ -181,13 +181,13 @@ <span class="sourceLineNo">173</span><a name="line.173"></a> <span class="sourceLineNo">174</span> List<Scan> scanList = tableMaps.get(tableName);<a name="line.174"></a> <span class="sourceLineNo">175</span> if (scanList == null) {<a name="line.175"></a> -<span class="sourceLineNo">176</span> scanList = new ArrayList<Scan>();<a name="line.176"></a> +<span class="sourceLineNo">176</span> scanList = new ArrayList<>();<a name="line.176"></a> <span class="sourceLineNo">177</span> tableMaps.put(tableName, scanList);<a name="line.177"></a> <span class="sourceLineNo">178</span> }<a name="line.178"></a> <span class="sourceLineNo">179</span> scanList.add(scan);<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> List<InputSplit> splits = new ArrayList<InputSplit>();<a name="line.182"></a> +<span class="sourceLineNo">182</span> List<InputSplit> splits = new ArrayList<>();<a name="line.182"></a> <span class="sourceLineNo">183</span> Iterator iter = tableMaps.entrySet().iterator();<a name="line.183"></a> <span class="sourceLineNo">184</span> while (iter.hasNext()) {<a name="line.184"></a> <span class="sourceLineNo">185</span> Map.Entry<TableName, List<Scan>> entry = (Map.Entry<TableName, List<Scan>>) iter.next();<a name="line.185"></a> http://git-wip-us.apache.org/repos/asf/hbase-site/blob/22cff34f/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/PutSortReducer.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/PutSortReducer.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/PutSortReducer.html index ac18540..c880dfd 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/PutSortReducer.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/PutSortReducer.html @@ -87,10 +87,10 @@ <span class="sourceLineNo">079</span> "putsortreducer.row.threshold", 1L * (1<<30));<a name="line.79"></a> <span class="sourceLineNo">080</span> Iterator<Put> iter = puts.iterator();<a name="line.80"></a> <span class="sourceLineNo">081</span> while (iter.hasNext()) {<a name="line.81"></a> -<span class="sourceLineNo">082</span> TreeSet<KeyValue> map = new TreeSet<KeyValue>(CellComparator.COMPARATOR);<a name="line.82"></a> +<span class="sourceLineNo">082</span> TreeSet<KeyValue> map = new TreeSet<>(CellComparator.COMPARATOR);<a name="line.82"></a> <span class="sourceLineNo">083</span> long curSize = 0;<a name="line.83"></a> <span class="sourceLineNo">084</span> // stop at the end or the RAM threshold<a name="line.84"></a> -<span class="sourceLineNo">085</span> List<Tag> tags = new ArrayList<Tag>();<a name="line.85"></a> +<span class="sourceLineNo">085</span> List<Tag> tags = new ArrayList<>();<a name="line.85"></a> <span class="sourceLineNo">086</span> while (iter.hasNext() && curSize < threshold) {<a name="line.86"></a> <span class="sourceLineNo">087</span> // clear the tags<a name="line.87"></a> <span class="sourceLineNo">088</span> tags.clear();<a name="line.88"></a> http://git-wip-us.apache.org/repos/asf/hbase-site/blob/22cff34f/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/ResultSerialization.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/ResultSerialization.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/ResultSerialization.html index 2a0b51a..dcdd181 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/ResultSerialization.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/ResultSerialization.html @@ -100,7 +100,7 @@ <span class="sourceLineNo">092</span> }<a name="line.92"></a> <span class="sourceLineNo">093</span> byte[] buf = new byte[totalBuffer];<a name="line.93"></a> <span class="sourceLineNo">094</span> readChunked(in, buf, 0, totalBuffer);<a name="line.94"></a> -<span class="sourceLineNo">095</span> List<Cell> kvs = new ArrayList<Cell>();<a name="line.95"></a> +<span class="sourceLineNo">095</span> List<Cell> kvs = new ArrayList<>();<a name="line.95"></a> <span class="sourceLineNo">096</span> int offset = 0;<a name="line.96"></a> <span class="sourceLineNo">097</span> while (offset < totalBuffer) {<a name="line.97"></a> <span class="sourceLineNo">098</span> int keyLength = Bytes.toInt(buf, offset);<a name="line.98"></a> http://git-wip-us.apache.org/repos/asf/hbase-site/blob/22cff34f/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/RowCounter.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/RowCounter.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/RowCounter.html index 87c73d9..97fbe11 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/RowCounter.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/RowCounter.html @@ -84,196 +84,195 @@ <span class="sourceLineNo">076</span> * @param values The columns.<a name="line.76"></a> <span class="sourceLineNo">077</span> * @param context The current context.<a name="line.77"></a> <span class="sourceLineNo">078</span> * @throws IOException When something is broken with the data.<a name="line.78"></a> -<span class="sourceLineNo">079</span> * @see org.apache.hadoop.mapreduce.Mapper#map(KEYIN, VALUEIN,<a name="line.79"></a> -<span class="sourceLineNo">080</span> * org.apache.hadoop.mapreduce.Mapper.Context)<a name="line.80"></a> -<span class="sourceLineNo">081</span> */<a name="line.81"></a> -<span class="sourceLineNo">082</span> @Override<a name="line.82"></a> -<span class="sourceLineNo">083</span> public void map(ImmutableBytesWritable row, Result values,<a name="line.83"></a> -<span class="sourceLineNo">084</span> Context context)<a name="line.84"></a> -<span class="sourceLineNo">085</span> throws IOException {<a name="line.85"></a> -<span class="sourceLineNo">086</span> // Count every row containing data, whether it's in qualifiers or values<a name="line.86"></a> -<span class="sourceLineNo">087</span> context.getCounter(Counters.ROWS).increment(1);<a name="line.87"></a> -<span class="sourceLineNo">088</span> }<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> /**<a name="line.91"></a> -<span class="sourceLineNo">092</span> * Sets up the actual job.<a name="line.92"></a> -<span class="sourceLineNo">093</span> *<a name="line.93"></a> -<span class="sourceLineNo">094</span> * @param conf The current configuration.<a name="line.94"></a> -<span class="sourceLineNo">095</span> * @param args The command line parameters.<a name="line.95"></a> -<span class="sourceLineNo">096</span> * @return The newly created job.<a name="line.96"></a> -<span class="sourceLineNo">097</span> * @throws IOException When setting up the job fails.<a name="line.97"></a> -<span class="sourceLineNo">098</span> */<a name="line.98"></a> -<span class="sourceLineNo">099</span> public static Job createSubmittableJob(Configuration conf, String[] args)<a name="line.99"></a> -<span class="sourceLineNo">100</span> throws IOException {<a name="line.100"></a> -<span class="sourceLineNo">101</span> String tableName = args[0];<a name="line.101"></a> -<span class="sourceLineNo">102</span> List<MultiRowRangeFilter.RowRange> rowRangeList = null;<a name="line.102"></a> -<span class="sourceLineNo">103</span> long startTime = 0;<a name="line.103"></a> -<span class="sourceLineNo">104</span> long endTime = 0;<a name="line.104"></a> -<span class="sourceLineNo">105</span><a name="line.105"></a> -<span class="sourceLineNo">106</span> StringBuilder sb = new StringBuilder();<a name="line.106"></a> -<span class="sourceLineNo">107</span><a name="line.107"></a> -<span class="sourceLineNo">108</span> final String rangeSwitch = "--range=";<a name="line.108"></a> -<span class="sourceLineNo">109</span> final String startTimeArgKey = "--starttime=";<a name="line.109"></a> -<span class="sourceLineNo">110</span> final String endTimeArgKey = "--endtime=";<a name="line.110"></a> -<span class="sourceLineNo">111</span> final String expectedCountArg = "--expected-count=";<a name="line.111"></a> -<span class="sourceLineNo">112</span><a name="line.112"></a> -<span class="sourceLineNo">113</span> // First argument is table name, starting from second<a name="line.113"></a> -<span class="sourceLineNo">114</span> for (int i = 1; i < args.length; i++) {<a name="line.114"></a> -<span class="sourceLineNo">115</span> if (args[i].startsWith(rangeSwitch)) {<a name="line.115"></a> -<span class="sourceLineNo">116</span> try {<a name="line.116"></a> -<span class="sourceLineNo">117</span> rowRangeList = parseRowRangeParameter(args[i], rangeSwitch);<a name="line.117"></a> -<span class="sourceLineNo">118</span> } catch (IllegalArgumentException e) {<a name="line.118"></a> -<span class="sourceLineNo">119</span> return null;<a name="line.119"></a> -<span class="sourceLineNo">120</span> }<a name="line.120"></a> -<span class="sourceLineNo">121</span> continue;<a name="line.121"></a> -<span class="sourceLineNo">122</span> }<a name="line.122"></a> -<span class="sourceLineNo">123</span> if (args[i].startsWith(startTimeArgKey)) {<a name="line.123"></a> -<span class="sourceLineNo">124</span> startTime = Long.parseLong(args[i].substring(startTimeArgKey.length()));<a name="line.124"></a> -<span class="sourceLineNo">125</span> continue;<a name="line.125"></a> -<span class="sourceLineNo">126</span> }<a name="line.126"></a> -<span class="sourceLineNo">127</span> if (args[i].startsWith(endTimeArgKey)) {<a name="line.127"></a> -<span class="sourceLineNo">128</span> endTime = Long.parseLong(args[i].substring(endTimeArgKey.length()));<a name="line.128"></a> -<span class="sourceLineNo">129</span> continue;<a name="line.129"></a> -<span class="sourceLineNo">130</span> }<a name="line.130"></a> -<span class="sourceLineNo">131</span> if (args[i].startsWith(expectedCountArg)) {<a name="line.131"></a> -<span class="sourceLineNo">132</span> conf.setLong(EXPECTED_COUNT_KEY,<a name="line.132"></a> -<span class="sourceLineNo">133</span> Long.parseLong(args[i].substring(expectedCountArg.length())));<a name="line.133"></a> -<span class="sourceLineNo">134</span> continue;<a name="line.134"></a> -<span class="sourceLineNo">135</span> }<a name="line.135"></a> -<span class="sourceLineNo">136</span> // if no switch, assume column names<a name="line.136"></a> -<span class="sourceLineNo">137</span> sb.append(args[i]);<a name="line.137"></a> -<span class="sourceLineNo">138</span> sb.append(" ");<a name="line.138"></a> -<span class="sourceLineNo">139</span> }<a name="line.139"></a> -<span class="sourceLineNo">140</span> if (endTime < startTime) {<a name="line.140"></a> -<span class="sourceLineNo">141</span> printUsage("--endtime=" + endTime + " needs to be greater than --starttime=" + startTime);<a name="line.141"></a> -<span class="sourceLineNo">142</span> return null;<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> Job job = Job.getInstance(conf, conf.get(JOB_NAME_CONF_KEY, NAME + "_" + tableName));<a name="line.145"></a> -<span class="sourceLineNo">146</span> job.setJarByClass(RowCounter.class);<a name="line.146"></a> -<span class="sourceLineNo">147</span> Scan scan = new Scan();<a name="line.147"></a> -<span class="sourceLineNo">148</span> scan.setCacheBlocks(false);<a name="line.148"></a> -<span class="sourceLineNo">149</span> setScanFilter(scan, rowRangeList);<a name="line.149"></a> -<span class="sourceLineNo">150</span> if (sb.length() > 0) {<a name="line.150"></a> -<span class="sourceLineNo">151</span> for (String columnName : sb.toString().trim().split(" ")) {<a name="line.151"></a> -<span class="sourceLineNo">152</span> String family = StringUtils.substringBefore(columnName, ":");<a name="line.152"></a> -<span class="sourceLineNo">153</span> String qualifier = StringUtils.substringAfter(columnName, ":");<a name="line.153"></a> -<span class="sourceLineNo">154</span><a name="line.154"></a> -<span class="sourceLineNo">155</span> if (StringUtils.isBlank(qualifier)) {<a name="line.155"></a> -<span class="sourceLineNo">156</span> scan.addFamily(Bytes.toBytes(family));<a name="line.156"></a> -<span class="sourceLineNo">157</span> }<a name="line.157"></a> -<span class="sourceLineNo">158</span> else {<a name="line.158"></a> -<span class="sourceLineNo">159</span> scan.addColumn(Bytes.toBytes(family), Bytes.toBytes(qualifier));<a name="line.159"></a> -<span class="sourceLineNo">160</span> }<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> scan.setTimeRange(startTime, endTime == 0 ? HConstants.LATEST_TIMESTAMP : endTime);<a name="line.163"></a> -<span class="sourceLineNo">164</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.164"></a> -<span class="sourceLineNo">165</span> TableMapReduceUtil.initTableMapperJob(tableName, scan,<a name="line.165"></a> -<span class="sourceLineNo">166</span> RowCounterMapper.class, ImmutableBytesWritable.class, Result.class, job);<a name="line.166"></a> -<span class="sourceLineNo">167</span> job.setNumReduceTasks(0);<a name="line.167"></a> -<span class="sourceLineNo">168</span> return job;<a name="line.168"></a> -<span class="sourceLineNo">169</span> }<a name="line.169"></a> -<span class="sourceLineNo">170</span><a name="line.170"></a> -<span class="sourceLineNo">171</span> private static List<MultiRowRangeFilter.RowRange> parseRowRangeParameter(<a name="line.171"></a> -<span class="sourceLineNo">172</span> String arg, String rangeSwitch) {<a name="line.172"></a> -<span class="sourceLineNo">173</span> final String[] ranges = arg.substring(rangeSwitch.length()).split(";");<a name="line.173"></a> -<span class="sourceLineNo">174</span> final List<MultiRowRangeFilter.RowRange> rangeList = new ArrayList<>();<a name="line.174"></a> -<span class="sourceLineNo">175</span> for (String range : ranges) {<a name="line.175"></a> -<span class="sourceLineNo">176</span> String[] startEnd = range.split(",", 2);<a name="line.176"></a> -<span class="sourceLineNo">177</span> if (startEnd.length != 2 || startEnd[1].contains(",")) {<a name="line.177"></a> -<span class="sourceLineNo">178</span> printUsage("Please specify range in such format as \"--range=a,b\" " +<a name="line.178"></a> -<span class="sourceLineNo">179</span> "or, with only one boundary, \"--range=,b\" or \"--range=a,\"");<a name="line.179"></a> -<span class="sourceLineNo">180</span> throw new IllegalArgumentException("Wrong range specification: " + range);<a name="line.180"></a> -<span class="sourceLineNo">181</span> }<a name="line.181"></a> -<span class="sourceLineNo">182</span> String startKey = startEnd[0];<a name="line.182"></a> -<span class="sourceLineNo">183</span> String endKey = startEnd[1];<a name="line.183"></a> -<span class="sourceLineNo">184</span> rangeList.add(new MultiRowRangeFilter.RowRange(<a name="line.184"></a> -<span class="sourceLineNo">185</span> Bytes.toBytesBinary(startKey), true,<a name="line.185"></a> -<span class="sourceLineNo">186</span> Bytes.toBytesBinary(endKey), false));<a name="line.186"></a> -<span class="sourceLineNo">187</span> }<a name="line.187"></a> -<span class="sourceLineNo">188</span> return rangeList;<a name="line.188"></a> -<span class="sourceLineNo">189</span> }<a name="line.189"></a> -<span class="sourceLineNo">190</span><a name="line.190"></a> -<span class="sourceLineNo">191</span> /**<a name="line.191"></a> -<span class="sourceLineNo">192</span> * Sets filter {@link FilterBase} to the {@link Scan} instance.<a name="line.192"></a> -<span class="sourceLineNo">193</span> * If provided rowRangeList contains more than one element,<a name="line.193"></a> -<span class="sourceLineNo">194</span> * method sets filter which is instance of {@link MultiRowRangeFilter}.<a name="line.194"></a> -<span class="sourceLineNo">195</span> * Otherwise, method sets filter which is instance of {@link FirstKeyOnlyFilter}.<a name="line.195"></a> -<span class="sourceLineNo">196</span> * If rowRangeList contains exactly one element, startRow and stopRow are set to the scan.<a name="line.196"></a> -<span class="sourceLineNo">197</span> * @param scan<a name="line.197"></a> -<span class="sourceLineNo">198</span> * @param rowRangeList<a name="line.198"></a> -<span class="sourceLineNo">199</span> */<a name="line.199"></a> -<span class="sourceLineNo">200</span> private static void setScanFilter(Scan scan, List<MultiRowRangeFilter.RowRange> rowRangeList) {<a name="line.200"></a> -<span class="sourceLineNo">201</span> final int size = rowRangeList == null ? 0 : rowRangeList.size();<a name="line.201"></a> -<span class="sourceLineNo">202</span> if (size <= 1) {<a name="line.202"></a> -<span class="sourceLineNo">203</span> scan.setFilter(new FirstKeyOnlyFilter());<a name="line.203"></a> -<span class="sourceLineNo">204</span> }<a name="line.204"></a> -<span class="sourceLineNo">205</span> if (size == 1) {<a name="line.205"></a> -<span class="sourceLineNo">206</span> MultiRowRangeFilter.RowRange range = rowRangeList.get(0);<a name="line.206"></a> -<span class="sourceLineNo">207</span> scan.setStartRow(range.getStartRow()); //inclusive<a name="line.207"></a> -<span class="sourceLineNo">208</span> scan.setStopRow(range.getStopRow()); //exclusive<a name="line.208"></a> -<span class="sourceLineNo">209</span> } else if (size > 1) {<a name="line.209"></a> -<span class="sourceLineNo">210</span> scan.setFilter(new MultiRowRangeFilter(rowRangeList));<a name="line.210"></a> -<span class="sourceLineNo">211</span> }<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> * @param errorMessage Can attach a message when error occurs.<a name="line.215"></a> -<span class="sourceLineNo">216</span> */<a name="line.216"></a> -<span class="sourceLineNo">217</span> private static void printUsage(String errorMessage) {<a name="line.217"></a> -<span class="sourceLineNo">218</span> System.err.println("ERROR: " + errorMessage);<a name="line.218"></a> -<span class="sourceLineNo">219</span> printUsage();<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">222</span> /**<a name="line.222"></a> -<span class="sourceLineNo">223</span> * Prints usage without error message.<a name="line.223"></a> -<span class="sourceLineNo">224</span> * Note that we don't document --expected-count, because it's intended for test.<a name="line.224"></a> -<span class="sourceLineNo">225</span> */<a name="line.225"></a> -<span class="sourceLineNo">226</span> private static void printUsage() {<a name="line.226"></a> -<span class="sourceLineNo">227</span> System.err.println("Usage: RowCounter [options] <tablename> " +<a name="line.227"></a> -<span class="sourceLineNo">228</span> "[--starttime=[start] --endtime=[end] " +<a name="line.228"></a> -<span class="sourceLineNo">229</span> "[--range=[startKey],[endKey][;[startKey],[endKey]...]] [<column1> <column2>...]");<a name="line.229"></a> -<span class="sourceLineNo">230</span> System.err.println("For performance consider the following options:\n"<a name="line.230"></a> -<span class="sourceLineNo">231</span> + "-Dhbase.client.scanner.caching=100\n"<a name="line.231"></a> -<span class="sourceLineNo">232</span> + "-Dmapreduce.map.speculative=false");<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> @Override<a name="line.235"></a> -<span class="sourceLineNo">236</span> public int run(String[] args) throws Exception {<a name="line.236"></a> -<span class="sourceLineNo">237</span> if (args.length < 1) {<a name="line.237"></a> -<span class="sourceLineNo">238</span> printUsage("Wrong number of parameters: " + args.length);<a name="line.238"></a> -<span class="sourceLineNo">239</span> return -1;<a name="line.239"></a> -<span class="sourceLineNo">240</span> }<a name="line.240"></a> -<span class="sourceLineNo">241</span> Job job = createSubmittableJob(getConf(), args);<a name="line.241"></a> -<span class="sourceLineNo">242</span> if (job == null) {<a name="line.242"></a> -<span class="sourceLineNo">243</span> return -1;<a name="line.243"></a> -<span class="sourceLineNo">244</span> }<a name="line.244"></a> -<span class="sourceLineNo">245</span> boolean success = job.waitForCompletion(true);<a name="line.245"></a> -<span class="sourceLineNo">246</span> final long expectedCount = getConf().getLong(EXPECTED_COUNT_KEY, -1);<a name="line.246"></a> -<span class="sourceLineNo">247</span> if (success && expectedCount != -1) {<a name="line.247"></a> -<span class="sourceLineNo">248</span> final Counter counter = job.getCounters().findCounter(RowCounterMapper.Counters.ROWS);<a name="line.248"></a> -<span class="sourceLineNo">249</span> success = expectedCount == counter.getValue();<a name="line.249"></a> -<span class="sourceLineNo">250</span> if (!success) {<a name="line.250"></a> -<span class="sourceLineNo">251</span> LOG.error("Failing job because count of '" + counter.getValue() +<a name="line.251"></a> -<span class="sourceLineNo">252</span> "' does not match expected count of '" + expectedCount + "'");<a name="line.252"></a> -<span class="sourceLineNo">253</span> }<a name="line.253"></a> -<span class="sourceLineNo">254</span> }<a name="line.254"></a> -<span class="sourceLineNo">255</span> return (success ? 0 : 1);<a name="line.255"></a> -<span class="sourceLineNo">256</span> }<a name="line.256"></a> -<span class="sourceLineNo">257</span><a name="line.257"></a> -<span class="sourceLineNo">258</span> /**<a name="line.258"></a> -<span class="sourceLineNo">259</span> * Main entry point.<a name="line.259"></a> -<span class="sourceLineNo">260</span> * @param args The command line parameters.<a name="line.260"></a> -<span class="sourceLineNo">261</span> * @throws Exception When running the job fails.<a name="line.261"></a> -<span class="sourceLineNo">262</span> */<a name="line.262"></a> -<span class="sourceLineNo">263</span> public static void main(String[] args) throws Exception {<a name="line.263"></a> -<span class="sourceLineNo">264</span> int errCode = ToolRunner.run(HBaseConfiguration.create(), new RowCounter(), args);<a name="line.264"></a> -<span class="sourceLineNo">265</span> System.exit(errCode);<a name="line.265"></a> -<span class="sourceLineNo">266</span> }<a name="line.266"></a> -<span class="sourceLineNo">267</span><a name="line.267"></a> -<span class="sourceLineNo">268</span>}<a name="line.268"></a> +<span class="sourceLineNo">079</span> * @see org.apache.hadoop.mapreduce.Mapper#map(Object, Object, Context)<a name="line.79"></a> +<span class="sourceLineNo">080</span> */<a name="line.80"></a> +<span class="sourceLineNo">081</span> @Override<a name="line.81"></a> +<span class="sourceLineNo">082</span> public void map(ImmutableBytesWritable row, Result values,<a name="line.82"></a> +<span class="sourceLineNo">083</span> Context context)<a name="line.83"></a> +<span class="sourceLineNo">084</span> throws IOException {<a name="line.84"></a> +<span class="sourceLineNo">085</span> // Count every row containing data, whether it's in qualifiers or values<a name="line.85"></a> +<span class="sourceLineNo">086</span> context.getCounter(Counters.ROWS).increment(1);<a name="line.86"></a> +<span class="sourceLineNo">087</span> }<a name="line.87"></a> +<span class="sourceLineNo">088</span> }<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> * Sets up the actual job.<a name="line.91"></a> +<span class="sourceLineNo">092</span> *<a name="line.92"></a> +<span class="sourceLineNo">093</span> * @param conf The current configuration.<a name="line.93"></a> +<span class="sourceLineNo">094</span> * @param args The command line parameters.<a name="line.94"></a> +<span class="sourceLineNo">095</span> * @return The newly created job.<a name="line.95"></a> +<span class="sourceLineNo">096</span> * @throws IOException When setting up the job fails.<a name="line.96"></a> +<span class="sourceLineNo">097</span> */<a name="line.97"></a> +<span class="sourceLineNo">098</span> public static Job createSubmittableJob(Configuration conf, String[] args)<a name="line.98"></a> +<span class="sourceLineNo">099</span> throws IOException {<a name="line.99"></a> +<span class="sourceLineNo">100</span> String tableName = args[0];<a name="line.100"></a> +<span class="sourceLineNo">101</span> List<MultiRowRangeFilter.RowRange> rowRangeList = null;<a name="line.101"></a> +<span class="sourceLineNo">102</span> long startTime = 0;<a name="line.102"></a> +<span class="sourceLineNo">103</span> long endTime = 0;<a name="line.103"></a> +<span class="sourceLineNo">104</span><a name="line.104"></a> +<span class="sourceLineNo">105</span> StringBuilder sb = new StringBuilder();<a name="line.105"></a> +<span class="sourceLineNo">106</span><a name="line.106"></a> +<span class="sourceLineNo">107</span> final String rangeSwitch = "--range=";<a name="line.107"></a> +<span class="sourceLineNo">108</span> final String startTimeArgKey = "--starttime=";<a name="line.108"></a> +<span class="sourceLineNo">109</span> final String endTimeArgKey = "--endtime=";<a name="line.109"></a> +<span class="sourceLineNo">110</span> final String expectedCountArg = "--expected-count=";<a name="line.110"></a> +<span class="sourceLineNo">111</span><a name="line.111"></a> +<span class="sourceLineNo">112</span> // First argument is table name, starting from second<a name="line.112"></a> +<span class="sourceLineNo">113</span> for (int i = 1; i < args.length; i++) {<a name="line.113"></a> +<span class="sourceLineNo">114</span> if (args[i].startsWith(rangeSwitch)) {<a name="line.114"></a> +<span class="sourceLineNo">115</span> try {<a name="line.115"></a> +<span class="sourceLineNo">116</span> rowRangeList = parseRowRangeParameter(args[i], rangeSwitch);<a name="line.116"></a> +<span class="sourceLineNo">117</span> } catch (IllegalArgumentException e) {<a name="line.117"></a> +<span class="sourceLineNo">118</span> return null;<a name="line.118"></a> +<span class="sourceLineNo">119</span> }<a name="line.119"></a> +<span class="sourceLineNo">120</span> continue;<a name="line.120"></a> +<span class="sourceLineNo">121</span> }<a name="line.121"></a> +<span class="sourceLineNo">122</span> if (args[i].startsWith(startTimeArgKey)) {<a name="line.122"></a> +<span class="sourceLineNo">123</span> startTime = Long.parseLong(args[i].substring(startTimeArgKey.length()));<a name="line.123"></a> +<span class="sourceLineNo">124</span> continue;<a name="line.124"></a> +<span class="sourceLineNo">125</span> }<a name="line.125"></a> +<span class="sourceLineNo">126</span> if (args[i].startsWith(endTimeArgKey)) {<a name="line.126"></a> +<span class="sourceLineNo">127</span> endTime = Long.parseLong(args[i].substring(endTimeArgKey.length()));<a name="line.127"></a> +<span class="sourceLineNo">128</span> continue;<a name="line.128"></a> +<span class="sourceLineNo">129</span> }<a name="line.129"></a> +<span class="sourceLineNo">130</span> if (args[i].startsWith(expectedCountArg)) {<a name="line.130"></a> +<span class="sourceLineNo">131</span> conf.setLong(EXPECTED_COUNT_KEY,<a name="line.131"></a> +<span class="sourceLineNo">132</span> Long.parseLong(args[i].substring(expectedCountArg.length())));<a name="line.132"></a> +<span class="sourceLineNo">133</span> continue;<a name="line.133"></a> +<span class="sourceLineNo">134</span> }<a name="line.134"></a> +<span class="sourceLineNo">135</span> // if no switch, assume column names<a name="line.135"></a> +<span class="sourceLineNo">136</span> sb.append(args[i]);<a name="line.136"></a> +<span class="sourceLineNo">137</span> sb.append(" ");<a name="line.137"></a> +<span class="sourceLineNo">138</span> }<a name="line.138"></a> +<span class="sourceLineNo">139</span> if (endTime < startTime) {<a name="line.139"></a> +<span class="sourceLineNo">140</span> printUsage("--endtime=" + endTime + " needs to be greater than --starttime=" + startTime);<a name="line.140"></a> +<span class="sourceLineNo">141</span> return null;<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> Job job = Job.getInstance(conf, conf.get(JOB_NAME_CONF_KEY, NAME + "_" + tableName));<a name="line.144"></a> +<span class="sourceLineNo">145</span> job.setJarByClass(RowCounter.class);<a name="line.145"></a> +<span class="sourceLineNo">146</span> Scan scan = new Scan();<a name="line.146"></a> +<span class="sourceLineNo">147</span> scan.setCacheBlocks(false);<a name="line.147"></a> +<span class="sourceLineNo">148</span> setScanFilter(scan, rowRangeList);<a name="line.148"></a> +<span class="sourceLineNo">149</span> if (sb.length() > 0) {<a name="line.149"></a> +<span class="sourceLineNo">150</span> for (String columnName : sb.toString().trim().split(" ")) {<a name="line.150"></a> +<span class="sourceLineNo">151</span> String family = StringUtils.substringBefore(columnName, ":");<a name="line.151"></a> +<span class="sourceLineNo">152</span> String qualifier = StringUtils.substringAfter(columnName, ":");<a name="line.152"></a> +<span class="sourceLineNo">153</span><a name="line.153"></a> +<span class="sourceLineNo">154</span> if (StringUtils.isBlank(qualifier)) {<a name="line.154"></a> +<span class="sourceLineNo">155</span> scan.addFamily(Bytes.toBytes(family));<a name="line.155"></a> +<span class="sourceLineNo">156</span> }<a name="line.156"></a> +<span class="sourceLineNo">157</span> else {<a name="line.157"></a> +<span class="sourceLineNo">158</span> scan.addColumn(Bytes.toBytes(family), Bytes.toBytes(qualifier));<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> }<a name="line.161"></a> +<span class="sourceLineNo">162</span> scan.setTimeRange(startTime, endTime == 0 ? HConstants.LATEST_TIMESTAMP : endTime);<a name="line.162"></a> +<span class="sourceLineNo">163</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.163"></a> +<span class="sourceLineNo">164</span> TableMapReduceUtil.initTableMapperJob(tableName, scan,<a name="line.164"></a> +<span class="sourceLineNo">165</span> RowCounterMapper.class, ImmutableBytesWritable.class, Result.class, job);<a name="line.165"></a> +<span class="sourceLineNo">166</span> job.setNumReduceTasks(0);<a name="line.166"></a> +<span class="sourceLineNo">167</span> return job;<a name="line.167"></a> +<span class="sourceLineNo">168</span> }<a name="line.168"></a> +<span class="sourceLineNo">169</span><a name="line.169"></a> +<span class="sourceLineNo">170</span> private static List<MultiRowRangeFilter.RowRange> parseRowRangeParameter(<a name="line.170"></a> +<span class="sourceLineNo">171</span> String arg, String rangeSwitch) {<a name="line.171"></a> +<span class="sourceLineNo">172</span> final String[] ranges = arg.substring(rangeSwitch.length()).split(";");<a name="line.172"></a> +<span class="sourceLineNo">173</span> final List<MultiRowRangeFilter.RowRange> rangeList = new ArrayList<>();<a name="line.173"></a> +<span class="sourceLineNo">174</span> for (String range : ranges) {<a name="line.174"></a> +<span class="sourceLineNo">175</span> String[] startEnd = range.split(",", 2);<a name="line.175"></a> +<span class="sourceLineNo">176</span> if (startEnd.length != 2 || startEnd[1].contains(",")) {<a name="line.176"></a> +<span class="sourceLineNo">177</span> printUsage("Please specify range in such format as \"--range=a,b\" " +<a name="line.177"></a> +<span class="sourceLineNo">178</span> "or, with only one boundary, \"--range=,b\" or \"--range=a,\"");<a name="line.178"></a> +<span class="sourceLineNo">179</span> throw new IllegalArgumentException("Wrong range specification: " + range);<a name="line.179"></a> +<span class="sourceLineNo">180</span> }<a name="line.180"></a> +<span class="sourceLineNo">181</span> String startKey = startEnd[0];<a name="line.181"></a> +<span class="sourceLineNo">182</span> String endKey = startEnd[1];<a name="line.182"></a> +<span class="sourceLineNo">183</span> rangeList.add(new MultiRowRangeFilter.RowRange(<a name="line.183"></a> +<span class="sourceLineNo">184</span> Bytes.toBytesBinary(startKey), true,<a name="line.184"></a> +<span class="sourceLineNo">185</span> Bytes.toBytesBinary(endKey), false));<a name="line.185"></a> +<span class="sourceLineNo">186</span> }<a name="line.186"></a> +<span class="sourceLineNo">187</span> return rangeList;<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> /**<a name="line.190"></a> +<span class="sourceLineNo">191</span> * Sets filter {@link FilterBase} to the {@link Scan} instance.<a name="line.191"></a> +<span class="sourceLineNo">192</span> * If provided rowRangeList contains more than one element,<a name="line.192"></a> +<span class="sourceLineNo">193</span> * method sets filter which is instance of {@link MultiRowRangeFilter}.<a name="line.193"></a> +<span class="sourceLineNo">194</span> * Otherwise, method sets filter which is instance of {@link FirstKeyOnlyFilter}.<a name="line.194"></a> +<span class="sourceLineNo">195</span> * If rowRangeList contains exactly one element, startRow and stopRow are set to the scan.<a name="line.195"></a> +<span class="sourceLineNo">196</span> * @param scan<a name="line.196"></a> +<span class="sourceLineNo">197</span> * @param rowRangeList<a name="line.197"></a> +<span class="sourceLineNo">198</span> */<a name="line.198"></a> +<span class="sourceLineNo">199</span> private static void setScanFilter(Scan scan, List<MultiRowRangeFilter.RowRange> rowRangeList) {<a name="line.199"></a> +<span class="sourceLineNo">200</span> final int size = rowRangeList == null ? 0 : rowRangeList.size();<a name="line.200"></a> +<span class="sourceLineNo">201</span> if (size <= 1) {<a name="line.201"></a> +<span class="sourceLineNo">202</span> scan.setFilter(new FirstKeyOnlyFilter());<a name="line.202"></a> +<span class="sourceLineNo">203</span> }<a name="line.203"></a> +<span class="sourceLineNo">204</span> if (size == 1) {<a name="line.204"></a> +<span class="sourceLineNo">205</span> MultiRowRangeFilter.RowRange range = rowRangeList.get(0);<a name="line.205"></a> +<span class="sourceLineNo">206</span> scan.setStartRow(range.getStartRow()); //inclusive<a name="line.206"></a> +<span class="sourceLineNo">207</span> scan.setStopRow(range.getStopRow()); //exclusive<a name="line.207"></a> +<span class="sourceLineNo">208</span> } else if (size > 1) {<a name="line.208"></a> +<span class="sourceLineNo">209</span> scan.setFilter(new MultiRowRangeFilter(rowRangeList));<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><a name="line.212"></a> +<span class="sourceLineNo">213</span> /*<a name="line.213"></a> +<span class="sourceLineNo">214</span> * @param errorMessage Can attach a message when error occurs.<a name="line.214"></a> +<span class="sourceLineNo">215</span> */<a name="line.215"></a> +<span class="sourceLineNo">216</span> private static void printUsage(String errorMessage) {<a name="line.216"></a> +<span class="sourceLineNo">217</span> System.err.println("ERROR: " + errorMessage);<a name="line.217"></a> +<span class="sourceLineNo">218</span> printUsage();<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">222</span> * Prints usage without error message.<a name="line.222"></a> +<span class="sourceLineNo">223</span> * Note that we don't document --expected-count, because it's intended for test.<a name="line.223"></a> +<span class="sourceLineNo">224</span> */<a name="line.224"></a> +<span class="sourceLineNo">225</span> private static void printUsage() {<a name="line.225"></a> +<span class="sourceLineNo">226</span> System.err.println("Usage: RowCounter [options] <tablename> " +<a name="line.226"></a> +<span class="sourceLineNo">227</span> "[--starttime=[start] --endtime=[end] " +<a name="line.227"></a> +<span class="sourceLineNo">228</span> "[--range=[startKey],[endKey][;[startKey],[endKey]...]] [<column1> <column2>...]");<a name="line.228"></a> +<span class="sourceLineNo">229</span> System.err.println("For performance consider the following options:\n"<a name="line.229"></a> +<span class="sourceLineNo">230</span> + "-Dhbase.client.scanner.caching=100\n"<a name="line.230"></a> +<span class="sourceLineNo">231</span> + "-Dmapreduce.map.speculative=false");<a name="line.231"></a> +<span class="sourceLineNo">232</span> }<a name="line.232"></a> +<span class="sourceLineNo">233</span><a name="line.233"></a> +<span class="sourceLineNo">234</span> @Override<a name="line.234"></a> +<span class="sourceLineNo">235</span> public int run(String[] args) throws Exception {<a name="line.235"></a> +<span class="sourceLineNo">236</span> if (args.length < 1) {<a name="line.236"></a> +<span class="sourceLineNo">237</span> printUsage("Wrong number of parameters: " + args.length);<a name="line.237"></a> +<span class="sourceLineNo">238</span> return -1;<a name="line.238"></a> +<span class="sourceLineNo">239</span> }<a name="line.239"></a> +<span class="sourceLineNo">240</span> Job job = createSubmittableJob(getConf(), args);<a name="line.240"></a> +<span class="sourceLineNo">241</span> if (job == null) {<a name="line.241"></a> +<span class="sourceLineNo">242</span> return -1;<a name="line.242"></a> +<span class="sourceLineNo">243</span> }<a name="line.243"></a> +<span class="sourceLineNo">244</span> boolean success = job.waitForCompletion(true);<a name="line.244"></a> +<span class="sourceLineNo">245</span> final long expectedCount = getConf().getLong(EXPECTED_COUNT_KEY, -1);<a name="line.245"></a> +<span class="sourceLineNo">246</span> if (success && expectedCount != -1) {<a name="line.246"></a> +<span class="sourceLineNo">247</span> final Counter counter = job.getCounters().findCounter(RowCounterMapper.Counters.ROWS);<a name="line.247"></a> +<span class="sourceLineNo">248</span> success = expectedCount == counter.getValue();<a name="line.248"></a> +<span class="sourceLineNo">249</span> if (!success) {<a name="line.249"></a> +<span class="sourceLineNo">250</span> LOG.error("Failing job because count of '" + counter.getValue() +<a name="line.250"></a> +<span class="sourceLineNo">251</span> "' does not match expected count of '" + expectedCount + "'");<a name="line.251"></a> +<span class="sourceLineNo">252</span> }<a name="line.252"></a> +<span class="sourceLineNo">253</span> }<a name="line.253"></a> +<span class="sourceLineNo">254</span> return (success ? 0 : 1);<a name="line.254"></a> +<span class="sourceLineNo">255</span> }<a name="line.255"></a> +<span class="sourceLineNo">256</span><a name="line.256"></a> +<span class="sourceLineNo">257</span> /**<a name="line.257"></a> +<span class="sourceLineNo">258</span> * Main entry point.<a name="line.258"></a> +<span class="sourceLineNo">259</span> * @param args The command line parameters.<a name="line.259"></a> +<span class="sourceLineNo">260</span> * @throws Exception When running the job fails.<a name="line.260"></a> +<span class="sourceLineNo">261</span> */<a name="line.261"></a> +<span class="sourceLineNo">262</span> public static void main(String[] args) throws Exception {<a name="line.262"></a> +<span class="sourceLineNo">263</span> int errCode = ToolRunner.run(HBaseConfiguration.create(), new RowCounter(), args);<a name="line.263"></a> +<span class="sourceLineNo">264</span> System.exit(errCode);<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>}<a name="line.267"></a>