| <!DOCTYPE HTML> |
| <html lang="en"> |
| <head> |
| <!-- Generated by javadoc (17) --> |
| <title>Source code</title> |
| <meta name="viewport" content="width=device-width, initial-scale=1"> |
| <meta name="description" content="source: package: org.apache.hadoop.hbase.mapreduce, class: Import"> |
| <meta name="generator" content="javadoc/SourceToHTMLConverter"> |
| <link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style"> |
| </head> |
| <body class="source-page"> |
| <main role="main"> |
| <div class="source-container"> |
| <pre><span class="source-line-no">001</span><span id="line-1">/*</span> |
| <span class="source-line-no">002</span><span id="line-2"> * Licensed to the Apache Software Foundation (ASF) under one</span> |
| <span class="source-line-no">003</span><span id="line-3"> * or more contributor license agreements. See the NOTICE file</span> |
| <span class="source-line-no">004</span><span id="line-4"> * distributed with this work for additional information</span> |
| <span class="source-line-no">005</span><span id="line-5"> * regarding copyright ownership. The ASF licenses this file</span> |
| <span class="source-line-no">006</span><span id="line-6"> * to you under the Apache License, Version 2.0 (the</span> |
| <span class="source-line-no">007</span><span id="line-7"> * "License"); you may not use this file except in compliance</span> |
| <span class="source-line-no">008</span><span id="line-8"> * with the License. You may obtain a copy of the License at</span> |
| <span class="source-line-no">009</span><span id="line-9"> *</span> |
| <span class="source-line-no">010</span><span id="line-10"> * http://www.apache.org/licenses/LICENSE-2.0</span> |
| <span class="source-line-no">011</span><span id="line-11"> *</span> |
| <span class="source-line-no">012</span><span id="line-12"> * Unless required by applicable law or agreed to in writing, software</span> |
| <span class="source-line-no">013</span><span id="line-13"> * distributed under the License is distributed on an "AS IS" BASIS,</span> |
| <span class="source-line-no">014</span><span id="line-14"> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span> |
| <span class="source-line-no">015</span><span id="line-15"> * See the License for the specific language governing permissions and</span> |
| <span class="source-line-no">016</span><span id="line-16"> * limitations under the License.</span> |
| <span class="source-line-no">017</span><span id="line-17"> */</span> |
| <span class="source-line-no">018</span><span id="line-18">package org.apache.hadoop.hbase.mapreduce;</span> |
| <span class="source-line-no">019</span><span id="line-19"></span> |
| <span class="source-line-no">020</span><span id="line-20">import java.io.ByteArrayInputStream;</span> |
| <span class="source-line-no">021</span><span id="line-21">import java.io.DataInput;</span> |
| <span class="source-line-no">022</span><span id="line-22">import java.io.DataInputStream;</span> |
| <span class="source-line-no">023</span><span id="line-23">import java.io.DataOutput;</span> |
| <span class="source-line-no">024</span><span id="line-24">import java.io.IOException;</span> |
| <span class="source-line-no">025</span><span id="line-25">import java.lang.reflect.InvocationTargetException;</span> |
| <span class="source-line-no">026</span><span id="line-26">import java.lang.reflect.Method;</span> |
| <span class="source-line-no">027</span><span id="line-27">import java.util.ArrayList;</span> |
| <span class="source-line-no">028</span><span id="line-28">import java.util.Collections;</span> |
| <span class="source-line-no">029</span><span id="line-29">import java.util.List;</span> |
| <span class="source-line-no">030</span><span id="line-30">import java.util.Locale;</span> |
| <span class="source-line-no">031</span><span id="line-31">import java.util.Map;</span> |
| <span class="source-line-no">032</span><span id="line-32">import java.util.TreeMap;</span> |
| <span class="source-line-no">033</span><span id="line-33">import java.util.UUID;</span> |
| <span class="source-line-no">034</span><span id="line-34">import org.apache.hadoop.conf.Configuration;</span> |
| <span class="source-line-no">035</span><span id="line-35">import org.apache.hadoop.conf.Configured;</span> |
| <span class="source-line-no">036</span><span id="line-36">import org.apache.hadoop.fs.FileSystem;</span> |
| <span class="source-line-no">037</span><span id="line-37">import org.apache.hadoop.fs.Path;</span> |
| <span class="source-line-no">038</span><span id="line-38">import org.apache.hadoop.hbase.Cell;</span> |
| <span class="source-line-no">039</span><span id="line-39">import org.apache.hadoop.hbase.CellComparator;</span> |
| <span class="source-line-no">040</span><span id="line-40">import org.apache.hadoop.hbase.CellUtil;</span> |
| <span class="source-line-no">041</span><span id="line-41">import org.apache.hadoop.hbase.ExtendedCell;</span> |
| <span class="source-line-no">042</span><span id="line-42">import org.apache.hadoop.hbase.HBaseConfiguration;</span> |
| <span class="source-line-no">043</span><span id="line-43">import org.apache.hadoop.hbase.KeyValue;</span> |
| <span class="source-line-no">044</span><span id="line-44">import org.apache.hadoop.hbase.KeyValueUtil;</span> |
| <span class="source-line-no">045</span><span id="line-45">import org.apache.hadoop.hbase.PrivateCellUtil;</span> |
| <span class="source-line-no">046</span><span id="line-46">import org.apache.hadoop.hbase.TableName;</span> |
| <span class="source-line-no">047</span><span id="line-47">import org.apache.hadoop.hbase.Tag;</span> |
| <span class="source-line-no">048</span><span id="line-48">import org.apache.hadoop.hbase.ZooKeeperConnectionException;</span> |
| <span class="source-line-no">049</span><span id="line-49">import org.apache.hadoop.hbase.client.Admin;</span> |
| <span class="source-line-no">050</span><span id="line-50">import org.apache.hadoop.hbase.client.Connection;</span> |
| <span class="source-line-no">051</span><span id="line-51">import org.apache.hadoop.hbase.client.ConnectionFactory;</span> |
| <span class="source-line-no">052</span><span id="line-52">import org.apache.hadoop.hbase.client.Delete;</span> |
| <span class="source-line-no">053</span><span id="line-53">import org.apache.hadoop.hbase.client.Durability;</span> |
| <span class="source-line-no">054</span><span id="line-54">import org.apache.hadoop.hbase.client.Mutation;</span> |
| <span class="source-line-no">055</span><span id="line-55">import org.apache.hadoop.hbase.client.PackagePrivateFieldAccessor;</span> |
| <span class="source-line-no">056</span><span id="line-56">import org.apache.hadoop.hbase.client.Put;</span> |
| <span class="source-line-no">057</span><span id="line-57">import org.apache.hadoop.hbase.client.RegionLocator;</span> |
| <span class="source-line-no">058</span><span id="line-58">import org.apache.hadoop.hbase.client.Result;</span> |
| <span class="source-line-no">059</span><span id="line-59">import org.apache.hadoop.hbase.client.Table;</span> |
| <span class="source-line-no">060</span><span id="line-60">import org.apache.hadoop.hbase.filter.Filter;</span> |
| <span class="source-line-no">061</span><span id="line-61">import org.apache.hadoop.hbase.io.ImmutableBytesWritable;</span> |
| <span class="source-line-no">062</span><span id="line-62">import org.apache.hadoop.hbase.util.Bytes;</span> |
| <span class="source-line-no">063</span><span id="line-63">import org.apache.hadoop.hbase.util.MapReduceExtendedCell;</span> |
| <span class="source-line-no">064</span><span id="line-64">import org.apache.hadoop.hbase.zookeeper.ZKClusterId;</span> |
| <span class="source-line-no">065</span><span id="line-65">import org.apache.hadoop.hbase.zookeeper.ZKWatcher;</span> |
| <span class="source-line-no">066</span><span id="line-66">import org.apache.hadoop.io.RawComparator;</span> |
| <span class="source-line-no">067</span><span id="line-67">import org.apache.hadoop.io.WritableComparable;</span> |
| <span class="source-line-no">068</span><span id="line-68">import org.apache.hadoop.io.WritableComparator;</span> |
| <span class="source-line-no">069</span><span id="line-69">import org.apache.hadoop.mapreduce.Job;</span> |
| <span class="source-line-no">070</span><span id="line-70">import org.apache.hadoop.mapreduce.Partitioner;</span> |
| <span class="source-line-no">071</span><span id="line-71">import org.apache.hadoop.mapreduce.Reducer;</span> |
| <span class="source-line-no">072</span><span id="line-72">import org.apache.hadoop.mapreduce.TaskCounter;</span> |
| <span class="source-line-no">073</span><span id="line-73">import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;</span> |
| <span class="source-line-no">074</span><span id="line-74">import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat;</span> |
| <span class="source-line-no">075</span><span id="line-75">import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;</span> |
| <span class="source-line-no">076</span><span id="line-76">import org.apache.hadoop.mapreduce.lib.partition.TotalOrderPartitioner;</span> |
| <span class="source-line-no">077</span><span id="line-77">import org.apache.hadoop.util.Tool;</span> |
| <span class="source-line-no">078</span><span id="line-78">import org.apache.hadoop.util.ToolRunner;</span> |
| <span class="source-line-no">079</span><span id="line-79">import org.apache.yetus.audience.InterfaceAudience;</span> |
| <span class="source-line-no">080</span><span id="line-80">import org.apache.zookeeper.KeeperException;</span> |
| <span class="source-line-no">081</span><span id="line-81">import org.slf4j.Logger;</span> |
| <span class="source-line-no">082</span><span id="line-82">import org.slf4j.LoggerFactory;</span> |
| <span class="source-line-no">083</span><span id="line-83"></span> |
| <span class="source-line-no">084</span><span id="line-84">/**</span> |
| <span class="source-line-no">085</span><span id="line-85"> * Import data written by {@link Export}.</span> |
| <span class="source-line-no">086</span><span id="line-86"> */</span> |
| <span class="source-line-no">087</span><span id="line-87">@InterfaceAudience.Public</span> |
| <span class="source-line-no">088</span><span id="line-88">public class Import extends Configured implements Tool {</span> |
| <span class="source-line-no">089</span><span id="line-89"> private static final Logger LOG = LoggerFactory.getLogger(Import.class);</span> |
| <span class="source-line-no">090</span><span id="line-90"> final static String NAME = "import";</span> |
| <span class="source-line-no">091</span><span id="line-91"> public final static String CF_RENAME_PROP = "HBASE_IMPORTER_RENAME_CFS";</span> |
| <span class="source-line-no">092</span><span id="line-92"> public final static String BULK_OUTPUT_CONF_KEY = "import.bulk.output";</span> |
| <span class="source-line-no">093</span><span id="line-93"> public final static String FILTER_CLASS_CONF_KEY = "import.filter.class";</span> |
| <span class="source-line-no">094</span><span id="line-94"> public final static String FILTER_ARGS_CONF_KEY = "import.filter.args";</span> |
| <span class="source-line-no">095</span><span id="line-95"> public final static String TABLE_NAME = "import.table.name";</span> |
| <span class="source-line-no">096</span><span id="line-96"> public final static String WAL_DURABILITY = "import.wal.durability";</span> |
| <span class="source-line-no">097</span><span id="line-97"> public final static String HAS_LARGE_RESULT = "import.bulk.hasLargeResult";</span> |
| <span class="source-line-no">098</span><span id="line-98"></span> |
| <span class="source-line-no">099</span><span id="line-99"> private final static String JOB_NAME_CONF_KEY = "mapreduce.job.name";</span> |
| <span class="source-line-no">100</span><span id="line-100"></span> |
| <span class="source-line-no">101</span><span id="line-101"> public static class CellWritableComparablePartitioner</span> |
| <span class="source-line-no">102</span><span id="line-102"> extends Partitioner<CellWritableComparable, Cell> {</span> |
| <span class="source-line-no">103</span><span id="line-103"> private static CellWritableComparable[] START_KEYS = null;</span> |
| <span class="source-line-no">104</span><span id="line-104"></span> |
| <span class="source-line-no">105</span><span id="line-105"> @Override</span> |
| <span class="source-line-no">106</span><span id="line-106"> public int getPartition(CellWritableComparable key, Cell value, int numPartitions) {</span> |
| <span class="source-line-no">107</span><span id="line-107"> for (int i = 0; i < START_KEYS.length; ++i) {</span> |
| <span class="source-line-no">108</span><span id="line-108"> if (key.compareTo(START_KEYS[i]) <= 0) {</span> |
| <span class="source-line-no">109</span><span id="line-109"> return i;</span> |
| <span class="source-line-no">110</span><span id="line-110"> }</span> |
| <span class="source-line-no">111</span><span id="line-111"> }</span> |
| <span class="source-line-no">112</span><span id="line-112"> return START_KEYS.length;</span> |
| <span class="source-line-no">113</span><span id="line-113"> }</span> |
| <span class="source-line-no">114</span><span id="line-114"></span> |
| <span class="source-line-no">115</span><span id="line-115"> }</span> |
| <span class="source-line-no">116</span><span id="line-116"></span> |
| <span class="source-line-no">117</span><span id="line-117"> public static class CellWritableComparable implements WritableComparable<CellWritableComparable> {</span> |
| <span class="source-line-no">118</span><span id="line-118"></span> |
| <span class="source-line-no">119</span><span id="line-119"> private ExtendedCell kv = null;</span> |
| <span class="source-line-no">120</span><span id="line-120"></span> |
| <span class="source-line-no">121</span><span id="line-121"> static {</span> |
| <span class="source-line-no">122</span><span id="line-122"> // register this comparator</span> |
| <span class="source-line-no">123</span><span id="line-123"> WritableComparator.define(CellWritableComparable.class, new CellWritableComparator());</span> |
| <span class="source-line-no">124</span><span id="line-124"> }</span> |
| <span class="source-line-no">125</span><span id="line-125"></span> |
| <span class="source-line-no">126</span><span id="line-126"> public CellWritableComparable() {</span> |
| <span class="source-line-no">127</span><span id="line-127"> }</span> |
| <span class="source-line-no">128</span><span id="line-128"></span> |
| <span class="source-line-no">129</span><span id="line-129"> public CellWritableComparable(Cell kv) {</span> |
| <span class="source-line-no">130</span><span id="line-130"> this.kv = (ExtendedCell) kv;</span> |
| <span class="source-line-no">131</span><span id="line-131"> }</span> |
| <span class="source-line-no">132</span><span id="line-132"></span> |
| <span class="source-line-no">133</span><span id="line-133"> @Override</span> |
| <span class="source-line-no">134</span><span id="line-134"> public void write(DataOutput out) throws IOException {</span> |
| <span class="source-line-no">135</span><span id="line-135"> out.writeInt(PrivateCellUtil.estimatedSerializedSizeOfKey(kv));</span> |
| <span class="source-line-no">136</span><span id="line-136"> out.writeInt(0);</span> |
| <span class="source-line-no">137</span><span id="line-137"> PrivateCellUtil.writeFlatKey(kv, out);</span> |
| <span class="source-line-no">138</span><span id="line-138"> }</span> |
| <span class="source-line-no">139</span><span id="line-139"></span> |
| <span class="source-line-no">140</span><span id="line-140"> @Override</span> |
| <span class="source-line-no">141</span><span id="line-141"> public void readFields(DataInput in) throws IOException {</span> |
| <span class="source-line-no">142</span><span id="line-142"> kv = KeyValue.create(in);</span> |
| <span class="source-line-no">143</span><span id="line-143"> }</span> |
| <span class="source-line-no">144</span><span id="line-144"></span> |
| <span class="source-line-no">145</span><span id="line-145"> @Override</span> |
| <span class="source-line-no">146</span><span id="line-146"> @edu.umd.cs.findbugs.annotations.SuppressWarnings(value = "EQ_COMPARETO_USE_OBJECT_EQUALS",</span> |
| <span class="source-line-no">147</span><span id="line-147"> justification = "This is wrong, yes, but we should be purging Writables, not fixing them")</span> |
| <span class="source-line-no">148</span><span id="line-148"> public int compareTo(CellWritableComparable o) {</span> |
| <span class="source-line-no">149</span><span id="line-149"> return CellComparator.getInstance().compare(this.kv, o.kv);</span> |
| <span class="source-line-no">150</span><span id="line-150"> }</span> |
| <span class="source-line-no">151</span><span id="line-151"></span> |
| <span class="source-line-no">152</span><span id="line-152"> public static class CellWritableComparator extends WritableComparator {</span> |
| <span class="source-line-no">153</span><span id="line-153"></span> |
| <span class="source-line-no">154</span><span id="line-154"> @Override</span> |
| <span class="source-line-no">155</span><span id="line-155"> public int compare(byte[] b1, int s1, int l1, byte[] b2, int s2, int l2) {</span> |
| <span class="source-line-no">156</span><span id="line-156"> try {</span> |
| <span class="source-line-no">157</span><span id="line-157"> CellWritableComparable kv1 = new CellWritableComparable();</span> |
| <span class="source-line-no">158</span><span id="line-158"> kv1.readFields(new DataInputStream(new ByteArrayInputStream(b1, s1, l1)));</span> |
| <span class="source-line-no">159</span><span id="line-159"> CellWritableComparable kv2 = new CellWritableComparable();</span> |
| <span class="source-line-no">160</span><span id="line-160"> kv2.readFields(new DataInputStream(new ByteArrayInputStream(b2, s2, l2)));</span> |
| <span class="source-line-no">161</span><span id="line-161"> return compare(kv1, kv2);</span> |
| <span class="source-line-no">162</span><span id="line-162"> } catch (IOException e) {</span> |
| <span class="source-line-no">163</span><span id="line-163"> throw new RuntimeException(e);</span> |
| <span class="source-line-no">164</span><span id="line-164"> }</span> |
| <span class="source-line-no">165</span><span id="line-165"> }</span> |
| <span class="source-line-no">166</span><span id="line-166"></span> |
| <span class="source-line-no">167</span><span id="line-167"> }</span> |
| <span class="source-line-no">168</span><span id="line-168"></span> |
| <span class="source-line-no">169</span><span id="line-169"> }</span> |
| <span class="source-line-no">170</span><span id="line-170"></span> |
| <span class="source-line-no">171</span><span id="line-171"> public static class CellReducer</span> |
| <span class="source-line-no">172</span><span id="line-172"> extends Reducer<CellWritableComparable, Cell, ImmutableBytesWritable, Cell> {</span> |
| <span class="source-line-no">173</span><span id="line-173"> protected void reduce(CellWritableComparable row, Iterable<Cell> kvs,</span> |
| <span class="source-line-no">174</span><span id="line-174"> Reducer<CellWritableComparable, Cell, ImmutableBytesWritable, Cell>.Context context)</span> |
| <span class="source-line-no">175</span><span id="line-175"> throws java.io.IOException, InterruptedException {</span> |
| <span class="source-line-no">176</span><span id="line-176"> int index = 0;</span> |
| <span class="source-line-no">177</span><span id="line-177"> for (Cell kv : kvs) {</span> |
| <span class="source-line-no">178</span><span id="line-178"> context.write(new ImmutableBytesWritable(CellUtil.cloneRow(kv)),</span> |
| <span class="source-line-no">179</span><span id="line-179"> new MapReduceExtendedCell(PrivateCellUtil.ensureExtendedCell(kv)));</span> |
| <span class="source-line-no">180</span><span id="line-180"> if (++index % 100 == 0) context.setStatus("Wrote " + index + " KeyValues, "</span> |
| <span class="source-line-no">181</span><span id="line-181"> + "and the rowkey whose is being wrote is " + Bytes.toString(kv.getRowArray()));</span> |
| <span class="source-line-no">182</span><span id="line-182"> }</span> |
| <span class="source-line-no">183</span><span id="line-183"> }</span> |
| <span class="source-line-no">184</span><span id="line-184"> }</span> |
| <span class="source-line-no">185</span><span id="line-185"></span> |
| <span class="source-line-no">186</span><span id="line-186"> public static class CellSortImporter extends TableMapper<CellWritableComparable, Cell> {</span> |
| <span class="source-line-no">187</span><span id="line-187"> private Map<byte[], byte[]> cfRenameMap;</span> |
| <span class="source-line-no">188</span><span id="line-188"> private Filter filter;</span> |
| <span class="source-line-no">189</span><span id="line-189"> private static final Logger LOG = LoggerFactory.getLogger(CellImporter.class);</span> |
| <span class="source-line-no">190</span><span id="line-190"></span> |
| <span class="source-line-no">191</span><span id="line-191"> /**</span> |
| <span class="source-line-no">192</span><span id="line-192"> * @param row The current table row key.</span> |
| <span class="source-line-no">193</span><span id="line-193"> * @param value The columns.</span> |
| <span class="source-line-no">194</span><span id="line-194"> * @param context The current context.</span> |
| <span class="source-line-no">195</span><span id="line-195"> * @throws IOException When something is broken with the data.</span> |
| <span class="source-line-no">196</span><span id="line-196"> */</span> |
| <span class="source-line-no">197</span><span id="line-197"> @Override</span> |
| <span class="source-line-no">198</span><span id="line-198"> public void map(ImmutableBytesWritable row, Result value, Context context) throws IOException {</span> |
| <span class="source-line-no">199</span><span id="line-199"> try {</span> |
| <span class="source-line-no">200</span><span id="line-200"> if (LOG.isTraceEnabled()) {</span> |
| <span class="source-line-no">201</span><span id="line-201"> LOG.trace(</span> |
| <span class="source-line-no">202</span><span id="line-202"> "Considering the row." + Bytes.toString(row.get(), row.getOffset(), row.getLength()));</span> |
| <span class="source-line-no">203</span><span id="line-203"> }</span> |
| <span class="source-line-no">204</span><span id="line-204"> if (</span> |
| <span class="source-line-no">205</span><span id="line-205"> filter == null || !filter.filterRowKey(</span> |
| <span class="source-line-no">206</span><span id="line-206"> PrivateCellUtil.createFirstOnRow(row.get(), row.getOffset(), (short) row.getLength()))</span> |
| <span class="source-line-no">207</span><span id="line-207"> ) {</span> |
| <span class="source-line-no">208</span><span id="line-208"> for (ExtendedCell kv : PackagePrivateFieldAccessor.getExtendedRawCells(value)) {</span> |
| <span class="source-line-no">209</span><span id="line-209"> kv = filterKv(filter, kv);</span> |
| <span class="source-line-no">210</span><span id="line-210"> // skip if we filtered it out</span> |
| <span class="source-line-no">211</span><span id="line-211"> if (kv == null) {</span> |
| <span class="source-line-no">212</span><span id="line-212"> continue;</span> |
| <span class="source-line-no">213</span><span id="line-213"> }</span> |
| <span class="source-line-no">214</span><span id="line-214"> Cell ret = convertKv(kv, cfRenameMap);</span> |
| <span class="source-line-no">215</span><span id="line-215"> context.write(new CellWritableComparable(ret), ret);</span> |
| <span class="source-line-no">216</span><span id="line-216"> }</span> |
| <span class="source-line-no">217</span><span id="line-217"> }</span> |
| <span class="source-line-no">218</span><span id="line-218"> } catch (InterruptedException e) {</span> |
| <span class="source-line-no">219</span><span id="line-219"> LOG.error("Interrupted while emitting Cell", e);</span> |
| <span class="source-line-no">220</span><span id="line-220"> Thread.currentThread().interrupt();</span> |
| <span class="source-line-no">221</span><span id="line-221"> }</span> |
| <span class="source-line-no">222</span><span id="line-222"> }</span> |
| <span class="source-line-no">223</span><span id="line-223"></span> |
| <span class="source-line-no">224</span><span id="line-224"> @Override</span> |
| <span class="source-line-no">225</span><span id="line-225"> public void setup(Context context) throws IOException {</span> |
| <span class="source-line-no">226</span><span id="line-226"> cfRenameMap = createCfRenameMap(context.getConfiguration());</span> |
| <span class="source-line-no">227</span><span id="line-227"> filter = instantiateFilter(context.getConfiguration());</span> |
| <span class="source-line-no">228</span><span id="line-228"> int reduceNum = context.getNumReduceTasks();</span> |
| <span class="source-line-no">229</span><span id="line-229"> Configuration conf = context.getConfiguration();</span> |
| <span class="source-line-no">230</span><span id="line-230"> TableName tableName = TableName.valueOf(context.getConfiguration().get(TABLE_NAME));</span> |
| <span class="source-line-no">231</span><span id="line-231"> try (Connection conn = ConnectionFactory.createConnection(conf);</span> |
| <span class="source-line-no">232</span><span id="line-232"> RegionLocator regionLocator = conn.getRegionLocator(tableName)) {</span> |
| <span class="source-line-no">233</span><span id="line-233"> byte[][] startKeys = regionLocator.getStartKeys();</span> |
| <span class="source-line-no">234</span><span id="line-234"> if (startKeys.length != reduceNum) {</span> |
| <span class="source-line-no">235</span><span id="line-235"> throw new IOException("Region split after job initialization");</span> |
| <span class="source-line-no">236</span><span id="line-236"> }</span> |
| <span class="source-line-no">237</span><span id="line-237"> CellWritableComparable[] startKeyWraps = new CellWritableComparable[startKeys.length - 1];</span> |
| <span class="source-line-no">238</span><span id="line-238"> for (int i = 1; i < startKeys.length; ++i) {</span> |
| <span class="source-line-no">239</span><span id="line-239"> startKeyWraps[i - 1] =</span> |
| <span class="source-line-no">240</span><span id="line-240"> new CellWritableComparable(KeyValueUtil.createFirstOnRow(startKeys[i]));</span> |
| <span class="source-line-no">241</span><span id="line-241"> }</span> |
| <span class="source-line-no">242</span><span id="line-242"> CellWritableComparablePartitioner.START_KEYS = startKeyWraps;</span> |
| <span class="source-line-no">243</span><span id="line-243"> }</span> |
| <span class="source-line-no">244</span><span id="line-244"> }</span> |
| <span class="source-line-no">245</span><span id="line-245"> }</span> |
| <span class="source-line-no">246</span><span id="line-246"></span> |
| <span class="source-line-no">247</span><span id="line-247"> /**</span> |
| <span class="source-line-no">248</span><span id="line-248"> * A mapper that just writes out KeyValues.</span> |
| <span class="source-line-no">249</span><span id="line-249"> */</span> |
| <span class="source-line-no">250</span><span id="line-250"> @edu.umd.cs.findbugs.annotations.SuppressWarnings(value = "EQ_COMPARETO_USE_OBJECT_EQUALS",</span> |
| <span class="source-line-no">251</span><span id="line-251"> justification = "Writables are going away and this has been this way forever")</span> |
| <span class="source-line-no">252</span><span id="line-252"> public static class CellImporter extends TableMapper<ImmutableBytesWritable, Cell> {</span> |
| <span class="source-line-no">253</span><span id="line-253"> private Map<byte[], byte[]> cfRenameMap;</span> |
| <span class="source-line-no">254</span><span id="line-254"> private Filter filter;</span> |
| <span class="source-line-no">255</span><span id="line-255"> private static final Logger LOG = LoggerFactory.getLogger(CellImporter.class);</span> |
| <span class="source-line-no">256</span><span id="line-256"></span> |
| <span class="source-line-no">257</span><span id="line-257"> /**</span> |
| <span class="source-line-no">258</span><span id="line-258"> * @param row The current table row key.</span> |
| <span class="source-line-no">259</span><span id="line-259"> * @param value The columns.</span> |
| <span class="source-line-no">260</span><span id="line-260"> * @param context The current context.</span> |
| <span class="source-line-no">261</span><span id="line-261"> * @throws IOException When something is broken with the data.</span> |
| <span class="source-line-no">262</span><span id="line-262"> */</span> |
| <span class="source-line-no">263</span><span id="line-263"> @Override</span> |
| <span class="source-line-no">264</span><span id="line-264"> public void map(ImmutableBytesWritable row, Result value, Context context) throws IOException {</span> |
| <span class="source-line-no">265</span><span id="line-265"> try {</span> |
| <span class="source-line-no">266</span><span id="line-266"> if (LOG.isTraceEnabled()) {</span> |
| <span class="source-line-no">267</span><span id="line-267"> LOG.trace(</span> |
| <span class="source-line-no">268</span><span id="line-268"> "Considering the row." + Bytes.toString(row.get(), row.getOffset(), row.getLength()));</span> |
| <span class="source-line-no">269</span><span id="line-269"> }</span> |
| <span class="source-line-no">270</span><span id="line-270"> if (</span> |
| <span class="source-line-no">271</span><span id="line-271"> filter == null || !filter.filterRowKey(</span> |
| <span class="source-line-no">272</span><span id="line-272"> PrivateCellUtil.createFirstOnRow(row.get(), row.getOffset(), (short) row.getLength()))</span> |
| <span class="source-line-no">273</span><span id="line-273"> ) {</span> |
| <span class="source-line-no">274</span><span id="line-274"> for (ExtendedCell kv : PackagePrivateFieldAccessor.getExtendedRawCells(value)) {</span> |
| <span class="source-line-no">275</span><span id="line-275"> kv = filterKv(filter, kv);</span> |
| <span class="source-line-no">276</span><span id="line-276"> // skip if we filtered it out</span> |
| <span class="source-line-no">277</span><span id="line-277"> if (kv == null) {</span> |
| <span class="source-line-no">278</span><span id="line-278"> continue;</span> |
| <span class="source-line-no">279</span><span id="line-279"> }</span> |
| <span class="source-line-no">280</span><span id="line-280"> context.write(row, new MapReduceExtendedCell(convertKv(kv, cfRenameMap)));</span> |
| <span class="source-line-no">281</span><span id="line-281"> }</span> |
| <span class="source-line-no">282</span><span id="line-282"> }</span> |
| <span class="source-line-no">283</span><span id="line-283"> } catch (InterruptedException e) {</span> |
| <span class="source-line-no">284</span><span id="line-284"> LOG.error("Interrupted while emitting Cell", e);</span> |
| <span class="source-line-no">285</span><span id="line-285"> Thread.currentThread().interrupt();</span> |
| <span class="source-line-no">286</span><span id="line-286"> }</span> |
| <span class="source-line-no">287</span><span id="line-287"> }</span> |
| <span class="source-line-no">288</span><span id="line-288"></span> |
| <span class="source-line-no">289</span><span id="line-289"> @Override</span> |
| <span class="source-line-no">290</span><span id="line-290"> public void setup(Context context) {</span> |
| <span class="source-line-no">291</span><span id="line-291"> cfRenameMap = createCfRenameMap(context.getConfiguration());</span> |
| <span class="source-line-no">292</span><span id="line-292"> filter = instantiateFilter(context.getConfiguration());</span> |
| <span class="source-line-no">293</span><span id="line-293"> }</span> |
| <span class="source-line-no">294</span><span id="line-294"> }</span> |
| <span class="source-line-no">295</span><span id="line-295"></span> |
| <span class="source-line-no">296</span><span id="line-296"> /**</span> |
| <span class="source-line-no">297</span><span id="line-297"> * Write table content out to files in hdfs.</span> |
| <span class="source-line-no">298</span><span id="line-298"> */</span> |
| <span class="source-line-no">299</span><span id="line-299"> public static class Importer extends TableMapper<ImmutableBytesWritable, Mutation> {</span> |
| <span class="source-line-no">300</span><span id="line-300"> private Map<byte[], byte[]> cfRenameMap;</span> |
| <span class="source-line-no">301</span><span id="line-301"> private List<UUID> clusterIds;</span> |
| <span class="source-line-no">302</span><span id="line-302"> private Filter filter;</span> |
| <span class="source-line-no">303</span><span id="line-303"> private Durability durability;</span> |
| <span class="source-line-no">304</span><span id="line-304"></span> |
| <span class="source-line-no">305</span><span id="line-305"> /**</span> |
| <span class="source-line-no">306</span><span id="line-306"> * @param row The current table row key.</span> |
| <span class="source-line-no">307</span><span id="line-307"> * @param value The columns.</span> |
| <span class="source-line-no">308</span><span id="line-308"> * @param context The current context.</span> |
| <span class="source-line-no">309</span><span id="line-309"> * @throws IOException When something is broken with the data.</span> |
| <span class="source-line-no">310</span><span id="line-310"> */</span> |
| <span class="source-line-no">311</span><span id="line-311"> @Override</span> |
| <span class="source-line-no">312</span><span id="line-312"> public void map(ImmutableBytesWritable row, Result value, Context context) throws IOException {</span> |
| <span class="source-line-no">313</span><span id="line-313"> try {</span> |
| <span class="source-line-no">314</span><span id="line-314"> writeResult(row, value, context);</span> |
| <span class="source-line-no">315</span><span id="line-315"> } catch (InterruptedException e) {</span> |
| <span class="source-line-no">316</span><span id="line-316"> LOG.error("Interrupted while writing result", e);</span> |
| <span class="source-line-no">317</span><span id="line-317"> Thread.currentThread().interrupt();</span> |
| <span class="source-line-no">318</span><span id="line-318"> }</span> |
| <span class="source-line-no">319</span><span id="line-319"> }</span> |
| <span class="source-line-no">320</span><span id="line-320"></span> |
| <span class="source-line-no">321</span><span id="line-321"> private void writeResult(ImmutableBytesWritable key, Result result, Context context)</span> |
| <span class="source-line-no">322</span><span id="line-322"> throws IOException, InterruptedException {</span> |
| <span class="source-line-no">323</span><span id="line-323"> Put put = null;</span> |
| <span class="source-line-no">324</span><span id="line-324"> Delete delete = null;</span> |
| <span class="source-line-no">325</span><span id="line-325"> if (LOG.isTraceEnabled()) {</span> |
| <span class="source-line-no">326</span><span id="line-326"> LOG.trace(</span> |
| <span class="source-line-no">327</span><span id="line-327"> "Considering the row." + Bytes.toString(key.get(), key.getOffset(), key.getLength()));</span> |
| <span class="source-line-no">328</span><span id="line-328"> }</span> |
| <span class="source-line-no">329</span><span id="line-329"> if (</span> |
| <span class="source-line-no">330</span><span id="line-330"> filter == null || !filter.filterRowKey(</span> |
| <span class="source-line-no">331</span><span id="line-331"> PrivateCellUtil.createFirstOnRow(key.get(), key.getOffset(), (short) key.getLength()))</span> |
| <span class="source-line-no">332</span><span id="line-332"> ) {</span> |
| <span class="source-line-no">333</span><span id="line-333"> processKV(key, result, context, put, delete);</span> |
| <span class="source-line-no">334</span><span id="line-334"> }</span> |
| <span class="source-line-no">335</span><span id="line-335"> }</span> |
| <span class="source-line-no">336</span><span id="line-336"></span> |
| <span class="source-line-no">337</span><span id="line-337"> protected void processKV(ImmutableBytesWritable key, Result result, Context context, Put put,</span> |
| <span class="source-line-no">338</span><span id="line-338"> Delete delete) throws IOException, InterruptedException {</span> |
| <span class="source-line-no">339</span><span id="line-339"> for (ExtendedCell kv : PackagePrivateFieldAccessor.getExtendedRawCells(result)) {</span> |
| <span class="source-line-no">340</span><span id="line-340"> kv = filterKv(filter, kv);</span> |
| <span class="source-line-no">341</span><span id="line-341"> // skip if we filter it out</span> |
| <span class="source-line-no">342</span><span id="line-342"> if (kv == null) {</span> |
| <span class="source-line-no">343</span><span id="line-343"> continue;</span> |
| <span class="source-line-no">344</span><span id="line-344"> }</span> |
| <span class="source-line-no">345</span><span id="line-345"></span> |
| <span class="source-line-no">346</span><span id="line-346"> kv = convertKv(kv, cfRenameMap);</span> |
| <span class="source-line-no">347</span><span id="line-347"> // Deletes and Puts are gathered and written when finished</span> |
| <span class="source-line-no">348</span><span id="line-348"> /*</span> |
| <span class="source-line-no">349</span><span id="line-349"> * If there are sequence of mutations and tombstones in an Export, and after Import the same</span> |
| <span class="source-line-no">350</span><span id="line-350"> * sequence should be restored as it is. If we combine all Delete tombstones into single</span> |
| <span class="source-line-no">351</span><span id="line-351"> * request then there is chance of ignoring few DeleteFamily tombstones, because if we</span> |
| <span class="source-line-no">352</span><span id="line-352"> * submit multiple DeleteFamily tombstones in single Delete request then we are maintaining</span> |
| <span class="source-line-no">353</span><span id="line-353"> * only newest in hbase table and ignoring other. Check - HBASE-12065</span> |
| <span class="source-line-no">354</span><span id="line-354"> */</span> |
| <span class="source-line-no">355</span><span id="line-355"> if (PrivateCellUtil.isDeleteFamily(kv)) {</span> |
| <span class="source-line-no">356</span><span id="line-356"> Delete deleteFamily = new Delete(key.get());</span> |
| <span class="source-line-no">357</span><span id="line-357"> deleteFamily.add(kv);</span> |
| <span class="source-line-no">358</span><span id="line-358"> if (durability != null) {</span> |
| <span class="source-line-no">359</span><span id="line-359"> deleteFamily.setDurability(durability);</span> |
| <span class="source-line-no">360</span><span id="line-360"> }</span> |
| <span class="source-line-no">361</span><span id="line-361"> deleteFamily.setClusterIds(clusterIds);</span> |
| <span class="source-line-no">362</span><span id="line-362"> context.write(key, deleteFamily);</span> |
| <span class="source-line-no">363</span><span id="line-363"> } else if (CellUtil.isDelete(kv)) {</span> |
| <span class="source-line-no">364</span><span id="line-364"> if (delete == null) {</span> |
| <span class="source-line-no">365</span><span id="line-365"> delete = new Delete(key.get());</span> |
| <span class="source-line-no">366</span><span id="line-366"> }</span> |
| <span class="source-line-no">367</span><span id="line-367"> delete.add(kv);</span> |
| <span class="source-line-no">368</span><span id="line-368"> } else {</span> |
| <span class="source-line-no">369</span><span id="line-369"> if (put == null) {</span> |
| <span class="source-line-no">370</span><span id="line-370"> put = new Put(key.get());</span> |
| <span class="source-line-no">371</span><span id="line-371"> }</span> |
| <span class="source-line-no">372</span><span id="line-372"> addPutToKv(put, kv);</span> |
| <span class="source-line-no">373</span><span id="line-373"> }</span> |
| <span class="source-line-no">374</span><span id="line-374"> }</span> |
| <span class="source-line-no">375</span><span id="line-375"> if (put != null) {</span> |
| <span class="source-line-no">376</span><span id="line-376"> if (durability != null) {</span> |
| <span class="source-line-no">377</span><span id="line-377"> put.setDurability(durability);</span> |
| <span class="source-line-no">378</span><span id="line-378"> }</span> |
| <span class="source-line-no">379</span><span id="line-379"> put.setClusterIds(clusterIds);</span> |
| <span class="source-line-no">380</span><span id="line-380"> context.write(key, put);</span> |
| <span class="source-line-no">381</span><span id="line-381"> }</span> |
| <span class="source-line-no">382</span><span id="line-382"> if (delete != null) {</span> |
| <span class="source-line-no">383</span><span id="line-383"> if (durability != null) {</span> |
| <span class="source-line-no">384</span><span id="line-384"> delete.setDurability(durability);</span> |
| <span class="source-line-no">385</span><span id="line-385"> }</span> |
| <span class="source-line-no">386</span><span id="line-386"> delete.setClusterIds(clusterIds);</span> |
| <span class="source-line-no">387</span><span id="line-387"> context.write(key, delete);</span> |
| <span class="source-line-no">388</span><span id="line-388"> }</span> |
| <span class="source-line-no">389</span><span id="line-389"> }</span> |
| <span class="source-line-no">390</span><span id="line-390"></span> |
| <span class="source-line-no">391</span><span id="line-391"> protected void addPutToKv(Put put, Cell kv) throws IOException {</span> |
| <span class="source-line-no">392</span><span id="line-392"> put.add(kv);</span> |
| <span class="source-line-no">393</span><span id="line-393"> }</span> |
| <span class="source-line-no">394</span><span id="line-394"></span> |
| <span class="source-line-no">395</span><span id="line-395"> @Override</span> |
| <span class="source-line-no">396</span><span id="line-396"> public void setup(Context context) {</span> |
| <span class="source-line-no">397</span><span id="line-397"> LOG.info("Setting up " + getClass() + " mapper.");</span> |
| <span class="source-line-no">398</span><span id="line-398"> Configuration conf = context.getConfiguration();</span> |
| <span class="source-line-no">399</span><span id="line-399"> cfRenameMap = createCfRenameMap(conf);</span> |
| <span class="source-line-no">400</span><span id="line-400"> filter = instantiateFilter(conf);</span> |
| <span class="source-line-no">401</span><span id="line-401"> String durabilityStr = conf.get(WAL_DURABILITY);</span> |
| <span class="source-line-no">402</span><span id="line-402"> if (durabilityStr != null) {</span> |
| <span class="source-line-no">403</span><span id="line-403"> durability = Durability.valueOf(durabilityStr.toUpperCase(Locale.ROOT));</span> |
| <span class="source-line-no">404</span><span id="line-404"> LOG.info("setting WAL durability to " + durability);</span> |
| <span class="source-line-no">405</span><span id="line-405"> } else {</span> |
| <span class="source-line-no">406</span><span id="line-406"> LOG.info("setting WAL durability to default.");</span> |
| <span class="source-line-no">407</span><span id="line-407"> }</span> |
| <span class="source-line-no">408</span><span id="line-408"> // TODO: This is kind of ugly doing setup of ZKW just to read the clusterid.</span> |
| <span class="source-line-no">409</span><span id="line-409"> ZKWatcher zkw = null;</span> |
| <span class="source-line-no">410</span><span id="line-410"> Exception ex = null;</span> |
| <span class="source-line-no">411</span><span id="line-411"> try {</span> |
| <span class="source-line-no">412</span><span id="line-412"> zkw = new ZKWatcher(conf, context.getTaskAttemptID().toString(), null);</span> |
| <span class="source-line-no">413</span><span id="line-413"> clusterIds = Collections.singletonList(ZKClusterId.getUUIDForCluster(zkw));</span> |
| <span class="source-line-no">414</span><span id="line-414"> } catch (ZooKeeperConnectionException e) {</span> |
| <span class="source-line-no">415</span><span id="line-415"> ex = e;</span> |
| <span class="source-line-no">416</span><span id="line-416"> LOG.error("Problem connecting to ZooKeper during task setup", e);</span> |
| <span class="source-line-no">417</span><span id="line-417"> } catch (KeeperException e) {</span> |
| <span class="source-line-no">418</span><span id="line-418"> ex = e;</span> |
| <span class="source-line-no">419</span><span id="line-419"> LOG.error("Problem reading ZooKeeper data during task setup", e);</span> |
| <span class="source-line-no">420</span><span id="line-420"> } catch (IOException e) {</span> |
| <span class="source-line-no">421</span><span id="line-421"> ex = e;</span> |
| <span class="source-line-no">422</span><span id="line-422"> LOG.error("Problem setting up task", e);</span> |
| <span class="source-line-no">423</span><span id="line-423"> } finally {</span> |
| <span class="source-line-no">424</span><span id="line-424"> if (zkw != null) zkw.close();</span> |
| <span class="source-line-no">425</span><span id="line-425"> }</span> |
| <span class="source-line-no">426</span><span id="line-426"> if (clusterIds == null) {</span> |
| <span class="source-line-no">427</span><span id="line-427"> // exit early if setup fails</span> |
| <span class="source-line-no">428</span><span id="line-428"> throw new RuntimeException(ex);</span> |
| <span class="source-line-no">429</span><span id="line-429"> }</span> |
| <span class="source-line-no">430</span><span id="line-430"> }</span> |
| <span class="source-line-no">431</span><span id="line-431"> }</span> |
| <span class="source-line-no">432</span><span id="line-432"></span> |
| <span class="source-line-no">433</span><span id="line-433"> /**</span> |
| <span class="source-line-no">434</span><span id="line-434"> * Create a {@link Filter} to apply to all incoming keys ({@link KeyValue KeyValues}) to</span> |
| <span class="source-line-no">435</span><span id="line-435"> * optionally not include in the job output</span> |
| <span class="source-line-no">436</span><span id="line-436"> * @param conf {@link Configuration} from which to load the filter</span> |
| <span class="source-line-no">437</span><span id="line-437"> * @return the filter to use for the task, or <tt>null</tt> if no filter to should be used</span> |
| <span class="source-line-no">438</span><span id="line-438"> * @throws IllegalArgumentException if the filter is misconfigured</span> |
| <span class="source-line-no">439</span><span id="line-439"> */</span> |
| <span class="source-line-no">440</span><span id="line-440"> public static Filter instantiateFilter(Configuration conf) {</span> |
| <span class="source-line-no">441</span><span id="line-441"> // get the filter, if it was configured</span> |
| <span class="source-line-no">442</span><span id="line-442"> Class<? extends Filter> filterClass = conf.getClass(FILTER_CLASS_CONF_KEY, null, Filter.class);</span> |
| <span class="source-line-no">443</span><span id="line-443"> if (filterClass == null) {</span> |
| <span class="source-line-no">444</span><span id="line-444"> LOG.debug("No configured filter class, accepting all keyvalues.");</span> |
| <span class="source-line-no">445</span><span id="line-445"> return null;</span> |
| <span class="source-line-no">446</span><span id="line-446"> }</span> |
| <span class="source-line-no">447</span><span id="line-447"> LOG.debug("Attempting to create filter:" + filterClass);</span> |
| <span class="source-line-no">448</span><span id="line-448"> String[] filterArgs = conf.getStrings(FILTER_ARGS_CONF_KEY);</span> |
| <span class="source-line-no">449</span><span id="line-449"> ArrayList<byte[]> quotedArgs = toQuotedByteArrays(filterArgs);</span> |
| <span class="source-line-no">450</span><span id="line-450"> try {</span> |
| <span class="source-line-no">451</span><span id="line-451"> Method m = filterClass.getMethod("createFilterFromArguments", ArrayList.class);</span> |
| <span class="source-line-no">452</span><span id="line-452"> return (Filter) m.invoke(null, quotedArgs);</span> |
| <span class="source-line-no">453</span><span id="line-453"> } catch (IllegalAccessException e) {</span> |
| <span class="source-line-no">454</span><span id="line-454"> LOG.error("Couldn't instantiate filter!", e);</span> |
| <span class="source-line-no">455</span><span id="line-455"> throw new RuntimeException(e);</span> |
| <span class="source-line-no">456</span><span id="line-456"> } catch (SecurityException e) {</span> |
| <span class="source-line-no">457</span><span id="line-457"> LOG.error("Couldn't instantiate filter!", e);</span> |
| <span class="source-line-no">458</span><span id="line-458"> throw new RuntimeException(e);</span> |
| <span class="source-line-no">459</span><span id="line-459"> } catch (NoSuchMethodException e) {</span> |
| <span class="source-line-no">460</span><span id="line-460"> LOG.error("Couldn't instantiate filter!", e);</span> |
| <span class="source-line-no">461</span><span id="line-461"> throw new RuntimeException(e);</span> |
| <span class="source-line-no">462</span><span id="line-462"> } catch (IllegalArgumentException e) {</span> |
| <span class="source-line-no">463</span><span id="line-463"> LOG.error("Couldn't instantiate filter!", e);</span> |
| <span class="source-line-no">464</span><span id="line-464"> throw new RuntimeException(e);</span> |
| <span class="source-line-no">465</span><span id="line-465"> } catch (InvocationTargetException e) {</span> |
| <span class="source-line-no">466</span><span id="line-466"> LOG.error("Couldn't instantiate filter!", e);</span> |
| <span class="source-line-no">467</span><span id="line-467"> throw new RuntimeException(e);</span> |
| <span class="source-line-no">468</span><span id="line-468"> }</span> |
| <span class="source-line-no">469</span><span id="line-469"> }</span> |
| <span class="source-line-no">470</span><span id="line-470"></span> |
| <span class="source-line-no">471</span><span id="line-471"> private static ArrayList<byte[]> toQuotedByteArrays(String... stringArgs) {</span> |
| <span class="source-line-no">472</span><span id="line-472"> ArrayList<byte[]> quotedArgs = new ArrayList<>();</span> |
| <span class="source-line-no">473</span><span id="line-473"> for (String stringArg : stringArgs) {</span> |
| <span class="source-line-no">474</span><span id="line-474"> // all the filters' instantiation methods expected quoted args since they are coming from</span> |
| <span class="source-line-no">475</span><span id="line-475"> // the shell, so add them here, though it shouldn't really be needed :-/</span> |
| <span class="source-line-no">476</span><span id="line-476"> quotedArgs.add(Bytes.toBytes("'" + stringArg + "'"));</span> |
| <span class="source-line-no">477</span><span id="line-477"> }</span> |
| <span class="source-line-no">478</span><span id="line-478"> return quotedArgs;</span> |
| <span class="source-line-no">479</span><span id="line-479"> }</span> |
| <span class="source-line-no">480</span><span id="line-480"></span> |
| <span class="source-line-no">481</span><span id="line-481"> /**</span> |
| <span class="source-line-no">482</span><span id="line-482"> * Attempt to filter out the keyvalue</span> |
| <span class="source-line-no">483</span><span id="line-483"> * @param c {@link Cell} on which to apply the filter</span> |
| <span class="source-line-no">484</span><span id="line-484"> * @return <tt>null</tt> if the key should not be written, otherwise returns the original</span> |
| <span class="source-line-no">485</span><span id="line-485"> * {@link Cell}</span> |
| <span class="source-line-no">486</span><span id="line-486"> */</span> |
| <span class="source-line-no">487</span><span id="line-487"> public static ExtendedCell filterKv(Filter filter, ExtendedCell c) throws IOException {</span> |
| <span class="source-line-no">488</span><span id="line-488"> // apply the filter and skip this kv if the filter doesn't apply</span> |
| <span class="source-line-no">489</span><span id="line-489"> if (filter != null) {</span> |
| <span class="source-line-no">490</span><span id="line-490"> Filter.ReturnCode code = filter.filterCell(c);</span> |
| <span class="source-line-no">491</span><span id="line-491"> if (LOG.isTraceEnabled()) {</span> |
| <span class="source-line-no">492</span><span id="line-492"> LOG.trace("Filter returned:" + code + " for the cell:" + c);</span> |
| <span class="source-line-no">493</span><span id="line-493"> }</span> |
| <span class="source-line-no">494</span><span id="line-494"> // if its not an accept type, then skip this kv</span> |
| <span class="source-line-no">495</span><span id="line-495"> if (</span> |
| <span class="source-line-no">496</span><span id="line-496"> !(code.equals(Filter.ReturnCode.INCLUDE)</span> |
| <span class="source-line-no">497</span><span id="line-497"> || code.equals(Filter.ReturnCode.INCLUDE_AND_NEXT_COL))</span> |
| <span class="source-line-no">498</span><span id="line-498"> ) {</span> |
| <span class="source-line-no">499</span><span id="line-499"> return null;</span> |
| <span class="source-line-no">500</span><span id="line-500"> }</span> |
| <span class="source-line-no">501</span><span id="line-501"> }</span> |
| <span class="source-line-no">502</span><span id="line-502"> return c;</span> |
| <span class="source-line-no">503</span><span id="line-503"> }</span> |
| <span class="source-line-no">504</span><span id="line-504"></span> |
| <span class="source-line-no">505</span><span id="line-505"> // helper: create a new KeyValue based on CF rename map</span> |
| <span class="source-line-no">506</span><span id="line-506"> private static ExtendedCell convertKv(ExtendedCell kv, Map<byte[], byte[]> cfRenameMap) {</span> |
| <span class="source-line-no">507</span><span id="line-507"> if (cfRenameMap != null) {</span> |
| <span class="source-line-no">508</span><span id="line-508"> // If there's a rename mapping for this CF, create a new KeyValue</span> |
| <span class="source-line-no">509</span><span id="line-509"> byte[] newCfName = cfRenameMap.get(CellUtil.cloneFamily(kv));</span> |
| <span class="source-line-no">510</span><span id="line-510"> if (newCfName != null) {</span> |
| <span class="source-line-no">511</span><span id="line-511"> List<Tag> tags = PrivateCellUtil.getTags(kv);</span> |
| <span class="source-line-no">512</span><span id="line-512"> kv = new KeyValue(kv.getRowArray(), // row buffer</span> |
| <span class="source-line-no">513</span><span id="line-513"> kv.getRowOffset(), // row offset</span> |
| <span class="source-line-no">514</span><span id="line-514"> kv.getRowLength(), // row length</span> |
| <span class="source-line-no">515</span><span id="line-515"> newCfName, // CF buffer</span> |
| <span class="source-line-no">516</span><span id="line-516"> 0, // CF offset</span> |
| <span class="source-line-no">517</span><span id="line-517"> newCfName.length, // CF length</span> |
| <span class="source-line-no">518</span><span id="line-518"> kv.getQualifierArray(), // qualifier buffer</span> |
| <span class="source-line-no">519</span><span id="line-519"> kv.getQualifierOffset(), // qualifier offset</span> |
| <span class="source-line-no">520</span><span id="line-520"> kv.getQualifierLength(), // qualifier length</span> |
| <span class="source-line-no">521</span><span id="line-521"> kv.getTimestamp(), // timestamp</span> |
| <span class="source-line-no">522</span><span id="line-522"> KeyValue.Type.codeToType(kv.getTypeByte()), // KV Type</span> |
| <span class="source-line-no">523</span><span id="line-523"> kv.getValueArray(), // value buffer</span> |
| <span class="source-line-no">524</span><span id="line-524"> kv.getValueOffset(), // value offset</span> |
| <span class="source-line-no">525</span><span id="line-525"> kv.getValueLength(), // value length</span> |
| <span class="source-line-no">526</span><span id="line-526"> tags.size() == 0 ? null : tags);</span> |
| <span class="source-line-no">527</span><span id="line-527"> }</span> |
| <span class="source-line-no">528</span><span id="line-528"> }</span> |
| <span class="source-line-no">529</span><span id="line-529"> return kv;</span> |
| <span class="source-line-no">530</span><span id="line-530"> }</span> |
| <span class="source-line-no">531</span><span id="line-531"></span> |
| <span class="source-line-no">532</span><span id="line-532"> // helper: make a map from sourceCfName to destCfName by parsing a config key</span> |
| <span class="source-line-no">533</span><span id="line-533"> private static Map<byte[], byte[]> createCfRenameMap(Configuration conf) {</span> |
| <span class="source-line-no">534</span><span id="line-534"> Map<byte[], byte[]> cfRenameMap = null;</span> |
| <span class="source-line-no">535</span><span id="line-535"> String allMappingsPropVal = conf.get(CF_RENAME_PROP);</span> |
| <span class="source-line-no">536</span><span id="line-536"> if (allMappingsPropVal != null) {</span> |
| <span class="source-line-no">537</span><span id="line-537"> // The conf value format should be sourceCf1:destCf1,sourceCf2:destCf2,...</span> |
| <span class="source-line-no">538</span><span id="line-538"> String[] allMappings = allMappingsPropVal.split(",");</span> |
| <span class="source-line-no">539</span><span id="line-539"> for (String mapping : allMappings) {</span> |
| <span class="source-line-no">540</span><span id="line-540"> if (cfRenameMap == null) {</span> |
| <span class="source-line-no">541</span><span id="line-541"> cfRenameMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);</span> |
| <span class="source-line-no">542</span><span id="line-542"> }</span> |
| <span class="source-line-no">543</span><span id="line-543"> String[] srcAndDest = mapping.split(":");</span> |
| <span class="source-line-no">544</span><span id="line-544"> if (srcAndDest.length != 2) {</span> |
| <span class="source-line-no">545</span><span id="line-545"> continue;</span> |
| <span class="source-line-no">546</span><span id="line-546"> }</span> |
| <span class="source-line-no">547</span><span id="line-547"> cfRenameMap.put(Bytes.toBytes(srcAndDest[0]), Bytes.toBytes(srcAndDest[1]));</span> |
| <span class="source-line-no">548</span><span id="line-548"> }</span> |
| <span class="source-line-no">549</span><span id="line-549"> }</span> |
| <span class="source-line-no">550</span><span id="line-550"> return cfRenameMap;</span> |
| <span class="source-line-no">551</span><span id="line-551"> }</span> |
| <span class="source-line-no">552</span><span id="line-552"></span> |
| <span class="source-line-no">553</span><span id="line-553"> /**</span> |
| <span class="source-line-no">554</span><span id="line-554"> * <p></span> |
| <span class="source-line-no">555</span><span id="line-555"> * Sets a configuration property with key {@link #CF_RENAME_PROP} in conf that tells the mapper</span> |
| <span class="source-line-no">556</span><span id="line-556"> * how to rename column families.</span> |
| <span class="source-line-no">557</span><span id="line-557"> * <p></span> |
| <span class="source-line-no">558</span><span id="line-558"> * Alternately, instead of calling this function, you could set the configuration key</span> |
| <span class="source-line-no">559</span><span id="line-559"> * {@link #CF_RENAME_PROP} yourself. The value should look like</span> |
| <span class="source-line-no">560</span><span id="line-560"> *</span> |
| <span class="source-line-no">561</span><span id="line-561"> * <pre></span> |
| <span class="source-line-no">562</span><span id="line-562"> * srcCf1:destCf1,srcCf2:destCf2,....</span> |
| <span class="source-line-no">563</span><span id="line-563"> * </pre></span> |
| <span class="source-line-no">564</span><span id="line-564"> *</span> |
| <span class="source-line-no">565</span><span id="line-565"> * . This would have the same effect on the mapper behavior.</span> |
| <span class="source-line-no">566</span><span id="line-566"> * @param conf the Configuration in which the {@link #CF_RENAME_PROP} key will be set</span> |
| <span class="source-line-no">567</span><span id="line-567"> * @param renameMap a mapping from source CF names to destination CF names</span> |
| <span class="source-line-no">568</span><span id="line-568"> */</span> |
| <span class="source-line-no">569</span><span id="line-569"> static public void configureCfRenaming(Configuration conf, Map<String, String> renameMap) {</span> |
| <span class="source-line-no">570</span><span id="line-570"> StringBuilder sb = new StringBuilder();</span> |
| <span class="source-line-no">571</span><span id="line-571"> for (Map.Entry<String, String> entry : renameMap.entrySet()) {</span> |
| <span class="source-line-no">572</span><span id="line-572"> String sourceCf = entry.getKey();</span> |
| <span class="source-line-no">573</span><span id="line-573"> String destCf = entry.getValue();</span> |
| <span class="source-line-no">574</span><span id="line-574"></span> |
| <span class="source-line-no">575</span><span id="line-575"> if (</span> |
| <span class="source-line-no">576</span><span id="line-576"> sourceCf.contains(":") || sourceCf.contains(",") || destCf.contains(":")</span> |
| <span class="source-line-no">577</span><span id="line-577"> || destCf.contains(",")</span> |
| <span class="source-line-no">578</span><span id="line-578"> ) {</span> |
| <span class="source-line-no">579</span><span id="line-579"> throw new IllegalArgumentException(</span> |
| <span class="source-line-no">580</span><span id="line-580"> "Illegal character in CF names: " + sourceCf + ", " + destCf);</span> |
| <span class="source-line-no">581</span><span id="line-581"> }</span> |
| <span class="source-line-no">582</span><span id="line-582"></span> |
| <span class="source-line-no">583</span><span id="line-583"> if (sb.length() != 0) {</span> |
| <span class="source-line-no">584</span><span id="line-584"> sb.append(",");</span> |
| <span class="source-line-no">585</span><span id="line-585"> }</span> |
| <span class="source-line-no">586</span><span id="line-586"> sb.append(sourceCf + ":" + destCf);</span> |
| <span class="source-line-no">587</span><span id="line-587"> }</span> |
| <span class="source-line-no">588</span><span id="line-588"> conf.set(CF_RENAME_PROP, sb.toString());</span> |
| <span class="source-line-no">589</span><span id="line-589"> }</span> |
| <span class="source-line-no">590</span><span id="line-590"></span> |
| <span class="source-line-no">591</span><span id="line-591"> /**</span> |
| <span class="source-line-no">592</span><span id="line-592"> * Add a Filter to be instantiated on import</span> |
| <span class="source-line-no">593</span><span id="line-593"> * @param conf Configuration to update (will be passed to the job)</span> |
| <span class="source-line-no">594</span><span id="line-594"> * @param clazz {@link Filter} subclass to instantiate on the server.</span> |
| <span class="source-line-no">595</span><span id="line-595"> * @param filterArgs List of arguments to pass to the filter on instantiation</span> |
| <span class="source-line-no">596</span><span id="line-596"> */</span> |
| <span class="source-line-no">597</span><span id="line-597"> public static void addFilterAndArguments(Configuration conf, Class<? extends Filter> clazz,</span> |
| <span class="source-line-no">598</span><span id="line-598"> List<String> filterArgs) throws IOException {</span> |
| <span class="source-line-no">599</span><span id="line-599"> conf.set(Import.FILTER_CLASS_CONF_KEY, clazz.getName());</span> |
| <span class="source-line-no">600</span><span id="line-600"> conf.setStrings(Import.FILTER_ARGS_CONF_KEY, filterArgs.toArray(new String[filterArgs.size()]));</span> |
| <span class="source-line-no">601</span><span id="line-601"> }</span> |
| <span class="source-line-no">602</span><span id="line-602"></span> |
| <span class="source-line-no">603</span><span id="line-603"> /**</span> |
| <span class="source-line-no">604</span><span id="line-604"> * Sets up the actual job.</span> |
| <span class="source-line-no">605</span><span id="line-605"> * @param conf The current configuration.</span> |
| <span class="source-line-no">606</span><span id="line-606"> * @param args The command line parameters.</span> |
| <span class="source-line-no">607</span><span id="line-607"> * @return The newly created job.</span> |
| <span class="source-line-no">608</span><span id="line-608"> * @throws IOException When setting up the job fails.</span> |
| <span class="source-line-no">609</span><span id="line-609"> */</span> |
| <span class="source-line-no">610</span><span id="line-610"> public static Job createSubmittableJob(Configuration conf, String[] args) throws IOException {</span> |
| <span class="source-line-no">611</span><span id="line-611"> TableName tableName = TableName.valueOf(args[0]);</span> |
| <span class="source-line-no">612</span><span id="line-612"> conf.set(TABLE_NAME, tableName.getNameAsString());</span> |
| <span class="source-line-no">613</span><span id="line-613"> Path inputDir = new Path(args[1]);</span> |
| <span class="source-line-no">614</span><span id="line-614"> Job job = Job.getInstance(conf, conf.get(JOB_NAME_CONF_KEY, NAME + "_" + tableName));</span> |
| <span class="source-line-no">615</span><span id="line-615"> job.setJarByClass(Importer.class);</span> |
| <span class="source-line-no">616</span><span id="line-616"> FileInputFormat.setInputPaths(job, inputDir);</span> |
| <span class="source-line-no">617</span><span id="line-617"> job.setInputFormatClass(SequenceFileInputFormat.class);</span> |
| <span class="source-line-no">618</span><span id="line-618"> String hfileOutPath = conf.get(BULK_OUTPUT_CONF_KEY);</span> |
| <span class="source-line-no">619</span><span id="line-619"></span> |
| <span class="source-line-no">620</span><span id="line-620"> // make sure we get the filter in the jars</span> |
| <span class="source-line-no">621</span><span id="line-621"> try {</span> |
| <span class="source-line-no">622</span><span id="line-622"> Class<? extends Filter> filter = conf.getClass(FILTER_CLASS_CONF_KEY, null, Filter.class);</span> |
| <span class="source-line-no">623</span><span id="line-623"> if (filter != null) {</span> |
| <span class="source-line-no">624</span><span id="line-624"> TableMapReduceUtil.addDependencyJarsForClasses(conf, filter);</span> |
| <span class="source-line-no">625</span><span id="line-625"> }</span> |
| <span class="source-line-no">626</span><span id="line-626"> } catch (Exception e) {</span> |
| <span class="source-line-no">627</span><span id="line-627"> throw new IOException(e);</span> |
| <span class="source-line-no">628</span><span id="line-628"> }</span> |
| <span class="source-line-no">629</span><span id="line-629"></span> |
| <span class="source-line-no">630</span><span id="line-630"> if (hfileOutPath != null && conf.getBoolean(HAS_LARGE_RESULT, false)) {</span> |
| <span class="source-line-no">631</span><span id="line-631"> LOG.info("Use Large Result!!");</span> |
| <span class="source-line-no">632</span><span id="line-632"> try (Connection conn = ConnectionFactory.createConnection(conf);</span> |
| <span class="source-line-no">633</span><span id="line-633"> Table table = conn.getTable(tableName);</span> |
| <span class="source-line-no">634</span><span id="line-634"> RegionLocator regionLocator = conn.getRegionLocator(tableName)) {</span> |
| <span class="source-line-no">635</span><span id="line-635"> HFileOutputFormat2.configureIncrementalLoad(job, table.getDescriptor(), regionLocator);</span> |
| <span class="source-line-no">636</span><span id="line-636"> job.setMapperClass(CellSortImporter.class);</span> |
| <span class="source-line-no">637</span><span id="line-637"> job.setReducerClass(CellReducer.class);</span> |
| <span class="source-line-no">638</span><span id="line-638"> Path outputDir = new Path(hfileOutPath);</span> |
| <span class="source-line-no">639</span><span id="line-639"> FileOutputFormat.setOutputPath(job, outputDir);</span> |
| <span class="source-line-no">640</span><span id="line-640"> job.setMapOutputKeyClass(CellWritableComparable.class);</span> |
| <span class="source-line-no">641</span><span id="line-641"> job.setMapOutputValueClass(MapReduceExtendedCell.class);</span> |
| <span class="source-line-no">642</span><span id="line-642"> job.getConfiguration().setClass("mapreduce.job.output.key.comparator.class",</span> |
| <span class="source-line-no">643</span><span id="line-643"> CellWritableComparable.CellWritableComparator.class, RawComparator.class);</span> |
| <span class="source-line-no">644</span><span id="line-644"> Path partitionsPath =</span> |
| <span class="source-line-no">645</span><span id="line-645"> new Path(TotalOrderPartitioner.getPartitionFile(job.getConfiguration()));</span> |
| <span class="source-line-no">646</span><span id="line-646"> FileSystem fs = FileSystem.get(job.getConfiguration());</span> |
| <span class="source-line-no">647</span><span id="line-647"> fs.deleteOnExit(partitionsPath);</span> |
| <span class="source-line-no">648</span><span id="line-648"> job.setPartitionerClass(CellWritableComparablePartitioner.class);</span> |
| <span class="source-line-no">649</span><span id="line-649"> job.setNumReduceTasks(regionLocator.getStartKeys().length);</span> |
| <span class="source-line-no">650</span><span id="line-650"> TableMapReduceUtil.addDependencyJarsForClasses(job.getConfiguration(),</span> |
| <span class="source-line-no">651</span><span id="line-651"> org.apache.hbase.thirdparty.com.google.common.base.Preconditions.class);</span> |
| <span class="source-line-no">652</span><span id="line-652"> }</span> |
| <span class="source-line-no">653</span><span id="line-653"> } else if (hfileOutPath != null) {</span> |
| <span class="source-line-no">654</span><span id="line-654"> LOG.info("writing to hfiles for bulk load.");</span> |
| <span class="source-line-no">655</span><span id="line-655"> job.setMapperClass(CellImporter.class);</span> |
| <span class="source-line-no">656</span><span id="line-656"> try (Connection conn = ConnectionFactory.createConnection(conf);</span> |
| <span class="source-line-no">657</span><span id="line-657"> Table table = conn.getTable(tableName);</span> |
| <span class="source-line-no">658</span><span id="line-658"> RegionLocator regionLocator = conn.getRegionLocator(tableName)) {</span> |
| <span class="source-line-no">659</span><span id="line-659"> job.setReducerClass(CellSortReducer.class);</span> |
| <span class="source-line-no">660</span><span id="line-660"> Path outputDir = new Path(hfileOutPath);</span> |
| <span class="source-line-no">661</span><span id="line-661"> FileOutputFormat.setOutputPath(job, outputDir);</span> |
| <span class="source-line-no">662</span><span id="line-662"> job.setMapOutputKeyClass(ImmutableBytesWritable.class);</span> |
| <span class="source-line-no">663</span><span id="line-663"> job.setMapOutputValueClass(MapReduceExtendedCell.class);</span> |
| <span class="source-line-no">664</span><span id="line-664"> HFileOutputFormat2.configureIncrementalLoad(job, table.getDescriptor(), regionLocator);</span> |
| <span class="source-line-no">665</span><span id="line-665"> TableMapReduceUtil.addDependencyJarsForClasses(job.getConfiguration(),</span> |
| <span class="source-line-no">666</span><span id="line-666"> org.apache.hbase.thirdparty.com.google.common.base.Preconditions.class);</span> |
| <span class="source-line-no">667</span><span id="line-667"> }</span> |
| <span class="source-line-no">668</span><span id="line-668"> } else {</span> |
| <span class="source-line-no">669</span><span id="line-669"> LOG.info("writing directly to table from Mapper.");</span> |
| <span class="source-line-no">670</span><span id="line-670"> // No reducers. Just write straight to table. Call initTableReducerJob</span> |
| <span class="source-line-no">671</span><span id="line-671"> // because it sets up the TableOutputFormat.</span> |
| <span class="source-line-no">672</span><span id="line-672"> job.setMapperClass(Importer.class);</span> |
| <span class="source-line-no">673</span><span id="line-673"> TableMapReduceUtil.initTableReducerJob(tableName.getNameAsString(), null, job);</span> |
| <span class="source-line-no">674</span><span id="line-674"> job.setNumReduceTasks(0);</span> |
| <span class="source-line-no">675</span><span id="line-675"> }</span> |
| <span class="source-line-no">676</span><span id="line-676"> return job;</span> |
| <span class="source-line-no">677</span><span id="line-677"> }</span> |
| <span class="source-line-no">678</span><span id="line-678"></span> |
| <span class="source-line-no">679</span><span id="line-679"> /*</span> |
| <span class="source-line-no">680</span><span id="line-680"> * @param errorMsg Error message. Can be null.</span> |
| <span class="source-line-no">681</span><span id="line-681"> */</span> |
| <span class="source-line-no">682</span><span id="line-682"> private static void usage(final String errorMsg) {</span> |
| <span class="source-line-no">683</span><span id="line-683"> if (errorMsg != null && errorMsg.length() > 0) {</span> |
| <span class="source-line-no">684</span><span id="line-684"> System.err.println("ERROR: " + errorMsg);</span> |
| <span class="source-line-no">685</span><span id="line-685"> }</span> |
| <span class="source-line-no">686</span><span id="line-686"> System.err.println("Usage: Import [options] <tablename> <inputdir>");</span> |
| <span class="source-line-no">687</span><span id="line-687"> System.err.println("By default Import will load data directly into HBase. To instead generate");</span> |
| <span class="source-line-no">688</span><span id="line-688"> System.err.println("HFiles of data to prepare for a bulk data load, pass the option:");</span> |
| <span class="source-line-no">689</span><span id="line-689"> System.err.println(" -D" + BULK_OUTPUT_CONF_KEY + "=/path/for/output");</span> |
| <span class="source-line-no">690</span><span id="line-690"> System.err.println("If there is a large result that includes too much Cell "</span> |
| <span class="source-line-no">691</span><span id="line-691"> + "whitch can occur OOME caused by the memery sort in reducer, pass the option:");</span> |
| <span class="source-line-no">692</span><span id="line-692"> System.err.println(" -D" + HAS_LARGE_RESULT + "=true");</span> |
| <span class="source-line-no">693</span><span id="line-693"> System.err</span> |
| <span class="source-line-no">694</span><span id="line-694"> .println(" To apply a generic org.apache.hadoop.hbase.filter.Filter to the input, use");</span> |
| <span class="source-line-no">695</span><span id="line-695"> System.err.println(" -D" + FILTER_CLASS_CONF_KEY + "=<name of filter class>");</span> |
| <span class="source-line-no">696</span><span id="line-696"> System.err.println(" -D" + FILTER_ARGS_CONF_KEY + "=<comma separated list of args for filter");</span> |
| <span class="source-line-no">697</span><span id="line-697"> System.err.println(" NOTE: The filter will be applied BEFORE doing key renames via the "</span> |
| <span class="source-line-no">698</span><span id="line-698"> + CF_RENAME_PROP + " property. Futher, filters will only use the"</span> |
| <span class="source-line-no">699</span><span id="line-699"> + " Filter#filterRowKey(byte[] buffer, int offset, int length) method to identify "</span> |
| <span class="source-line-no">700</span><span id="line-700"> + " whether the current row needs to be ignored completely for processing and "</span> |
| <span class="source-line-no">701</span><span id="line-701"> + " Filter#filterCell(Cell) method to determine if the Cell should be added;"</span> |
| <span class="source-line-no">702</span><span id="line-702"> + " Filter.ReturnCode#INCLUDE and #INCLUDE_AND_NEXT_COL will be considered as including"</span> |
| <span class="source-line-no">703</span><span id="line-703"> + " the Cell.");</span> |
| <span class="source-line-no">704</span><span id="line-704"> System.err.println("To import data exported from HBase 0.94, use");</span> |
| <span class="source-line-no">705</span><span id="line-705"> System.err.println(" -Dhbase.import.version=0.94");</span> |
| <span class="source-line-no">706</span><span id="line-706"> System.err.println(" -D " + JOB_NAME_CONF_KEY</span> |
| <span class="source-line-no">707</span><span id="line-707"> + "=jobName - use the specified mapreduce job name for the import");</span> |
| <span class="source-line-no">708</span><span id="line-708"> System.err.println("For performance consider the following options:\n"</span> |
| <span class="source-line-no">709</span><span id="line-709"> + " -Dmapreduce.map.speculative=false\n" + " -Dmapreduce.reduce.speculative=false\n"</span> |
| <span class="source-line-no">710</span><span id="line-710"> + " -D" + WAL_DURABILITY + "=<Used while writing data to hbase."</span> |
| <span class="source-line-no">711</span><span id="line-711"> + " Allowed values are the supported durability values"</span> |
| <span class="source-line-no">712</span><span id="line-712"> + " like SKIP_WAL/ASYNC_WAL/SYNC_WAL/...>");</span> |
| <span class="source-line-no">713</span><span id="line-713"> }</span> |
| <span class="source-line-no">714</span><span id="line-714"></span> |
| <span class="source-line-no">715</span><span id="line-715"> /**</span> |
| <span class="source-line-no">716</span><span id="line-716"> * If the durability is set to {@link Durability#SKIP_WAL} and the data is imported to hbase, we</span> |
| <span class="source-line-no">717</span><span id="line-717"> * need to flush all the regions of the table as the data is held in memory and is also not</span> |
| <span class="source-line-no">718</span><span id="line-718"> * present in the Write Ahead Log to replay in scenarios of a crash. This method flushes all the</span> |
| <span class="source-line-no">719</span><span id="line-719"> * regions of the table in the scenarios of import data to hbase with {@link Durability#SKIP_WAL}</span> |
| <span class="source-line-no">720</span><span id="line-720"> */</span> |
| <span class="source-line-no">721</span><span id="line-721"> public static void flushRegionsIfNecessary(Configuration conf)</span> |
| <span class="source-line-no">722</span><span id="line-722"> throws IOException, InterruptedException {</span> |
| <span class="source-line-no">723</span><span id="line-723"> String tableName = conf.get(TABLE_NAME);</span> |
| <span class="source-line-no">724</span><span id="line-724"> Admin hAdmin = null;</span> |
| <span class="source-line-no">725</span><span id="line-725"> Connection connection = null;</span> |
| <span class="source-line-no">726</span><span id="line-726"> String durability = conf.get(WAL_DURABILITY);</span> |
| <span class="source-line-no">727</span><span id="line-727"> // Need to flush if the data is written to hbase and skip wal is enabled.</span> |
| <span class="source-line-no">728</span><span id="line-728"> if (</span> |
| <span class="source-line-no">729</span><span id="line-729"> conf.get(BULK_OUTPUT_CONF_KEY) == null && durability != null</span> |
| <span class="source-line-no">730</span><span id="line-730"> && Durability.SKIP_WAL.name().equalsIgnoreCase(durability)</span> |
| <span class="source-line-no">731</span><span id="line-731"> ) {</span> |
| <span class="source-line-no">732</span><span id="line-732"> LOG.info("Flushing all data that skipped the WAL.");</span> |
| <span class="source-line-no">733</span><span id="line-733"> try {</span> |
| <span class="source-line-no">734</span><span id="line-734"> connection = ConnectionFactory.createConnection(conf);</span> |
| <span class="source-line-no">735</span><span id="line-735"> hAdmin = connection.getAdmin();</span> |
| <span class="source-line-no">736</span><span id="line-736"> hAdmin.flush(TableName.valueOf(tableName));</span> |
| <span class="source-line-no">737</span><span id="line-737"> } finally {</span> |
| <span class="source-line-no">738</span><span id="line-738"> if (hAdmin != null) {</span> |
| <span class="source-line-no">739</span><span id="line-739"> hAdmin.close();</span> |
| <span class="source-line-no">740</span><span id="line-740"> }</span> |
| <span class="source-line-no">741</span><span id="line-741"> if (connection != null) {</span> |
| <span class="source-line-no">742</span><span id="line-742"> connection.close();</span> |
| <span class="source-line-no">743</span><span id="line-743"> }</span> |
| <span class="source-line-no">744</span><span id="line-744"> }</span> |
| <span class="source-line-no">745</span><span id="line-745"> }</span> |
| <span class="source-line-no">746</span><span id="line-746"> }</span> |
| <span class="source-line-no">747</span><span id="line-747"></span> |
| <span class="source-line-no">748</span><span id="line-748"> @Override</span> |
| <span class="source-line-no">749</span><span id="line-749"> public int run(String[] args) throws Exception {</span> |
| <span class="source-line-no">750</span><span id="line-750"> if (args.length < 2) {</span> |
| <span class="source-line-no">751</span><span id="line-751"> usage("Wrong number of arguments: " + args.length);</span> |
| <span class="source-line-no">752</span><span id="line-752"> return -1;</span> |
| <span class="source-line-no">753</span><span id="line-753"> }</span> |
| <span class="source-line-no">754</span><span id="line-754"> String inputVersionString = System.getProperty(ResultSerialization.IMPORT_FORMAT_VER);</span> |
| <span class="source-line-no">755</span><span id="line-755"> if (inputVersionString != null) {</span> |
| <span class="source-line-no">756</span><span id="line-756"> getConf().set(ResultSerialization.IMPORT_FORMAT_VER, inputVersionString);</span> |
| <span class="source-line-no">757</span><span id="line-757"> }</span> |
| <span class="source-line-no">758</span><span id="line-758"> Job job = createSubmittableJob(getConf(), args);</span> |
| <span class="source-line-no">759</span><span id="line-759"> boolean isJobSuccessful = job.waitForCompletion(true);</span> |
| <span class="source-line-no">760</span><span id="line-760"> if (isJobSuccessful) {</span> |
| <span class="source-line-no">761</span><span id="line-761"> // Flush all the regions of the table</span> |
| <span class="source-line-no">762</span><span id="line-762"> flushRegionsIfNecessary(getConf());</span> |
| <span class="source-line-no">763</span><span id="line-763"> }</span> |
| <span class="source-line-no">764</span><span id="line-764"> long inputRecords = job.getCounters().findCounter(TaskCounter.MAP_INPUT_RECORDS).getValue();</span> |
| <span class="source-line-no">765</span><span id="line-765"> long outputRecords = job.getCounters().findCounter(TaskCounter.MAP_OUTPUT_RECORDS).getValue();</span> |
| <span class="source-line-no">766</span><span id="line-766"> if (outputRecords < inputRecords) {</span> |
| <span class="source-line-no">767</span><span id="line-767"> System.err.println("Warning, not all records were imported (maybe filtered out).");</span> |
| <span class="source-line-no">768</span><span id="line-768"> if (outputRecords == 0) {</span> |
| <span class="source-line-no">769</span><span id="line-769"> System.err.println("If the data was exported from HBase 0.94 "</span> |
| <span class="source-line-no">770</span><span id="line-770"> + "consider using -Dhbase.import.version=0.94.");</span> |
| <span class="source-line-no">771</span><span id="line-771"> }</span> |
| <span class="source-line-no">772</span><span id="line-772"> }</span> |
| <span class="source-line-no">773</span><span id="line-773"></span> |
| <span class="source-line-no">774</span><span id="line-774"> return (isJobSuccessful ? 0 : 1);</span> |
| <span class="source-line-no">775</span><span id="line-775"> }</span> |
| <span class="source-line-no">776</span><span id="line-776"></span> |
| <span class="source-line-no">777</span><span id="line-777"> /**</span> |
| <span class="source-line-no">778</span><span id="line-778"> * Main entry point.</span> |
| <span class="source-line-no">779</span><span id="line-779"> * @param args The command line parameters.</span> |
| <span class="source-line-no">780</span><span id="line-780"> * @throws Exception When running the job fails.</span> |
| <span class="source-line-no">781</span><span id="line-781"> */</span> |
| <span class="source-line-no">782</span><span id="line-782"> public static void main(String[] args) throws Exception {</span> |
| <span class="source-line-no">783</span><span id="line-783"> int errCode = ToolRunner.run(HBaseConfiguration.create(), new Import(), args);</span> |
| <span class="source-line-no">784</span><span id="line-784"> System.exit(errCode);</span> |
| <span class="source-line-no">785</span><span id="line-785"> }</span> |
| <span class="source-line-no">786</span><span id="line-786"></span> |
| <span class="source-line-no">787</span><span id="line-787">}</span> |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| </pre> |
| </div> |
| </main> |
| </body> |
| </html> |
