| <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd"> |
| <html lang="en"> |
| <head> |
| <title>Source code</title> |
| <link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style"> |
| </head> |
| <body> |
| <div class="sourceContainer"> |
| <pre><span class="sourceLineNo">001</span>/**<a name="line.1"></a> |
| <span class="sourceLineNo">002</span> *<a name="line.2"></a> |
| <span class="sourceLineNo">003</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.3"></a> |
| <span class="sourceLineNo">004</span> * or more contributor license agreements. See the NOTICE file<a name="line.4"></a> |
| <span class="sourceLineNo">005</span> * distributed with this work for additional information<a name="line.5"></a> |
| <span class="sourceLineNo">006</span> * regarding copyright ownership. The ASF licenses this file<a name="line.6"></a> |
| <span class="sourceLineNo">007</span> * to you under the Apache License, Version 2.0 (the<a name="line.7"></a> |
| <span class="sourceLineNo">008</span> * "License"); you may not use this file except in compliance<a name="line.8"></a> |
| <span class="sourceLineNo">009</span> * with the License. You may obtain a copy of the License at<a name="line.9"></a> |
| <span class="sourceLineNo">010</span> *<a name="line.10"></a> |
| <span class="sourceLineNo">011</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.11"></a> |
| <span class="sourceLineNo">012</span> *<a name="line.12"></a> |
| <span class="sourceLineNo">013</span> * Unless required by applicable law or agreed to in writing, software<a name="line.13"></a> |
| <span class="sourceLineNo">014</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.14"></a> |
| <span class="sourceLineNo">015</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.15"></a> |
| <span class="sourceLineNo">016</span> * See the License for the specific language governing permissions and<a name="line.16"></a> |
| <span class="sourceLineNo">017</span> * limitations under the License.<a name="line.17"></a> |
| <span class="sourceLineNo">018</span> */<a name="line.18"></a> |
| <span class="sourceLineNo">019</span>package org.apache.hadoop.hbase.util;<a name="line.19"></a> |
| <span class="sourceLineNo">020</span><a name="line.20"></a> |
| <span class="sourceLineNo">021</span>import java.io.IOException;<a name="line.21"></a> |
| <span class="sourceLineNo">022</span>import java.math.BigInteger;<a name="line.22"></a> |
| <span class="sourceLineNo">023</span>import java.util.Arrays;<a name="line.23"></a> |
| <span class="sourceLineNo">024</span>import java.util.Collection;<a name="line.24"></a> |
| <span class="sourceLineNo">025</span>import java.util.LinkedList;<a name="line.25"></a> |
| <span class="sourceLineNo">026</span>import java.util.List;<a name="line.26"></a> |
| <span class="sourceLineNo">027</span>import java.util.Map;<a name="line.27"></a> |
| <span class="sourceLineNo">028</span>import java.util.Set;<a name="line.28"></a> |
| <span class="sourceLineNo">029</span>import java.util.TreeMap;<a name="line.29"></a> |
| <span class="sourceLineNo">030</span>import org.apache.commons.lang3.ArrayUtils;<a name="line.30"></a> |
| <span class="sourceLineNo">031</span>import org.apache.commons.lang3.StringUtils;<a name="line.31"></a> |
| <span class="sourceLineNo">032</span>import org.apache.hadoop.conf.Configuration;<a name="line.32"></a> |
| <span class="sourceLineNo">033</span>import org.apache.hadoop.fs.FSDataInputStream;<a name="line.33"></a> |
| <span class="sourceLineNo">034</span>import org.apache.hadoop.fs.FSDataOutputStream;<a name="line.34"></a> |
| <span class="sourceLineNo">035</span>import org.apache.hadoop.fs.FileSystem;<a name="line.35"></a> |
| <span class="sourceLineNo">036</span>import org.apache.hadoop.fs.Path;<a name="line.36"></a> |
| <span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.37"></a> |
| <span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.HConstants;<a name="line.38"></a> |
| <span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.HRegionLocation;<a name="line.39"></a> |
| <span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.ServerName;<a name="line.40"></a> |
| <span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.TableName;<a name="line.41"></a> |
| <span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.client.Admin;<a name="line.42"></a> |
| <span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.43"></a> |
| <span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.44"></a> |
| <span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.45"></a> |
| <span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.46"></a> |
| <span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.client.NoServerForRegionException;<a name="line.47"></a> |
| <span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.client.RegionInfo;<a name="line.48"></a> |
| <span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.client.RegionLocator;<a name="line.49"></a> |
| <span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.client.Table;<a name="line.50"></a> |
| <span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.51"></a> |
| <span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.52"></a> |
| <span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.regionserver.HRegionFileSystem;<a name="line.53"></a> |
| <span class="sourceLineNo">054</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.54"></a> |
| <span class="sourceLineNo">055</span>import org.slf4j.Logger;<a name="line.55"></a> |
| <span class="sourceLineNo">056</span>import org.slf4j.LoggerFactory;<a name="line.56"></a> |
| <span class="sourceLineNo">057</span><a name="line.57"></a> |
| <span class="sourceLineNo">058</span>import org.apache.hbase.thirdparty.com.google.common.base.Preconditions;<a name="line.58"></a> |
| <span class="sourceLineNo">059</span>import org.apache.hbase.thirdparty.com.google.common.collect.Lists;<a name="line.59"></a> |
| <span class="sourceLineNo">060</span>import org.apache.hbase.thirdparty.com.google.common.collect.Maps;<a name="line.60"></a> |
| <span class="sourceLineNo">061</span>import org.apache.hbase.thirdparty.com.google.common.collect.Sets;<a name="line.61"></a> |
| <span class="sourceLineNo">062</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.62"></a> |
| <span class="sourceLineNo">063</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.GnuParser;<a name="line.63"></a> |
| <span class="sourceLineNo">064</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.HelpFormatter;<a name="line.64"></a> |
| <span class="sourceLineNo">065</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.OptionBuilder;<a name="line.65"></a> |
| <span class="sourceLineNo">066</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.Options;<a name="line.66"></a> |
| <span class="sourceLineNo">067</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.ParseException;<a name="line.67"></a> |
| <span class="sourceLineNo">068</span><a name="line.68"></a> |
| <span class="sourceLineNo">069</span>/**<a name="line.69"></a> |
| <span class="sourceLineNo">070</span> * The {@link RegionSplitter} class provides several utilities to help in the<a name="line.70"></a> |
| <span class="sourceLineNo">071</span> * administration lifecycle for developers who choose to manually split regions<a name="line.71"></a> |
| <span class="sourceLineNo">072</span> * instead of having HBase handle that automatically. The most useful utilities<a name="line.72"></a> |
| <span class="sourceLineNo">073</span> * are:<a name="line.73"></a> |
| <span class="sourceLineNo">074</span> * <p><a name="line.74"></a> |
| <span class="sourceLineNo">075</span> * <ul><a name="line.75"></a> |
| <span class="sourceLineNo">076</span> * <li>Create a table with a specified number of pre-split regions<a name="line.76"></a> |
| <span class="sourceLineNo">077</span> * <li>Execute a rolling split of all regions on an existing table<a name="line.77"></a> |
| <span class="sourceLineNo">078</span> * </ul><a name="line.78"></a> |
| <span class="sourceLineNo">079</span> * <p><a name="line.79"></a> |
| <span class="sourceLineNo">080</span> * Both operations can be safely done on a live server.<a name="line.80"></a> |
| <span class="sourceLineNo">081</span> * <p><a name="line.81"></a> |
| <span class="sourceLineNo">082</span> * <b>Question:</b> How do I turn off automatic splitting? <br><a name="line.82"></a> |
| <span class="sourceLineNo">083</span> * <b>Answer:</b> Automatic splitting is determined by the configuration value<a name="line.83"></a> |
| <span class="sourceLineNo">084</span> * <i>HConstants.HREGION_MAX_FILESIZE</i>. It is not recommended that you set this<a name="line.84"></a> |
| <span class="sourceLineNo">085</span> * to Long.MAX_VALUE in case you forget about manual splits. A suggested setting<a name="line.85"></a> |
| <span class="sourceLineNo">086</span> * is 100GB, which would result in &gt; 1hr major compactions if reached.<a name="line.86"></a> |
| <span class="sourceLineNo">087</span> * <p><a name="line.87"></a> |
| <span class="sourceLineNo">088</span> * <b>Question:</b> Why did the original authors decide to manually split? <br><a name="line.88"></a> |
| <span class="sourceLineNo">089</span> * <b>Answer:</b> Specific workload characteristics of our use case allowed us<a name="line.89"></a> |
| <span class="sourceLineNo">090</span> * to benefit from a manual split system.<a name="line.90"></a> |
| <span class="sourceLineNo">091</span> * <p><a name="line.91"></a> |
| <span class="sourceLineNo">092</span> * <ul><a name="line.92"></a> |
| <span class="sourceLineNo">093</span> * <li>Data (~1k) that would grow instead of being replaced<a name="line.93"></a> |
| <span class="sourceLineNo">094</span> * <li>Data growth was roughly uniform across all regions<a name="line.94"></a> |
| <span class="sourceLineNo">095</span> * <li>OLTP workload. Data loss is a big deal.<a name="line.95"></a> |
| <span class="sourceLineNo">096</span> * </ul><a name="line.96"></a> |
| <span class="sourceLineNo">097</span> * <p><a name="line.97"></a> |
| <span class="sourceLineNo">098</span> * <b>Question:</b> Why is manual splitting good for this workload? <br><a name="line.98"></a> |
| <span class="sourceLineNo">099</span> * <b>Answer:</b> Although automated splitting is not a bad option, there are<a name="line.99"></a> |
| <span class="sourceLineNo">100</span> * benefits to manual splitting.<a name="line.100"></a> |
| <span class="sourceLineNo">101</span> * <p><a name="line.101"></a> |
| <span class="sourceLineNo">102</span> * <ul><a name="line.102"></a> |
| <span class="sourceLineNo">103</span> * <li>With growing amounts of data, splits will continually be needed. Since<a name="line.103"></a> |
| <span class="sourceLineNo">104</span> * you always know exactly what regions you have, long-term debugging and<a name="line.104"></a> |
| <span class="sourceLineNo">105</span> * profiling is much easier with manual splits. It is hard to trace the logs to<a name="line.105"></a> |
| <span class="sourceLineNo">106</span> * understand region level problems if it keeps splitting and getting renamed.<a name="line.106"></a> |
| <span class="sourceLineNo">107</span> * <li>Data offlining bugs + unknown number of split regions == oh crap! If an<a name="line.107"></a> |
| <span class="sourceLineNo">108</span> * WAL or StoreFile was mistakenly unprocessed by HBase due to a weird bug and<a name="line.108"></a> |
| <span class="sourceLineNo">109</span> * you notice it a day or so later, you can be assured that the regions<a name="line.109"></a> |
| <span class="sourceLineNo">110</span> * specified in these files are the same as the current regions and you have<a name="line.110"></a> |
| <span class="sourceLineNo">111</span> * less headaches trying to restore/replay your data.<a name="line.111"></a> |
| <span class="sourceLineNo">112</span> * <li>You can finely tune your compaction algorithm. With roughly uniform data<a name="line.112"></a> |
| <span class="sourceLineNo">113</span> * growth, it's easy to cause split / compaction storms as the regions all<a name="line.113"></a> |
| <span class="sourceLineNo">114</span> * roughly hit the same data size at the same time. With manual splits, you can<a name="line.114"></a> |
| <span class="sourceLineNo">115</span> * let staggered, time-based major compactions spread out your network IO load.<a name="line.115"></a> |
| <span class="sourceLineNo">116</span> * </ul><a name="line.116"></a> |
| <span class="sourceLineNo">117</span> * <p><a name="line.117"></a> |
| <span class="sourceLineNo">118</span> * <b>Question:</b> What's the optimal number of pre-split regions to create? <br><a name="line.118"></a> |
| <span class="sourceLineNo">119</span> * <b>Answer:</b> Mileage will vary depending upon your application.<a name="line.119"></a> |
| <span class="sourceLineNo">120</span> * <p><a name="line.120"></a> |
| <span class="sourceLineNo">121</span> * The short answer for our application is that we started with 10 pre-split<a name="line.121"></a> |
| <span class="sourceLineNo">122</span> * regions / server and watched our data growth over time. It's better to err on<a name="line.122"></a> |
| <span class="sourceLineNo">123</span> * the side of too little regions and rolling split later.<a name="line.123"></a> |
| <span class="sourceLineNo">124</span> * <p><a name="line.124"></a> |
| <span class="sourceLineNo">125</span> * The more complicated answer is that this depends upon the largest storefile<a name="line.125"></a> |
| <span class="sourceLineNo">126</span> * in your region. With a growing data size, this will get larger over time. You<a name="line.126"></a> |
| <span class="sourceLineNo">127</span> * want the largest region to be just big enough that the<a name="line.127"></a> |
| <span class="sourceLineNo">128</span> * {@link org.apache.hadoop.hbase.regionserver.HStore} compact<a name="line.128"></a> |
| <span class="sourceLineNo">129</span> * selection algorithm only compacts it due to a timed major. If you don't, your<a name="line.129"></a> |
| <span class="sourceLineNo">130</span> * cluster can be prone to compaction storms as the algorithm decides to run<a name="line.130"></a> |
| <span class="sourceLineNo">131</span> * major compactions on a large series of regions all at once. Note that<a name="line.131"></a> |
| <span class="sourceLineNo">132</span> * compaction storms are due to the uniform data growth, not the manual split<a name="line.132"></a> |
| <span class="sourceLineNo">133</span> * decision.<a name="line.133"></a> |
| <span class="sourceLineNo">134</span> * <p><a name="line.134"></a> |
| <span class="sourceLineNo">135</span> * If you pre-split your regions too thin, you can increase the major compaction<a name="line.135"></a> |
| <span class="sourceLineNo">136</span> * interval by configuring HConstants.MAJOR_COMPACTION_PERIOD. If your data size<a name="line.136"></a> |
| <span class="sourceLineNo">137</span> * grows too large, use this script to perform a network IO safe rolling split<a name="line.137"></a> |
| <span class="sourceLineNo">138</span> * of all regions.<a name="line.138"></a> |
| <span class="sourceLineNo">139</span> */<a name="line.139"></a> |
| <span class="sourceLineNo">140</span>@InterfaceAudience.Private<a name="line.140"></a> |
| <span class="sourceLineNo">141</span>public class RegionSplitter {<a name="line.141"></a> |
| <span class="sourceLineNo">142</span> private static final Logger LOG = LoggerFactory.getLogger(RegionSplitter.class);<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> * A generic interface for the RegionSplitter code to use for all it's<a name="line.145"></a> |
| <span class="sourceLineNo">146</span> * functionality. Note that the original authors of this code use<a name="line.146"></a> |
| <span class="sourceLineNo">147</span> * {@link HexStringSplit} to partition their table and set it as default, but<a name="line.147"></a> |
| <span class="sourceLineNo">148</span> * provided this for your custom algorithm. To use, create a new derived class<a name="line.148"></a> |
| <span class="sourceLineNo">149</span> * from this interface and call {@link RegionSplitter#createPresplitTable} or<a name="line.149"></a> |
| <span class="sourceLineNo">150</span> * RegionSplitter#rollingSplit(TableName, SplitAlgorithm, Configuration) with the<a name="line.150"></a> |
| <span class="sourceLineNo">151</span> * argument splitClassName giving the name of your class.<a name="line.151"></a> |
| <span class="sourceLineNo">152</span> */<a name="line.152"></a> |
| <span class="sourceLineNo">153</span> public interface SplitAlgorithm {<a name="line.153"></a> |
| <span class="sourceLineNo">154</span> /**<a name="line.154"></a> |
| <span class="sourceLineNo">155</span> * Split a pre-existing region into 2 regions.<a name="line.155"></a> |
| <span class="sourceLineNo">156</span> *<a name="line.156"></a> |
| <span class="sourceLineNo">157</span> * @param start<a name="line.157"></a> |
| <span class="sourceLineNo">158</span> * first row (inclusive)<a name="line.158"></a> |
| <span class="sourceLineNo">159</span> * @param end<a name="line.159"></a> |
| <span class="sourceLineNo">160</span> * last row (exclusive)<a name="line.160"></a> |
| <span class="sourceLineNo">161</span> * @return the split row to use<a name="line.161"></a> |
| <span class="sourceLineNo">162</span> */<a name="line.162"></a> |
| <span class="sourceLineNo">163</span> byte[] split(byte[] start, byte[] end);<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> * Split an entire table.<a name="line.166"></a> |
| <span class="sourceLineNo">167</span> *<a name="line.167"></a> |
| <span class="sourceLineNo">168</span> * @param numRegions<a name="line.168"></a> |
| <span class="sourceLineNo">169</span> * number of regions to split the table into<a name="line.169"></a> |
| <span class="sourceLineNo">170</span> *<a name="line.170"></a> |
| <span class="sourceLineNo">171</span> * @throws RuntimeException<a name="line.171"></a> |
| <span class="sourceLineNo">172</span> * user input is validated at this time. may throw a runtime<a name="line.172"></a> |
| <span class="sourceLineNo">173</span> * exception in response to a parse failure<a name="line.173"></a> |
| <span class="sourceLineNo">174</span> * @return array of split keys for the initial regions of the table. The<a name="line.174"></a> |
| <span class="sourceLineNo">175</span> * length of the returned array should be numRegions-1.<a name="line.175"></a> |
| <span class="sourceLineNo">176</span> */<a name="line.176"></a> |
| <span class="sourceLineNo">177</span> byte[][] split(int numRegions);<a name="line.177"></a> |
| <span class="sourceLineNo">178</span><a name="line.178"></a> |
| <span class="sourceLineNo">179</span> /**<a name="line.179"></a> |
| <span class="sourceLineNo">180</span> * Some MapReduce jobs may want to run multiple mappers per region,<a name="line.180"></a> |
| <span class="sourceLineNo">181</span> * this is intended for such usecase.<a name="line.181"></a> |
| <span class="sourceLineNo">182</span> *<a name="line.182"></a> |
| <span class="sourceLineNo">183</span> * @param start first row (inclusive)<a name="line.183"></a> |
| <span class="sourceLineNo">184</span> * @param end last row (exclusive)<a name="line.184"></a> |
| <span class="sourceLineNo">185</span> * @param numSplits number of splits to generate<a name="line.185"></a> |
| <span class="sourceLineNo">186</span> * @param inclusive whether start and end are returned as split points<a name="line.186"></a> |
| <span class="sourceLineNo">187</span> */<a name="line.187"></a> |
| <span class="sourceLineNo">188</span> byte[][] split(byte[] start, byte[] end, int numSplits, boolean inclusive);<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> * In HBase, the first row is represented by an empty byte array. This might<a name="line.191"></a> |
| <span class="sourceLineNo">192</span> * cause problems with your split algorithm or row printing. All your APIs<a name="line.192"></a> |
| <span class="sourceLineNo">193</span> * will be passed firstRow() instead of empty array.<a name="line.193"></a> |
| <span class="sourceLineNo">194</span> *<a name="line.194"></a> |
| <span class="sourceLineNo">195</span> * @return your representation of your first row<a name="line.195"></a> |
| <span class="sourceLineNo">196</span> */<a name="line.196"></a> |
| <span class="sourceLineNo">197</span> byte[] firstRow();<a name="line.197"></a> |
| <span class="sourceLineNo">198</span><a name="line.198"></a> |
| <span class="sourceLineNo">199</span> /**<a name="line.199"></a> |
| <span class="sourceLineNo">200</span> * In HBase, the last row is represented by an empty byte array. This might<a name="line.200"></a> |
| <span class="sourceLineNo">201</span> * cause problems with your split algorithm or row printing. All your APIs<a name="line.201"></a> |
| <span class="sourceLineNo">202</span> * will be passed firstRow() instead of empty array.<a name="line.202"></a> |
| <span class="sourceLineNo">203</span> *<a name="line.203"></a> |
| <span class="sourceLineNo">204</span> * @return your representation of your last row<a name="line.204"></a> |
| <span class="sourceLineNo">205</span> */<a name="line.205"></a> |
| <span class="sourceLineNo">206</span> byte[] lastRow();<a name="line.206"></a> |
| <span class="sourceLineNo">207</span><a name="line.207"></a> |
| <span class="sourceLineNo">208</span> /**<a name="line.208"></a> |
| <span class="sourceLineNo">209</span> * In HBase, the last row is represented by an empty byte array. Set this<a name="line.209"></a> |
| <span class="sourceLineNo">210</span> * value to help the split code understand how to evenly divide the first<a name="line.210"></a> |
| <span class="sourceLineNo">211</span> * region.<a name="line.211"></a> |
| <span class="sourceLineNo">212</span> *<a name="line.212"></a> |
| <span class="sourceLineNo">213</span> * @param userInput<a name="line.213"></a> |
| <span class="sourceLineNo">214</span> * raw user input (may throw RuntimeException on parse failure)<a name="line.214"></a> |
| <span class="sourceLineNo">215</span> */<a name="line.215"></a> |
| <span class="sourceLineNo">216</span> void setFirstRow(String userInput);<a name="line.216"></a> |
| <span class="sourceLineNo">217</span><a name="line.217"></a> |
| <span class="sourceLineNo">218</span> /**<a name="line.218"></a> |
| <span class="sourceLineNo">219</span> * In HBase, the last row is represented by an empty byte array. Set this<a name="line.219"></a> |
| <span class="sourceLineNo">220</span> * value to help the split code understand how to evenly divide the last<a name="line.220"></a> |
| <span class="sourceLineNo">221</span> * region. Note that this last row is inclusive for all rows sharing the<a name="line.221"></a> |
| <span class="sourceLineNo">222</span> * same prefix.<a name="line.222"></a> |
| <span class="sourceLineNo">223</span> *<a name="line.223"></a> |
| <span class="sourceLineNo">224</span> * @param userInput<a name="line.224"></a> |
| <span class="sourceLineNo">225</span> * raw user input (may throw RuntimeException on parse failure)<a name="line.225"></a> |
| <span class="sourceLineNo">226</span> */<a name="line.226"></a> |
| <span class="sourceLineNo">227</span> void setLastRow(String userInput);<a name="line.227"></a> |
| <span class="sourceLineNo">228</span><a name="line.228"></a> |
| <span class="sourceLineNo">229</span> /**<a name="line.229"></a> |
| <span class="sourceLineNo">230</span> * @param input<a name="line.230"></a> |
| <span class="sourceLineNo">231</span> * user or file input for row<a name="line.231"></a> |
| <span class="sourceLineNo">232</span> * @return byte array representation of this row for HBase<a name="line.232"></a> |
| <span class="sourceLineNo">233</span> */<a name="line.233"></a> |
| <span class="sourceLineNo">234</span> byte[] strToRow(String input);<a name="line.234"></a> |
| <span class="sourceLineNo">235</span><a name="line.235"></a> |
| <span class="sourceLineNo">236</span> /**<a name="line.236"></a> |
| <span class="sourceLineNo">237</span> * @param row<a name="line.237"></a> |
| <span class="sourceLineNo">238</span> * byte array representing a row in HBase<a name="line.238"></a> |
| <span class="sourceLineNo">239</span> * @return String to use for debug &amp; file printing<a name="line.239"></a> |
| <span class="sourceLineNo">240</span> */<a name="line.240"></a> |
| <span class="sourceLineNo">241</span> String rowToStr(byte[] row);<a name="line.241"></a> |
| <span class="sourceLineNo">242</span><a name="line.242"></a> |
| <span class="sourceLineNo">243</span> /**<a name="line.243"></a> |
| <span class="sourceLineNo">244</span> * @return the separator character to use when storing / printing the row<a name="line.244"></a> |
| <span class="sourceLineNo">245</span> */<a name="line.245"></a> |
| <span class="sourceLineNo">246</span> String separator();<a name="line.246"></a> |
| <span class="sourceLineNo">247</span><a name="line.247"></a> |
| <span class="sourceLineNo">248</span> /**<a name="line.248"></a> |
| <span class="sourceLineNo">249</span> * Set the first row<a name="line.249"></a> |
| <span class="sourceLineNo">250</span> * @param userInput byte array of the row key.<a name="line.250"></a> |
| <span class="sourceLineNo">251</span> */<a name="line.251"></a> |
| <span class="sourceLineNo">252</span> void setFirstRow(byte[] userInput);<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> * Set the last row<a name="line.255"></a> |
| <span class="sourceLineNo">256</span> * @param userInput byte array of the row key.<a name="line.256"></a> |
| <span class="sourceLineNo">257</span> */<a name="line.257"></a> |
| <span class="sourceLineNo">258</span> void setLastRow(byte[] userInput);<a name="line.258"></a> |
| <span class="sourceLineNo">259</span> }<a name="line.259"></a> |
| <span class="sourceLineNo">260</span><a name="line.260"></a> |
| <span class="sourceLineNo">261</span> /**<a name="line.261"></a> |
| <span class="sourceLineNo">262</span> * The main function for the RegionSplitter application. Common uses:<a name="line.262"></a> |
| <span class="sourceLineNo">263</span> * <p><a name="line.263"></a> |
| <span class="sourceLineNo">264</span> * <ul><a name="line.264"></a> |
| <span class="sourceLineNo">265</span> * <li>create a table named 'myTable' with 60 pre-split regions containing 2<a name="line.265"></a> |
| <span class="sourceLineNo">266</span> * column families 'test' &amp; 'rs', assuming the keys are hex-encoded ASCII:<a name="line.266"></a> |
| <span class="sourceLineNo">267</span> * <ul><a name="line.267"></a> |
| <span class="sourceLineNo">268</span> * <li>bin/hbase org.apache.hadoop.hbase.util.RegionSplitter -c 60 -f test:rs<a name="line.268"></a> |
| <span class="sourceLineNo">269</span> * myTable HexStringSplit<a name="line.269"></a> |
| <span class="sourceLineNo">270</span> * </ul><a name="line.270"></a> |
| <span class="sourceLineNo">271</span> * <li>create a table named 'myTable' with 50 pre-split regions,<a name="line.271"></a> |
| <span class="sourceLineNo">272</span> * assuming the keys are decimal-encoded ASCII:<a name="line.272"></a> |
| <span class="sourceLineNo">273</span> * <ul><a name="line.273"></a> |
| <span class="sourceLineNo">274</span> * <li>bin/hbase org.apache.hadoop.hbase.util.RegionSplitter -c 50<a name="line.274"></a> |
| <span class="sourceLineNo">275</span> * myTable DecimalStringSplit<a name="line.275"></a> |
| <span class="sourceLineNo">276</span> * </ul><a name="line.276"></a> |
| <span class="sourceLineNo">277</span> * <li>perform a rolling split of 'myTable' (i.e. 60 =&gt; 120 regions), # 2<a name="line.277"></a> |
| <span class="sourceLineNo">278</span> * outstanding splits at a time, assuming keys are uniformly distributed<a name="line.278"></a> |
| <span class="sourceLineNo">279</span> * bytes:<a name="line.279"></a> |
| <span class="sourceLineNo">280</span> * <ul><a name="line.280"></a> |
| <span class="sourceLineNo">281</span> * <li>bin/hbase org.apache.hadoop.hbase.util.RegionSplitter -r -o 2 myTable<a name="line.281"></a> |
| <span class="sourceLineNo">282</span> * UniformSplit<a name="line.282"></a> |
| <span class="sourceLineNo">283</span> * </ul><a name="line.283"></a> |
| <span class="sourceLineNo">284</span> * </ul><a name="line.284"></a> |
| <span class="sourceLineNo">285</span> *<a name="line.285"></a> |
| <span class="sourceLineNo">286</span> * There are three SplitAlgorithms built into RegionSplitter, HexStringSplit,<a name="line.286"></a> |
| <span class="sourceLineNo">287</span> * DecimalStringSplit, and UniformSplit. These are different strategies for<a name="line.287"></a> |
| <span class="sourceLineNo">288</span> * choosing region boundaries. See their source code for details.<a name="line.288"></a> |
| <span class="sourceLineNo">289</span> *<a name="line.289"></a> |
| <span class="sourceLineNo">290</span> * @param args<a name="line.290"></a> |
| <span class="sourceLineNo">291</span> * Usage: RegionSplitter &lt;TABLE&gt; &lt;SPLITALGORITHM&gt;<a name="line.291"></a> |
| <span class="sourceLineNo">292</span> * &lt;-c &lt;# regions&gt; -f &lt;family:family:...&gt; | -r<a name="line.292"></a> |
| <span class="sourceLineNo">293</span> * [-o &lt;# outstanding splits&gt;]&gt;<a name="line.293"></a> |
| <span class="sourceLineNo">294</span> * [-D &lt;conf.param=value&gt;]<a name="line.294"></a> |
| <span class="sourceLineNo">295</span> * @throws IOException<a name="line.295"></a> |
| <span class="sourceLineNo">296</span> * HBase IO problem<a name="line.296"></a> |
| <span class="sourceLineNo">297</span> * @throws InterruptedException<a name="line.297"></a> |
| <span class="sourceLineNo">298</span> * user requested exit<a name="line.298"></a> |
| <span class="sourceLineNo">299</span> * @throws ParseException<a name="line.299"></a> |
| <span class="sourceLineNo">300</span> * problem parsing user input<a name="line.300"></a> |
| <span class="sourceLineNo">301</span> */<a name="line.301"></a> |
| <span class="sourceLineNo">302</span> @SuppressWarnings("static-access")<a name="line.302"></a> |
| <span class="sourceLineNo">303</span> public static void main(String[] args) throws IOException,<a name="line.303"></a> |
| <span class="sourceLineNo">304</span> InterruptedException, ParseException {<a name="line.304"></a> |
| <span class="sourceLineNo">305</span> Configuration conf = HBaseConfiguration.create();<a name="line.305"></a> |
| <span class="sourceLineNo">306</span><a name="line.306"></a> |
| <span class="sourceLineNo">307</span> // parse user input<a name="line.307"></a> |
| <span class="sourceLineNo">308</span> Options opt = new Options();<a name="line.308"></a> |
| <span class="sourceLineNo">309</span> opt.addOption(OptionBuilder.withArgName("property=value").hasArg()<a name="line.309"></a> |
| <span class="sourceLineNo">310</span> .withDescription("Override HBase Configuration Settings").create("D"));<a name="line.310"></a> |
| <span class="sourceLineNo">311</span> opt.addOption(OptionBuilder.withArgName("region count").hasArg()<a name="line.311"></a> |
| <span class="sourceLineNo">312</span> .withDescription(<a name="line.312"></a> |
| <span class="sourceLineNo">313</span> "Create a new table with a pre-split number of regions")<a name="line.313"></a> |
| <span class="sourceLineNo">314</span> .create("c"));<a name="line.314"></a> |
| <span class="sourceLineNo">315</span> opt.addOption(OptionBuilder.withArgName("family:family:...").hasArg()<a name="line.315"></a> |
| <span class="sourceLineNo">316</span> .withDescription(<a name="line.316"></a> |
| <span class="sourceLineNo">317</span> "Column Families to create with new table. Required with -c")<a name="line.317"></a> |
| <span class="sourceLineNo">318</span> .create("f"));<a name="line.318"></a> |
| <span class="sourceLineNo">319</span> opt.addOption("h", false, "Print this usage help");<a name="line.319"></a> |
| <span class="sourceLineNo">320</span> opt.addOption("r", false, "Perform a rolling split of an existing region");<a name="line.320"></a> |
| <span class="sourceLineNo">321</span> opt.addOption(OptionBuilder.withArgName("count").hasArg().withDescription(<a name="line.321"></a> |
| <span class="sourceLineNo">322</span> "Max outstanding splits that have unfinished major compactions")<a name="line.322"></a> |
| <span class="sourceLineNo">323</span> .create("o"));<a name="line.323"></a> |
| <span class="sourceLineNo">324</span> opt.addOption(null, "firstrow", true,<a name="line.324"></a> |
| <span class="sourceLineNo">325</span> "First Row in Table for Split Algorithm");<a name="line.325"></a> |
| <span class="sourceLineNo">326</span> opt.addOption(null, "lastrow", true,<a name="line.326"></a> |
| <span class="sourceLineNo">327</span> "Last Row in Table for Split Algorithm");<a name="line.327"></a> |
| <span class="sourceLineNo">328</span> opt.addOption(null, "risky", false,<a name="line.328"></a> |
| <span class="sourceLineNo">329</span> "Skip verification steps to complete quickly. "<a name="line.329"></a> |
| <span class="sourceLineNo">330</span> + "STRONGLY DISCOURAGED for production systems. ");<a name="line.330"></a> |
| <span class="sourceLineNo">331</span> CommandLine cmd = new GnuParser().parse(opt, args);<a name="line.331"></a> |
| <span class="sourceLineNo">332</span><a name="line.332"></a> |
| <span class="sourceLineNo">333</span> if (cmd.hasOption("D")) {<a name="line.333"></a> |
| <span class="sourceLineNo">334</span> for (String confOpt : cmd.getOptionValues("D")) {<a name="line.334"></a> |
| <span class="sourceLineNo">335</span> String[] kv = confOpt.split("=", 2);<a name="line.335"></a> |
| <span class="sourceLineNo">336</span> if (kv.length == 2) {<a name="line.336"></a> |
| <span class="sourceLineNo">337</span> conf.set(kv[0], kv[1]);<a name="line.337"></a> |
| <span class="sourceLineNo">338</span> LOG.debug("-D configuration override: " + kv[0] + "=" + kv[1]);<a name="line.338"></a> |
| <span class="sourceLineNo">339</span> } else {<a name="line.339"></a> |
| <span class="sourceLineNo">340</span> throw new ParseException("-D option format invalid: " + confOpt);<a name="line.340"></a> |
| <span class="sourceLineNo">341</span> }<a name="line.341"></a> |
| <span class="sourceLineNo">342</span> }<a name="line.342"></a> |
| <span class="sourceLineNo">343</span> }<a name="line.343"></a> |
| <span class="sourceLineNo">344</span><a name="line.344"></a> |
| <span class="sourceLineNo">345</span> if (cmd.hasOption("risky")) {<a name="line.345"></a> |
| <span class="sourceLineNo">346</span> conf.setBoolean("split.verify", false);<a name="line.346"></a> |
| <span class="sourceLineNo">347</span> }<a name="line.347"></a> |
| <span class="sourceLineNo">348</span><a name="line.348"></a> |
| <span class="sourceLineNo">349</span> boolean createTable = cmd.hasOption("c") && cmd.hasOption("f");<a name="line.349"></a> |
| <span class="sourceLineNo">350</span> boolean rollingSplit = cmd.hasOption("r");<a name="line.350"></a> |
| <span class="sourceLineNo">351</span> boolean oneOperOnly = createTable ^ rollingSplit;<a name="line.351"></a> |
| <span class="sourceLineNo">352</span><a name="line.352"></a> |
| <span class="sourceLineNo">353</span> if (2 != cmd.getArgList().size() || !oneOperOnly || cmd.hasOption("h")) {<a name="line.353"></a> |
| <span class="sourceLineNo">354</span> new HelpFormatter().printHelp("bin/hbase regionsplitter <TABLE> <SPLITALGORITHM>\n"+<a name="line.354"></a> |
| <span class="sourceLineNo">355</span> "SPLITALGORITHM is the java class name of a class implementing " +<a name="line.355"></a> |
| <span class="sourceLineNo">356</span> "SplitAlgorithm, or one of the special strings HexStringSplit or " +<a name="line.356"></a> |
| <span class="sourceLineNo">357</span> "DecimalStringSplit or UniformSplit, which are built-in split algorithms. " +<a name="line.357"></a> |
| <span class="sourceLineNo">358</span> "HexStringSplit treats keys as hexadecimal ASCII, and " +<a name="line.358"></a> |
| <span class="sourceLineNo">359</span> "DecimalStringSplit treats keys as decimal ASCII, and " +<a name="line.359"></a> |
| <span class="sourceLineNo">360</span> "UniformSplit treats keys as arbitrary bytes.", opt);<a name="line.360"></a> |
| <span class="sourceLineNo">361</span> return;<a name="line.361"></a> |
| <span class="sourceLineNo">362</span> }<a name="line.362"></a> |
| <span class="sourceLineNo">363</span> TableName tableName = TableName.valueOf(cmd.getArgs()[0]);<a name="line.363"></a> |
| <span class="sourceLineNo">364</span> String splitClass = cmd.getArgs()[1];<a name="line.364"></a> |
| <span class="sourceLineNo">365</span> SplitAlgorithm splitAlgo = newSplitAlgoInstance(conf, splitClass);<a name="line.365"></a> |
| <span class="sourceLineNo">366</span><a name="line.366"></a> |
| <span class="sourceLineNo">367</span> if (cmd.hasOption("firstrow")) {<a name="line.367"></a> |
| <span class="sourceLineNo">368</span> splitAlgo.setFirstRow(cmd.getOptionValue("firstrow"));<a name="line.368"></a> |
| <span class="sourceLineNo">369</span> }<a name="line.369"></a> |
| <span class="sourceLineNo">370</span> if (cmd.hasOption("lastrow")) {<a name="line.370"></a> |
| <span class="sourceLineNo">371</span> splitAlgo.setLastRow(cmd.getOptionValue("lastrow"));<a name="line.371"></a> |
| <span class="sourceLineNo">372</span> }<a name="line.372"></a> |
| <span class="sourceLineNo">373</span><a name="line.373"></a> |
| <span class="sourceLineNo">374</span> if (createTable) {<a name="line.374"></a> |
| <span class="sourceLineNo">375</span> conf.set("split.count", cmd.getOptionValue("c"));<a name="line.375"></a> |
| <span class="sourceLineNo">376</span> createPresplitTable(tableName, splitAlgo, cmd.getOptionValue("f").split(":"), conf);<a name="line.376"></a> |
| <span class="sourceLineNo">377</span> }<a name="line.377"></a> |
| <span class="sourceLineNo">378</span><a name="line.378"></a> |
| <span class="sourceLineNo">379</span> if (rollingSplit) {<a name="line.379"></a> |
| <span class="sourceLineNo">380</span> if (cmd.hasOption("o")) {<a name="line.380"></a> |
| <span class="sourceLineNo">381</span> conf.set("split.outstanding", cmd.getOptionValue("o"));<a name="line.381"></a> |
| <span class="sourceLineNo">382</span> }<a name="line.382"></a> |
| <span class="sourceLineNo">383</span> rollingSplit(tableName, splitAlgo, conf);<a name="line.383"></a> |
| <span class="sourceLineNo">384</span> }<a name="line.384"></a> |
| <span class="sourceLineNo">385</span> }<a name="line.385"></a> |
| <span class="sourceLineNo">386</span><a name="line.386"></a> |
| <span class="sourceLineNo">387</span> static void createPresplitTable(TableName tableName, SplitAlgorithm splitAlgo,<a name="line.387"></a> |
| <span class="sourceLineNo">388</span> String[] columnFamilies, Configuration conf)<a name="line.388"></a> |
| <span class="sourceLineNo">389</span> throws IOException, InterruptedException {<a name="line.389"></a> |
| <span class="sourceLineNo">390</span> final int splitCount = conf.getInt("split.count", 0);<a name="line.390"></a> |
| <span class="sourceLineNo">391</span> Preconditions.checkArgument(splitCount > 1, "Split count must be > 1");<a name="line.391"></a> |
| <span class="sourceLineNo">392</span><a name="line.392"></a> |
| <span class="sourceLineNo">393</span> Preconditions.checkArgument(columnFamilies.length > 0,<a name="line.393"></a> |
| <span class="sourceLineNo">394</span> "Must specify at least one column family. ");<a name="line.394"></a> |
| <span class="sourceLineNo">395</span> LOG.debug("Creating table " + tableName + " with " + columnFamilies.length<a name="line.395"></a> |
| <span class="sourceLineNo">396</span> + " column families. Presplitting to " + splitCount + " regions");<a name="line.396"></a> |
| <span class="sourceLineNo">397</span><a name="line.397"></a> |
| <span class="sourceLineNo">398</span> TableDescriptorBuilder builder = TableDescriptorBuilder.newBuilder(tableName);<a name="line.398"></a> |
| <span class="sourceLineNo">399</span> for (String cf : columnFamilies) {<a name="line.399"></a> |
| <span class="sourceLineNo">400</span> builder.setColumnFamily(ColumnFamilyDescriptorBuilder.of(cf));<a name="line.400"></a> |
| <span class="sourceLineNo">401</span> }<a name="line.401"></a> |
| <span class="sourceLineNo">402</span> try (Connection connection = ConnectionFactory.createConnection(conf)) {<a name="line.402"></a> |
| <span class="sourceLineNo">403</span> Admin admin = connection.getAdmin();<a name="line.403"></a> |
| <span class="sourceLineNo">404</span> try {<a name="line.404"></a> |
| <span class="sourceLineNo">405</span> Preconditions.checkArgument(!admin.tableExists(tableName),<a name="line.405"></a> |
| <span class="sourceLineNo">406</span> "Table already exists: " + tableName);<a name="line.406"></a> |
| <span class="sourceLineNo">407</span> admin.createTable(builder.build(), splitAlgo.split(splitCount));<a name="line.407"></a> |
| <span class="sourceLineNo">408</span> } finally {<a name="line.408"></a> |
| <span class="sourceLineNo">409</span> admin.close();<a name="line.409"></a> |
| <span class="sourceLineNo">410</span> }<a name="line.410"></a> |
| <span class="sourceLineNo">411</span> LOG.debug("Table created! Waiting for regions to show online in META...");<a name="line.411"></a> |
| <span class="sourceLineNo">412</span> if (!conf.getBoolean("split.verify", true)) {<a name="line.412"></a> |
| <span class="sourceLineNo">413</span> // NOTE: createTable is synchronous on the table, but not on the regions<a name="line.413"></a> |
| <span class="sourceLineNo">414</span> int onlineRegions = 0;<a name="line.414"></a> |
| <span class="sourceLineNo">415</span> try (RegionLocator locator = connection.getRegionLocator(tableName)) {<a name="line.415"></a> |
| <span class="sourceLineNo">416</span> while (onlineRegions < splitCount) {<a name="line.416"></a> |
| <span class="sourceLineNo">417</span> onlineRegions = locator.getAllRegionLocations().size();<a name="line.417"></a> |
| <span class="sourceLineNo">418</span> LOG.debug(onlineRegions + " of " + splitCount + " regions online...");<a name="line.418"></a> |
| <span class="sourceLineNo">419</span> if (onlineRegions < splitCount) {<a name="line.419"></a> |
| <span class="sourceLineNo">420</span> Thread.sleep(10 * 1000); // sleep<a name="line.420"></a> |
| <span class="sourceLineNo">421</span> }<a name="line.421"></a> |
| <span class="sourceLineNo">422</span> }<a name="line.422"></a> |
| <span class="sourceLineNo">423</span> }<a name="line.423"></a> |
| <span class="sourceLineNo">424</span> }<a name="line.424"></a> |
| <span class="sourceLineNo">425</span> LOG.debug("Finished creating table with " + splitCount + " regions");<a name="line.425"></a> |
| <span class="sourceLineNo">426</span> }<a name="line.426"></a> |
| <span class="sourceLineNo">427</span> }<a name="line.427"></a> |
| <span class="sourceLineNo">428</span><a name="line.428"></a> |
| <span class="sourceLineNo">429</span> /**<a name="line.429"></a> |
| <span class="sourceLineNo">430</span> * Alternative getCurrentNrHRS which is no longer available.<a name="line.430"></a> |
| <span class="sourceLineNo">431</span> * @param connection<a name="line.431"></a> |
| <span class="sourceLineNo">432</span> * @return Rough count of regionservers out on cluster.<a name="line.432"></a> |
| <span class="sourceLineNo">433</span> * @throws IOException if a remote or network exception occurs<a name="line.433"></a> |
| <span class="sourceLineNo">434</span> */<a name="line.434"></a> |
| <span class="sourceLineNo">435</span> private static int getRegionServerCount(final Connection connection) throws IOException {<a name="line.435"></a> |
| <span class="sourceLineNo">436</span> try (Admin admin = connection.getAdmin()) {<a name="line.436"></a> |
| <span class="sourceLineNo">437</span> Collection<ServerName> servers = admin.getRegionServers();<a name="line.437"></a> |
| <span class="sourceLineNo">438</span> return servers == null || servers.isEmpty()? 0: servers.size();<a name="line.438"></a> |
| <span class="sourceLineNo">439</span> }<a name="line.439"></a> |
| <span class="sourceLineNo">440</span> }<a name="line.440"></a> |
| <span class="sourceLineNo">441</span><a name="line.441"></a> |
| <span class="sourceLineNo">442</span> private static byte [] readFile(final FileSystem fs, final Path path) throws IOException {<a name="line.442"></a> |
| <span class="sourceLineNo">443</span> FSDataInputStream tmpIn = fs.open(path);<a name="line.443"></a> |
| <span class="sourceLineNo">444</span> try {<a name="line.444"></a> |
| <span class="sourceLineNo">445</span> byte [] rawData = new byte[tmpIn.available()];<a name="line.445"></a> |
| <span class="sourceLineNo">446</span> tmpIn.readFully(rawData);<a name="line.446"></a> |
| <span class="sourceLineNo">447</span> return rawData;<a name="line.447"></a> |
| <span class="sourceLineNo">448</span> } finally {<a name="line.448"></a> |
| <span class="sourceLineNo">449</span> tmpIn.close();<a name="line.449"></a> |
| <span class="sourceLineNo">450</span> }<a name="line.450"></a> |
| <span class="sourceLineNo">451</span> }<a name="line.451"></a> |
| <span class="sourceLineNo">452</span><a name="line.452"></a> |
| <span class="sourceLineNo">453</span> static void rollingSplit(TableName tableName, SplitAlgorithm splitAlgo, Configuration conf)<a name="line.453"></a> |
| <span class="sourceLineNo">454</span> throws IOException, InterruptedException {<a name="line.454"></a> |
| <span class="sourceLineNo">455</span> final int minOS = conf.getInt("split.outstanding", 2);<a name="line.455"></a> |
| <span class="sourceLineNo">456</span> try (Connection connection = ConnectionFactory.createConnection(conf)) {<a name="line.456"></a> |
| <span class="sourceLineNo">457</span> // Max outstanding splits. default == 50% of servers<a name="line.457"></a> |
| <span class="sourceLineNo">458</span> final int MAX_OUTSTANDING = Math.max(getRegionServerCount(connection) / 2, minOS);<a name="line.458"></a> |
| <span class="sourceLineNo">459</span><a name="line.459"></a> |
| <span class="sourceLineNo">460</span> Path hbDir = CommonFSUtils.getRootDir(conf);<a name="line.460"></a> |
| <span class="sourceLineNo">461</span> Path tableDir = CommonFSUtils.getTableDir(hbDir, tableName);<a name="line.461"></a> |
| <span class="sourceLineNo">462</span> Path splitFile = new Path(tableDir, "_balancedSplit");<a name="line.462"></a> |
| <span class="sourceLineNo">463</span> FileSystem fs = FileSystem.get(conf);<a name="line.463"></a> |
| <span class="sourceLineNo">464</span><a name="line.464"></a> |
| <span class="sourceLineNo">465</span> // Get a list of daughter regions to create<a name="line.465"></a> |
| <span class="sourceLineNo">466</span> LinkedList<Pair<byte[], byte[]>> tmpRegionSet = null;<a name="line.466"></a> |
| <span class="sourceLineNo">467</span> try (Table table = connection.getTable(tableName)) {<a name="line.467"></a> |
| <span class="sourceLineNo">468</span> tmpRegionSet = getSplits(connection, tableName, splitAlgo);<a name="line.468"></a> |
| <span class="sourceLineNo">469</span> }<a name="line.469"></a> |
| <span class="sourceLineNo">470</span> LinkedList<Pair<byte[], byte[]>> outstanding = Lists.newLinkedList();<a name="line.470"></a> |
| <span class="sourceLineNo">471</span> int splitCount = 0;<a name="line.471"></a> |
| <span class="sourceLineNo">472</span> final int origCount = tmpRegionSet.size();<a name="line.472"></a> |
| <span class="sourceLineNo">473</span><a name="line.473"></a> |
| <span class="sourceLineNo">474</span> // all splits must compact & we have 1 compact thread, so 2 split<a name="line.474"></a> |
| <span class="sourceLineNo">475</span> // requests to the same RS can stall the outstanding split queue.<a name="line.475"></a> |
| <span class="sourceLineNo">476</span> // To fix, group the regions into an RS pool and round-robin through it<a name="line.476"></a> |
| <span class="sourceLineNo">477</span> LOG.debug("Bucketing regions by regionserver...");<a name="line.477"></a> |
| <span class="sourceLineNo">478</span> TreeMap<ServerName, LinkedList<Pair<byte[], byte[]>>> daughterRegions =<a name="line.478"></a> |
| <span class="sourceLineNo">479</span> Maps.newTreeMap();<a name="line.479"></a> |
| <span class="sourceLineNo">480</span> // Get a regionLocator. Need it in below.<a name="line.480"></a> |
| <span class="sourceLineNo">481</span> try (RegionLocator regionLocator = connection.getRegionLocator(tableName)) {<a name="line.481"></a> |
| <span class="sourceLineNo">482</span> for (Pair<byte[], byte[]> dr : tmpRegionSet) {<a name="line.482"></a> |
| <span class="sourceLineNo">483</span> ServerName rsLocation = regionLocator.getRegionLocation(dr.getSecond()).getServerName();<a name="line.483"></a> |
| <span class="sourceLineNo">484</span> if (!daughterRegions.containsKey(rsLocation)) {<a name="line.484"></a> |
| <span class="sourceLineNo">485</span> LinkedList<Pair<byte[], byte[]>> entry = Lists.newLinkedList();<a name="line.485"></a> |
| <span class="sourceLineNo">486</span> daughterRegions.put(rsLocation, entry);<a name="line.486"></a> |
| <span class="sourceLineNo">487</span> }<a name="line.487"></a> |
| <span class="sourceLineNo">488</span> daughterRegions.get(rsLocation).add(dr);<a name="line.488"></a> |
| <span class="sourceLineNo">489</span> }<a name="line.489"></a> |
| <span class="sourceLineNo">490</span> LOG.debug("Done with bucketing. Split time!");<a name="line.490"></a> |
| <span class="sourceLineNo">491</span> long startTime = System.currentTimeMillis();<a name="line.491"></a> |
| <span class="sourceLineNo">492</span><a name="line.492"></a> |
| <span class="sourceLineNo">493</span> // Open the split file and modify it as splits finish<a name="line.493"></a> |
| <span class="sourceLineNo">494</span> byte[] rawData = readFile(fs, splitFile);<a name="line.494"></a> |
| <span class="sourceLineNo">495</span><a name="line.495"></a> |
| <span class="sourceLineNo">496</span> FSDataOutputStream splitOut = fs.create(splitFile);<a name="line.496"></a> |
| <span class="sourceLineNo">497</span> try {<a name="line.497"></a> |
| <span class="sourceLineNo">498</span> splitOut.write(rawData);<a name="line.498"></a> |
| <span class="sourceLineNo">499</span><a name="line.499"></a> |
| <span class="sourceLineNo">500</span> try {<a name="line.500"></a> |
| <span class="sourceLineNo">501</span> // *** split code ***<a name="line.501"></a> |
| <span class="sourceLineNo">502</span> while (!daughterRegions.isEmpty()) {<a name="line.502"></a> |
| <span class="sourceLineNo">503</span> LOG.debug(daughterRegions.size() + " RS have regions to splt.");<a name="line.503"></a> |
| <span class="sourceLineNo">504</span><a name="line.504"></a> |
| <span class="sourceLineNo">505</span> // Get ServerName to region count mapping<a name="line.505"></a> |
| <span class="sourceLineNo">506</span> final TreeMap<ServerName, Integer> rsSizes = Maps.newTreeMap();<a name="line.506"></a> |
| <span class="sourceLineNo">507</span> List<HRegionLocation> hrls = regionLocator.getAllRegionLocations();<a name="line.507"></a> |
| <span class="sourceLineNo">508</span> for (HRegionLocation hrl: hrls) {<a name="line.508"></a> |
| <span class="sourceLineNo">509</span> ServerName sn = hrl.getServerName();<a name="line.509"></a> |
| <span class="sourceLineNo">510</span> if (rsSizes.containsKey(sn)) {<a name="line.510"></a> |
| <span class="sourceLineNo">511</span> rsSizes.put(sn, rsSizes.get(sn) + 1);<a name="line.511"></a> |
| <span class="sourceLineNo">512</span> } else {<a name="line.512"></a> |
| <span class="sourceLineNo">513</span> rsSizes.put(sn, 1);<a name="line.513"></a> |
| <span class="sourceLineNo">514</span> }<a name="line.514"></a> |
| <span class="sourceLineNo">515</span> }<a name="line.515"></a> |
| <span class="sourceLineNo">516</span><a name="line.516"></a> |
| <span class="sourceLineNo">517</span> // Round-robin through the ServerName list. Choose the lightest-loaded servers<a name="line.517"></a> |
| <span class="sourceLineNo">518</span> // first to keep the master from load-balancing regions as we split.<a name="line.518"></a> |
| <span class="sourceLineNo">519</span> for (Map.Entry<ServerName, LinkedList<Pair<byte[], byte[]>>> daughterRegion :<a name="line.519"></a> |
| <span class="sourceLineNo">520</span> daughterRegions.entrySet()) {<a name="line.520"></a> |
| <span class="sourceLineNo">521</span> Pair<byte[], byte[]> dr = null;<a name="line.521"></a> |
| <span class="sourceLineNo">522</span> ServerName rsLoc = daughterRegion.getKey();<a name="line.522"></a> |
| <span class="sourceLineNo">523</span> LinkedList<Pair<byte[], byte[]>> regionList = daughterRegion.getValue();<a name="line.523"></a> |
| <span class="sourceLineNo">524</span><a name="line.524"></a> |
| <span class="sourceLineNo">525</span> // Find a region in the ServerName list that hasn't been moved<a name="line.525"></a> |
| <span class="sourceLineNo">526</span> LOG.debug("Finding a region on " + rsLoc);<a name="line.526"></a> |
| <span class="sourceLineNo">527</span> while (!regionList.isEmpty()) {<a name="line.527"></a> |
| <span class="sourceLineNo">528</span> dr = regionList.pop();<a name="line.528"></a> |
| <span class="sourceLineNo">529</span><a name="line.529"></a> |
| <span class="sourceLineNo">530</span> // get current region info<a name="line.530"></a> |
| <span class="sourceLineNo">531</span> byte[] split = dr.getSecond();<a name="line.531"></a> |
| <span class="sourceLineNo">532</span> HRegionLocation regionLoc = regionLocator.getRegionLocation(split);<a name="line.532"></a> |
| <span class="sourceLineNo">533</span><a name="line.533"></a> |
| <span class="sourceLineNo">534</span> // if this region moved locations<a name="line.534"></a> |
| <span class="sourceLineNo">535</span> ServerName newRs = regionLoc.getServerName();<a name="line.535"></a> |
| <span class="sourceLineNo">536</span> if (newRs.compareTo(rsLoc) != 0) {<a name="line.536"></a> |
| <span class="sourceLineNo">537</span> LOG.debug("Region with " + splitAlgo.rowToStr(split)<a name="line.537"></a> |
| <span class="sourceLineNo">538</span> + " moved to " + newRs + ". Relocating...");<a name="line.538"></a> |
| <span class="sourceLineNo">539</span> // relocate it, don't use it right now<a name="line.539"></a> |
| <span class="sourceLineNo">540</span> if (!daughterRegions.containsKey(newRs)) {<a name="line.540"></a> |
| <span class="sourceLineNo">541</span> LinkedList<Pair<byte[], byte[]>> entry = Lists.newLinkedList();<a name="line.541"></a> |
| <span class="sourceLineNo">542</span> daughterRegions.put(newRs, entry);<a name="line.542"></a> |
| <span class="sourceLineNo">543</span> }<a name="line.543"></a> |
| <span class="sourceLineNo">544</span> daughterRegions.get(newRs).add(dr);<a name="line.544"></a> |
| <span class="sourceLineNo">545</span> dr = null;<a name="line.545"></a> |
| <span class="sourceLineNo">546</span> continue;<a name="line.546"></a> |
| <span class="sourceLineNo">547</span> }<a name="line.547"></a> |
| <span class="sourceLineNo">548</span><a name="line.548"></a> |
| <span class="sourceLineNo">549</span> // make sure this region wasn't already split<a name="line.549"></a> |
| <span class="sourceLineNo">550</span> byte[] sk = regionLoc.getRegion().getStartKey();<a name="line.550"></a> |
| <span class="sourceLineNo">551</span> if (sk.length != 0) {<a name="line.551"></a> |
| <span class="sourceLineNo">552</span> if (Bytes.equals(split, sk)) {<a name="line.552"></a> |
| <span class="sourceLineNo">553</span> LOG.debug("Region already split on "<a name="line.553"></a> |
| <span class="sourceLineNo">554</span> + splitAlgo.rowToStr(split) + ". Skipping this region...");<a name="line.554"></a> |
| <span class="sourceLineNo">555</span> ++splitCount;<a name="line.555"></a> |
| <span class="sourceLineNo">556</span> dr = null;<a name="line.556"></a> |
| <span class="sourceLineNo">557</span> continue;<a name="line.557"></a> |
| <span class="sourceLineNo">558</span> }<a name="line.558"></a> |
| <span class="sourceLineNo">559</span> byte[] start = dr.getFirst();<a name="line.559"></a> |
| <span class="sourceLineNo">560</span> Preconditions.checkArgument(Bytes.equals(start, sk), splitAlgo<a name="line.560"></a> |
| <span class="sourceLineNo">561</span> .rowToStr(start) + " != " + splitAlgo.rowToStr(sk));<a name="line.561"></a> |
| <span class="sourceLineNo">562</span> }<a name="line.562"></a> |
| <span class="sourceLineNo">563</span><a name="line.563"></a> |
| <span class="sourceLineNo">564</span> // passed all checks! found a good region<a name="line.564"></a> |
| <span class="sourceLineNo">565</span> break;<a name="line.565"></a> |
| <span class="sourceLineNo">566</span> }<a name="line.566"></a> |
| <span class="sourceLineNo">567</span> if (regionList.isEmpty()) {<a name="line.567"></a> |
| <span class="sourceLineNo">568</span> daughterRegions.remove(rsLoc);<a name="line.568"></a> |
| <span class="sourceLineNo">569</span> }<a name="line.569"></a> |
| <span class="sourceLineNo">570</span> if (dr == null)<a name="line.570"></a> |
| <span class="sourceLineNo">571</span> continue;<a name="line.571"></a> |
| <span class="sourceLineNo">572</span><a name="line.572"></a> |
| <span class="sourceLineNo">573</span> // we have a good region, time to split!<a name="line.573"></a> |
| <span class="sourceLineNo">574</span> byte[] split = dr.getSecond();<a name="line.574"></a> |
| <span class="sourceLineNo">575</span> LOG.debug("Splitting at " + splitAlgo.rowToStr(split));<a name="line.575"></a> |
| <span class="sourceLineNo">576</span> try (Admin admin = connection.getAdmin()) {<a name="line.576"></a> |
| <span class="sourceLineNo">577</span> admin.split(tableName, split);<a name="line.577"></a> |
| <span class="sourceLineNo">578</span> }<a name="line.578"></a> |
| <span class="sourceLineNo">579</span><a name="line.579"></a> |
| <span class="sourceLineNo">580</span> LinkedList<Pair<byte[], byte[]>> finished = Lists.newLinkedList();<a name="line.580"></a> |
| <span class="sourceLineNo">581</span> LinkedList<Pair<byte[], byte[]>> local_finished = Lists.newLinkedList();<a name="line.581"></a> |
| <span class="sourceLineNo">582</span> if (conf.getBoolean("split.verify", true)) {<a name="line.582"></a> |
| <span class="sourceLineNo">583</span> // we need to verify and rate-limit our splits<a name="line.583"></a> |
| <span class="sourceLineNo">584</span> outstanding.addLast(dr);<a name="line.584"></a> |
| <span class="sourceLineNo">585</span> // with too many outstanding splits, wait for some to finish<a name="line.585"></a> |
| <span class="sourceLineNo">586</span> while (outstanding.size() >= MAX_OUTSTANDING) {<a name="line.586"></a> |
| <span class="sourceLineNo">587</span> LOG.debug("Wait for outstanding splits " + outstanding.size());<a name="line.587"></a> |
| <span class="sourceLineNo">588</span> local_finished = splitScan(outstanding, connection, tableName, splitAlgo);<a name="line.588"></a> |
| <span class="sourceLineNo">589</span> if (local_finished.isEmpty()) {<a name="line.589"></a> |
| <span class="sourceLineNo">590</span> Thread.sleep(30 * 1000);<a name="line.590"></a> |
| <span class="sourceLineNo">591</span> } else {<a name="line.591"></a> |
| <span class="sourceLineNo">592</span> finished.addAll(local_finished);<a name="line.592"></a> |
| <span class="sourceLineNo">593</span> outstanding.removeAll(local_finished);<a name="line.593"></a> |
| <span class="sourceLineNo">594</span> LOG.debug(local_finished.size() + " outstanding splits finished");<a name="line.594"></a> |
| <span class="sourceLineNo">595</span> }<a name="line.595"></a> |
| <span class="sourceLineNo">596</span> }<a name="line.596"></a> |
| <span class="sourceLineNo">597</span> } else {<a name="line.597"></a> |
| <span class="sourceLineNo">598</span> finished.add(dr);<a name="line.598"></a> |
| <span class="sourceLineNo">599</span> }<a name="line.599"></a> |
| <span class="sourceLineNo">600</span><a name="line.600"></a> |
| <span class="sourceLineNo">601</span> // mark each finished region as successfully split.<a name="line.601"></a> |
| <span class="sourceLineNo">602</span> for (Pair<byte[], byte[]> region : finished) {<a name="line.602"></a> |
| <span class="sourceLineNo">603</span> splitOut.writeChars("- " + splitAlgo.rowToStr(region.getFirst())<a name="line.603"></a> |
| <span class="sourceLineNo">604</span> + " " + splitAlgo.rowToStr(region.getSecond()) + "\n");<a name="line.604"></a> |
| <span class="sourceLineNo">605</span> splitCount++;<a name="line.605"></a> |
| <span class="sourceLineNo">606</span> if (splitCount % 10 == 0) {<a name="line.606"></a> |
| <span class="sourceLineNo">607</span> long tDiff = (System.currentTimeMillis() - startTime)<a name="line.607"></a> |
| <span class="sourceLineNo">608</span> / splitCount;<a name="line.608"></a> |
| <span class="sourceLineNo">609</span> LOG.debug("STATUS UPDATE: " + splitCount + " / " + origCount<a name="line.609"></a> |
| <span class="sourceLineNo">610</span> + ". Avg Time / Split = "<a name="line.610"></a> |
| <span class="sourceLineNo">611</span> + org.apache.hadoop.util.StringUtils.formatTime(tDiff));<a name="line.611"></a> |
| <span class="sourceLineNo">612</span> }<a name="line.612"></a> |
| <span class="sourceLineNo">613</span> }<a name="line.613"></a> |
| <span class="sourceLineNo">614</span> }<a name="line.614"></a> |
| <span class="sourceLineNo">615</span> }<a name="line.615"></a> |
| <span class="sourceLineNo">616</span> if (conf.getBoolean("split.verify", true)) {<a name="line.616"></a> |
| <span class="sourceLineNo">617</span> while (!outstanding.isEmpty()) {<a name="line.617"></a> |
| <span class="sourceLineNo">618</span> LOG.debug("Finally Wait for outstanding splits " + outstanding.size());<a name="line.618"></a> |
| <span class="sourceLineNo">619</span> LinkedList<Pair<byte[], byte[]>> finished = splitScan(outstanding,<a name="line.619"></a> |
| <span class="sourceLineNo">620</span> connection, tableName, splitAlgo);<a name="line.620"></a> |
| <span class="sourceLineNo">621</span> if (finished.isEmpty()) {<a name="line.621"></a> |
| <span class="sourceLineNo">622</span> Thread.sleep(30 * 1000);<a name="line.622"></a> |
| <span class="sourceLineNo">623</span> } else {<a name="line.623"></a> |
| <span class="sourceLineNo">624</span> outstanding.removeAll(finished);<a name="line.624"></a> |
| <span class="sourceLineNo">625</span> for (Pair<byte[], byte[]> region : finished) {<a name="line.625"></a> |
| <span class="sourceLineNo">626</span> splitOut.writeChars("- " + splitAlgo.rowToStr(region.getFirst())<a name="line.626"></a> |
| <span class="sourceLineNo">627</span> + " " + splitAlgo.rowToStr(region.getSecond()) + "\n");<a name="line.627"></a> |
| <span class="sourceLineNo">628</span> splitCount++;<a name="line.628"></a> |
| <span class="sourceLineNo">629</span> }<a name="line.629"></a> |
| <span class="sourceLineNo">630</span> LOG.debug("Finally " + finished.size() + " outstanding splits finished");<a name="line.630"></a> |
| <span class="sourceLineNo">631</span> }<a name="line.631"></a> |
| <span class="sourceLineNo">632</span> }<a name="line.632"></a> |
| <span class="sourceLineNo">633</span> }<a name="line.633"></a> |
| <span class="sourceLineNo">634</span> LOG.debug("All regions have been successfully split!");<a name="line.634"></a> |
| <span class="sourceLineNo">635</span> } finally {<a name="line.635"></a> |
| <span class="sourceLineNo">636</span> long tDiff = System.currentTimeMillis() - startTime;<a name="line.636"></a> |
| <span class="sourceLineNo">637</span> LOG.debug("TOTAL TIME = "<a name="line.637"></a> |
| <span class="sourceLineNo">638</span> + org.apache.hadoop.util.StringUtils.formatTime(tDiff));<a name="line.638"></a> |
| <span class="sourceLineNo">639</span> LOG.debug("Splits = " + splitCount);<a name="line.639"></a> |
| <span class="sourceLineNo">640</span> if (0 < splitCount) {<a name="line.640"></a> |
| <span class="sourceLineNo">641</span> LOG.debug("Avg Time / Split = "<a name="line.641"></a> |
| <span class="sourceLineNo">642</span> + org.apache.hadoop.util.StringUtils.formatTime(tDiff / splitCount));<a name="line.642"></a> |
| <span class="sourceLineNo">643</span> }<a name="line.643"></a> |
| <span class="sourceLineNo">644</span> }<a name="line.644"></a> |
| <span class="sourceLineNo">645</span> } finally {<a name="line.645"></a> |
| <span class="sourceLineNo">646</span> splitOut.close();<a name="line.646"></a> |
| <span class="sourceLineNo">647</span> fs.delete(splitFile, false);<a name="line.647"></a> |
| <span class="sourceLineNo">648</span> }<a name="line.648"></a> |
| <span class="sourceLineNo">649</span> }<a name="line.649"></a> |
| <span class="sourceLineNo">650</span> }<a name="line.650"></a> |
| <span class="sourceLineNo">651</span> }<a name="line.651"></a> |
| <span class="sourceLineNo">652</span><a name="line.652"></a> |
| <span class="sourceLineNo">653</span> /**<a name="line.653"></a> |
| <span class="sourceLineNo">654</span> * @throws IOException if the specified SplitAlgorithm class couldn't be<a name="line.654"></a> |
| <span class="sourceLineNo">655</span> * instantiated<a name="line.655"></a> |
| <span class="sourceLineNo">656</span> */<a name="line.656"></a> |
| <span class="sourceLineNo">657</span> public static SplitAlgorithm newSplitAlgoInstance(Configuration conf,<a name="line.657"></a> |
| <span class="sourceLineNo">658</span> String splitClassName) throws IOException {<a name="line.658"></a> |
| <span class="sourceLineNo">659</span> Class<?> splitClass;<a name="line.659"></a> |
| <span class="sourceLineNo">660</span><a name="line.660"></a> |
| <span class="sourceLineNo">661</span> // For split algorithms builtin to RegionSplitter, the user can specify<a name="line.661"></a> |
| <span class="sourceLineNo">662</span> // their simple class name instead of a fully qualified class name.<a name="line.662"></a> |
| <span class="sourceLineNo">663</span> if(splitClassName.equals(HexStringSplit.class.getSimpleName())) {<a name="line.663"></a> |
| <span class="sourceLineNo">664</span> splitClass = HexStringSplit.class;<a name="line.664"></a> |
| <span class="sourceLineNo">665</span> } else if (splitClassName.equals(DecimalStringSplit.class.getSimpleName())) {<a name="line.665"></a> |
| <span class="sourceLineNo">666</span> splitClass = DecimalStringSplit.class;<a name="line.666"></a> |
| <span class="sourceLineNo">667</span> } else if (splitClassName.equals(UniformSplit.class.getSimpleName())) {<a name="line.667"></a> |
| <span class="sourceLineNo">668</span> splitClass = UniformSplit.class;<a name="line.668"></a> |
| <span class="sourceLineNo">669</span> } else {<a name="line.669"></a> |
| <span class="sourceLineNo">670</span> try {<a name="line.670"></a> |
| <span class="sourceLineNo">671</span> splitClass = conf.getClassByName(splitClassName);<a name="line.671"></a> |
| <span class="sourceLineNo">672</span> } catch (ClassNotFoundException e) {<a name="line.672"></a> |
| <span class="sourceLineNo">673</span> throw new IOException("Couldn't load split class " + splitClassName, e);<a name="line.673"></a> |
| <span class="sourceLineNo">674</span> }<a name="line.674"></a> |
| <span class="sourceLineNo">675</span> if(splitClass == null) {<a name="line.675"></a> |
| <span class="sourceLineNo">676</span> throw new IOException("Failed loading split class " + splitClassName);<a name="line.676"></a> |
| <span class="sourceLineNo">677</span> }<a name="line.677"></a> |
| <span class="sourceLineNo">678</span> if(!SplitAlgorithm.class.isAssignableFrom(splitClass)) {<a name="line.678"></a> |
| <span class="sourceLineNo">679</span> throw new IOException(<a name="line.679"></a> |
| <span class="sourceLineNo">680</span> "Specified split class doesn't implement SplitAlgorithm");<a name="line.680"></a> |
| <span class="sourceLineNo">681</span> }<a name="line.681"></a> |
| <span class="sourceLineNo">682</span> }<a name="line.682"></a> |
| <span class="sourceLineNo">683</span> try {<a name="line.683"></a> |
| <span class="sourceLineNo">684</span> return splitClass.asSubclass(SplitAlgorithm.class).getDeclaredConstructor().newInstance();<a name="line.684"></a> |
| <span class="sourceLineNo">685</span> } catch (Exception e) {<a name="line.685"></a> |
| <span class="sourceLineNo">686</span> throw new IOException("Problem loading split algorithm: ", e);<a name="line.686"></a> |
| <span class="sourceLineNo">687</span> }<a name="line.687"></a> |
| <span class="sourceLineNo">688</span> }<a name="line.688"></a> |
| <span class="sourceLineNo">689</span><a name="line.689"></a> |
| <span class="sourceLineNo">690</span> static LinkedList<Pair<byte[], byte[]>> splitScan(<a name="line.690"></a> |
| <span class="sourceLineNo">691</span> LinkedList<Pair<byte[], byte[]>> regionList,<a name="line.691"></a> |
| <span class="sourceLineNo">692</span> final Connection connection,<a name="line.692"></a> |
| <span class="sourceLineNo">693</span> final TableName tableName,<a name="line.693"></a> |
| <span class="sourceLineNo">694</span> SplitAlgorithm splitAlgo)<a name="line.694"></a> |
| <span class="sourceLineNo">695</span> throws IOException, InterruptedException {<a name="line.695"></a> |
| <span class="sourceLineNo">696</span> LinkedList<Pair<byte[], byte[]>> finished = Lists.newLinkedList();<a name="line.696"></a> |
| <span class="sourceLineNo">697</span> LinkedList<Pair<byte[], byte[]>> logicalSplitting = Lists.newLinkedList();<a name="line.697"></a> |
| <span class="sourceLineNo">698</span> LinkedList<Pair<byte[], byte[]>> physicalSplitting = Lists.newLinkedList();<a name="line.698"></a> |
| <span class="sourceLineNo">699</span><a name="line.699"></a> |
| <span class="sourceLineNo">700</span> // Get table info<a name="line.700"></a> |
| <span class="sourceLineNo">701</span> Pair<Path, Path> tableDirAndSplitFile =<a name="line.701"></a> |
| <span class="sourceLineNo">702</span> getTableDirAndSplitFile(connection.getConfiguration(), tableName);<a name="line.702"></a> |
| <span class="sourceLineNo">703</span> Path tableDir = tableDirAndSplitFile.getFirst();<a name="line.703"></a> |
| <span class="sourceLineNo">704</span> FileSystem fs = tableDir.getFileSystem(connection.getConfiguration());<a name="line.704"></a> |
| <span class="sourceLineNo">705</span> // Clear the cache to forcibly refresh region information<a name="line.705"></a> |
| <span class="sourceLineNo">706</span> connection.clearRegionLocationCache();<a name="line.706"></a> |
| <span class="sourceLineNo">707</span> TableDescriptor htd = null;<a name="line.707"></a> |
| <span class="sourceLineNo">708</span> try (Table table = connection.getTable(tableName)) {<a name="line.708"></a> |
| <span class="sourceLineNo">709</span> htd = table.getDescriptor();<a name="line.709"></a> |
| <span class="sourceLineNo">710</span> }<a name="line.710"></a> |
| <span class="sourceLineNo">711</span> try (RegionLocator regionLocator = connection.getRegionLocator(tableName)) {<a name="line.711"></a> |
| <span class="sourceLineNo">712</span> // for every region that hasn't been verified as a finished split<a name="line.712"></a> |
| <span class="sourceLineNo">713</span> for (Pair<byte[], byte[]> region : regionList) {<a name="line.713"></a> |
| <span class="sourceLineNo">714</span> byte[] start = region.getFirst();<a name="line.714"></a> |
| <span class="sourceLineNo">715</span> byte[] split = region.getSecond();<a name="line.715"></a> |
| <span class="sourceLineNo">716</span><a name="line.716"></a> |
| <span class="sourceLineNo">717</span> // see if the new split daughter region has come online<a name="line.717"></a> |
| <span class="sourceLineNo">718</span> try {<a name="line.718"></a> |
| <span class="sourceLineNo">719</span> RegionInfo dri = regionLocator.getRegionLocation(split, true).getRegion();<a name="line.719"></a> |
| <span class="sourceLineNo">720</span> if (dri.isOffline() || !Bytes.equals(dri.getStartKey(), split)) {<a name="line.720"></a> |
| <span class="sourceLineNo">721</span> logicalSplitting.add(region);<a name="line.721"></a> |
| <span class="sourceLineNo">722</span> continue;<a name="line.722"></a> |
| <span class="sourceLineNo">723</span> }<a name="line.723"></a> |
| <span class="sourceLineNo">724</span> } catch (NoServerForRegionException nsfre) {<a name="line.724"></a> |
| <span class="sourceLineNo">725</span> // NSFRE will occur if the old hbase:meta entry has no server assigned<a name="line.725"></a> |
| <span class="sourceLineNo">726</span> LOG.info(nsfre.toString(), nsfre);<a name="line.726"></a> |
| <span class="sourceLineNo">727</span> logicalSplitting.add(region);<a name="line.727"></a> |
| <span class="sourceLineNo">728</span> continue;<a name="line.728"></a> |
| <span class="sourceLineNo">729</span> }<a name="line.729"></a> |
| <span class="sourceLineNo">730</span><a name="line.730"></a> |
| <span class="sourceLineNo">731</span> try {<a name="line.731"></a> |
| <span class="sourceLineNo">732</span> // when a daughter region is opened, a compaction is triggered<a name="line.732"></a> |
| <span class="sourceLineNo">733</span> // wait until compaction completes for both daughter regions<a name="line.733"></a> |
| <span class="sourceLineNo">734</span> LinkedList<RegionInfo> check = Lists.newLinkedList();<a name="line.734"></a> |
| <span class="sourceLineNo">735</span> check.add(regionLocator.getRegionLocation(start).getRegion());<a name="line.735"></a> |
| <span class="sourceLineNo">736</span> check.add(regionLocator.getRegionLocation(split).getRegion());<a name="line.736"></a> |
| <span class="sourceLineNo">737</span> for (RegionInfo hri : check.toArray(new RegionInfo[check.size()])) {<a name="line.737"></a> |
| <span class="sourceLineNo">738</span> byte[] sk = hri.getStartKey();<a name="line.738"></a> |
| <span class="sourceLineNo">739</span> if (sk.length == 0)<a name="line.739"></a> |
| <span class="sourceLineNo">740</span> sk = splitAlgo.firstRow();<a name="line.740"></a> |
| <span class="sourceLineNo">741</span><a name="line.741"></a> |
| <span class="sourceLineNo">742</span> HRegionFileSystem regionFs = HRegionFileSystem.openRegionFromFileSystem(<a name="line.742"></a> |
| <span class="sourceLineNo">743</span> connection.getConfiguration(), fs, tableDir, hri, true);<a name="line.743"></a> |
| <span class="sourceLineNo">744</span><a name="line.744"></a> |
| <span class="sourceLineNo">745</span> // Check every Column Family for that region -- check does not have references.<a name="line.745"></a> |
| <span class="sourceLineNo">746</span> boolean refFound = false;<a name="line.746"></a> |
| <span class="sourceLineNo">747</span> for (ColumnFamilyDescriptor c : htd.getColumnFamilies()) {<a name="line.747"></a> |
| <span class="sourceLineNo">748</span> if ((refFound = regionFs.hasReferences(c.getNameAsString()))) {<a name="line.748"></a> |
| <span class="sourceLineNo">749</span> break;<a name="line.749"></a> |
| <span class="sourceLineNo">750</span> }<a name="line.750"></a> |
| <span class="sourceLineNo">751</span> }<a name="line.751"></a> |
| <span class="sourceLineNo">752</span><a name="line.752"></a> |
| <span class="sourceLineNo">753</span> // compaction is completed when all reference files are gone<a name="line.753"></a> |
| <span class="sourceLineNo">754</span> if (!refFound) {<a name="line.754"></a> |
| <span class="sourceLineNo">755</span> check.remove(hri);<a name="line.755"></a> |
| <span class="sourceLineNo">756</span> }<a name="line.756"></a> |
| <span class="sourceLineNo">757</span> }<a name="line.757"></a> |
| <span class="sourceLineNo">758</span> if (check.isEmpty()) {<a name="line.758"></a> |
| <span class="sourceLineNo">759</span> finished.add(region);<a name="line.759"></a> |
| <span class="sourceLineNo">760</span> } else {<a name="line.760"></a> |
| <span class="sourceLineNo">761</span> physicalSplitting.add(region);<a name="line.761"></a> |
| <span class="sourceLineNo">762</span> }<a name="line.762"></a> |
| <span class="sourceLineNo">763</span> } catch (NoServerForRegionException nsfre) {<a name="line.763"></a> |
| <span class="sourceLineNo">764</span> LOG.debug("No Server Exception thrown for: " + splitAlgo.rowToStr(start));<a name="line.764"></a> |
| <span class="sourceLineNo">765</span> physicalSplitting.add(region);<a name="line.765"></a> |
| <span class="sourceLineNo">766</span> connection.clearRegionLocationCache();<a name="line.766"></a> |
| <span class="sourceLineNo">767</span> }<a name="line.767"></a> |
| <span class="sourceLineNo">768</span> }<a name="line.768"></a> |
| <span class="sourceLineNo">769</span><a name="line.769"></a> |
| <span class="sourceLineNo">770</span> LOG.debug("Split Scan: " + finished.size() + " finished / "<a name="line.770"></a> |
| <span class="sourceLineNo">771</span> + logicalSplitting.size() + " split wait / "<a name="line.771"></a> |
| <span class="sourceLineNo">772</span> + physicalSplitting.size() + " reference wait");<a name="line.772"></a> |
| <span class="sourceLineNo">773</span><a name="line.773"></a> |
| <span class="sourceLineNo">774</span> return finished;<a name="line.774"></a> |
| <span class="sourceLineNo">775</span> }<a name="line.775"></a> |
| <span class="sourceLineNo">776</span> }<a name="line.776"></a> |
| <span class="sourceLineNo">777</span><a name="line.777"></a> |
| <span class="sourceLineNo">778</span> /**<a name="line.778"></a> |
| <span class="sourceLineNo">779</span> * @param conf<a name="line.779"></a> |
| <span class="sourceLineNo">780</span> * @param tableName<a name="line.780"></a> |
| <span class="sourceLineNo">781</span> * @return A Pair where first item is table dir and second is the split file.<a name="line.781"></a> |
| <span class="sourceLineNo">782</span> * @throws IOException if a remote or network exception occurs<a name="line.782"></a> |
| <span class="sourceLineNo">783</span> */<a name="line.783"></a> |
| <span class="sourceLineNo">784</span> private static Pair<Path, Path> getTableDirAndSplitFile(final Configuration conf,<a name="line.784"></a> |
| <span class="sourceLineNo">785</span> final TableName tableName) throws IOException {<a name="line.785"></a> |
| <span class="sourceLineNo">786</span> Path hbDir = CommonFSUtils.getRootDir(conf);<a name="line.786"></a> |
| <span class="sourceLineNo">787</span> Path tableDir = CommonFSUtils.getTableDir(hbDir, tableName);<a name="line.787"></a> |
| <span class="sourceLineNo">788</span> Path splitFile = new Path(tableDir, "_balancedSplit");<a name="line.788"></a> |
| <span class="sourceLineNo">789</span> return new Pair<>(tableDir, splitFile);<a name="line.789"></a> |
| <span class="sourceLineNo">790</span> }<a name="line.790"></a> |
| <span class="sourceLineNo">791</span><a name="line.791"></a> |
| <span class="sourceLineNo">792</span> static LinkedList<Pair<byte[], byte[]>> getSplits(final Connection connection,<a name="line.792"></a> |
| <span class="sourceLineNo">793</span> TableName tableName, SplitAlgorithm splitAlgo)<a name="line.793"></a> |
| <span class="sourceLineNo">794</span> throws IOException {<a name="line.794"></a> |
| <span class="sourceLineNo">795</span> Pair<Path, Path> tableDirAndSplitFile =<a name="line.795"></a> |
| <span class="sourceLineNo">796</span> getTableDirAndSplitFile(connection.getConfiguration(), tableName);<a name="line.796"></a> |
| <span class="sourceLineNo">797</span> Path tableDir = tableDirAndSplitFile.getFirst();<a name="line.797"></a> |
| <span class="sourceLineNo">798</span> Path splitFile = tableDirAndSplitFile.getSecond();<a name="line.798"></a> |
| <span class="sourceLineNo">799</span><a name="line.799"></a> |
| <span class="sourceLineNo">800</span> FileSystem fs = tableDir.getFileSystem(connection.getConfiguration());<a name="line.800"></a> |
| <span class="sourceLineNo">801</span><a name="line.801"></a> |
| <span class="sourceLineNo">802</span> // Using strings because (new byte[]{0}).equals(new byte[]{0}) == false<a name="line.802"></a> |
| <span class="sourceLineNo">803</span> Set<Pair<String, String>> daughterRegions = Sets.newHashSet();<a name="line.803"></a> |
| <span class="sourceLineNo">804</span><a name="line.804"></a> |
| <span class="sourceLineNo">805</span> // Does a split file exist?<a name="line.805"></a> |
| <span class="sourceLineNo">806</span> if (!fs.exists(splitFile)) {<a name="line.806"></a> |
| <span class="sourceLineNo">807</span> // NO = fresh start. calculate splits to make<a name="line.807"></a> |
| <span class="sourceLineNo">808</span> LOG.debug("No " + splitFile.getName() + " file. Calculating splits ");<a name="line.808"></a> |
| <span class="sourceLineNo">809</span><a name="line.809"></a> |
| <span class="sourceLineNo">810</span> // Query meta for all regions in the table<a name="line.810"></a> |
| <span class="sourceLineNo">811</span> Set<Pair<byte[], byte[]>> rows = Sets.newHashSet();<a name="line.811"></a> |
| <span class="sourceLineNo">812</span> Pair<byte[][], byte[][]> tmp = null;<a name="line.812"></a> |
| <span class="sourceLineNo">813</span> try (RegionLocator regionLocator = connection.getRegionLocator(tableName)) {<a name="line.813"></a> |
| <span class="sourceLineNo">814</span> tmp = regionLocator.getStartEndKeys();<a name="line.814"></a> |
| <span class="sourceLineNo">815</span> }<a name="line.815"></a> |
| <span class="sourceLineNo">816</span> Preconditions.checkArgument(tmp.getFirst().length == tmp.getSecond().length,<a name="line.816"></a> |
| <span class="sourceLineNo">817</span> "Start and End rows should be equivalent");<a name="line.817"></a> |
| <span class="sourceLineNo">818</span> for (int i = 0; i < tmp.getFirst().length; ++i) {<a name="line.818"></a> |
| <span class="sourceLineNo">819</span> byte[] start = tmp.getFirst()[i], end = tmp.getSecond()[i];<a name="line.819"></a> |
| <span class="sourceLineNo">820</span> if (start.length == 0)<a name="line.820"></a> |
| <span class="sourceLineNo">821</span> start = splitAlgo.firstRow();<a name="line.821"></a> |
| <span class="sourceLineNo">822</span> if (end.length == 0)<a name="line.822"></a> |
| <span class="sourceLineNo">823</span> end = splitAlgo.lastRow();<a name="line.823"></a> |
| <span class="sourceLineNo">824</span> rows.add(Pair.newPair(start, end));<a name="line.824"></a> |
| <span class="sourceLineNo">825</span> }<a name="line.825"></a> |
| <span class="sourceLineNo">826</span> LOG.debug("Table " + tableName + " has " + rows.size() + " regions that will be split.");<a name="line.826"></a> |
| <span class="sourceLineNo">827</span><a name="line.827"></a> |
| <span class="sourceLineNo">828</span> // prepare the split file<a name="line.828"></a> |
| <span class="sourceLineNo">829</span> Path tmpFile = new Path(tableDir, "_balancedSplit_prepare");<a name="line.829"></a> |
| <span class="sourceLineNo">830</span> FSDataOutputStream tmpOut = fs.create(tmpFile);<a name="line.830"></a> |
| <span class="sourceLineNo">831</span><a name="line.831"></a> |
| <span class="sourceLineNo">832</span> // calculate all the splits == [daughterRegions] = [(start, splitPoint)]<a name="line.832"></a> |
| <span class="sourceLineNo">833</span> for (Pair<byte[], byte[]> r : rows) {<a name="line.833"></a> |
| <span class="sourceLineNo">834</span> byte[] splitPoint = splitAlgo.split(r.getFirst(), r.getSecond());<a name="line.834"></a> |
| <span class="sourceLineNo">835</span> String startStr = splitAlgo.rowToStr(r.getFirst());<a name="line.835"></a> |
| <span class="sourceLineNo">836</span> String splitStr = splitAlgo.rowToStr(splitPoint);<a name="line.836"></a> |
| <span class="sourceLineNo">837</span> daughterRegions.add(Pair.newPair(startStr, splitStr));<a name="line.837"></a> |
| <span class="sourceLineNo">838</span> LOG.debug("Will Split [" + startStr + " , "<a name="line.838"></a> |
| <span class="sourceLineNo">839</span> + splitAlgo.rowToStr(r.getSecond()) + ") at " + splitStr);<a name="line.839"></a> |
| <span class="sourceLineNo">840</span> tmpOut.writeChars("+ " + startStr + splitAlgo.separator() + splitStr<a name="line.840"></a> |
| <span class="sourceLineNo">841</span> + "\n");<a name="line.841"></a> |
| <span class="sourceLineNo">842</span> }<a name="line.842"></a> |
| <span class="sourceLineNo">843</span> tmpOut.close();<a name="line.843"></a> |
| <span class="sourceLineNo">844</span> fs.rename(tmpFile, splitFile);<a name="line.844"></a> |
| <span class="sourceLineNo">845</span> } else {<a name="line.845"></a> |
| <span class="sourceLineNo">846</span> LOG.debug("_balancedSplit file found. Replay log to restore state...");<a name="line.846"></a> |
| <span class="sourceLineNo">847</span> RecoverLeaseFSUtils.recoverFileLease(fs, splitFile, connection.getConfiguration(), null);<a name="line.847"></a> |
| <span class="sourceLineNo">848</span><a name="line.848"></a> |
| <span class="sourceLineNo">849</span> // parse split file and process remaining splits<a name="line.849"></a> |
| <span class="sourceLineNo">850</span> FSDataInputStream tmpIn = fs.open(splitFile);<a name="line.850"></a> |
| <span class="sourceLineNo">851</span> StringBuilder sb = new StringBuilder(tmpIn.available());<a name="line.851"></a> |
| <span class="sourceLineNo">852</span> while (tmpIn.available() > 0) {<a name="line.852"></a> |
| <span class="sourceLineNo">853</span> sb.append(tmpIn.readChar());<a name="line.853"></a> |
| <span class="sourceLineNo">854</span> }<a name="line.854"></a> |
| <span class="sourceLineNo">855</span> tmpIn.close();<a name="line.855"></a> |
| <span class="sourceLineNo">856</span> for (String line : sb.toString().split("\n")) {<a name="line.856"></a> |
| <span class="sourceLineNo">857</span> String[] cmd = line.split(splitAlgo.separator());<a name="line.857"></a> |
| <span class="sourceLineNo">858</span> Preconditions.checkArgument(3 == cmd.length);<a name="line.858"></a> |
| <span class="sourceLineNo">859</span> byte[] start = splitAlgo.strToRow(cmd[1]);<a name="line.859"></a> |
| <span class="sourceLineNo">860</span> String startStr = splitAlgo.rowToStr(start);<a name="line.860"></a> |
| <span class="sourceLineNo">861</span> byte[] splitPoint = splitAlgo.strToRow(cmd[2]);<a name="line.861"></a> |
| <span class="sourceLineNo">862</span> String splitStr = splitAlgo.rowToStr(splitPoint);<a name="line.862"></a> |
| <span class="sourceLineNo">863</span> Pair<String, String> r = Pair.newPair(startStr, splitStr);<a name="line.863"></a> |
| <span class="sourceLineNo">864</span> if (cmd[0].equals("+")) {<a name="line.864"></a> |
| <span class="sourceLineNo">865</span> LOG.debug("Adding: " + r);<a name="line.865"></a> |
| <span class="sourceLineNo">866</span> daughterRegions.add(r);<a name="line.866"></a> |
| <span class="sourceLineNo">867</span> } else {<a name="line.867"></a> |
| <span class="sourceLineNo">868</span> LOG.debug("Removing: " + r);<a name="line.868"></a> |
| <span class="sourceLineNo">869</span> Preconditions.checkArgument(cmd[0].equals("-"),<a name="line.869"></a> |
| <span class="sourceLineNo">870</span> "Unknown option: " + cmd[0]);<a name="line.870"></a> |
| <span class="sourceLineNo">871</span> Preconditions.checkState(daughterRegions.contains(r),<a name="line.871"></a> |
| <span class="sourceLineNo">872</span> "Missing row: " + r);<a name="line.872"></a> |
| <span class="sourceLineNo">873</span> daughterRegions.remove(r);<a name="line.873"></a> |
| <span class="sourceLineNo">874</span> }<a name="line.874"></a> |
| <span class="sourceLineNo">875</span> }<a name="line.875"></a> |
| <span class="sourceLineNo">876</span> LOG.debug("Done reading. " + daughterRegions.size() + " regions left.");<a name="line.876"></a> |
| <span class="sourceLineNo">877</span> }<a name="line.877"></a> |
| <span class="sourceLineNo">878</span> LinkedList<Pair<byte[], byte[]>> ret = Lists.newLinkedList();<a name="line.878"></a> |
| <span class="sourceLineNo">879</span> for (Pair<String, String> r : daughterRegions) {<a name="line.879"></a> |
| <span class="sourceLineNo">880</span> ret.add(Pair.newPair(splitAlgo.strToRow(r.getFirst()), splitAlgo<a name="line.880"></a> |
| <span class="sourceLineNo">881</span> .strToRow(r.getSecond())));<a name="line.881"></a> |
| <span class="sourceLineNo">882</span> }<a name="line.882"></a> |
| <span class="sourceLineNo">883</span> return ret;<a name="line.883"></a> |
| <span class="sourceLineNo">884</span> }<a name="line.884"></a> |
| <span class="sourceLineNo">885</span><a name="line.885"></a> |
| <span class="sourceLineNo">886</span> /**<a name="line.886"></a> |
| <span class="sourceLineNo">887</span> * HexStringSplit is a well-known {@link SplitAlgorithm} for choosing region<a name="line.887"></a> |
| <span class="sourceLineNo">888</span> * boundaries. The format of a HexStringSplit region boundary is the ASCII<a name="line.888"></a> |
| <span class="sourceLineNo">889</span> * representation of an MD5 checksum, or any other uniformly distributed<a name="line.889"></a> |
| <span class="sourceLineNo">890</span> * hexadecimal value. Row are hex-encoded long values in the range<a name="line.890"></a> |
| <span class="sourceLineNo">891</span> * <b>"00000000" =&gt; "FFFFFFFF"</b> and are left-padded with zeros to keep the<a name="line.891"></a> |
| <span class="sourceLineNo">892</span> * same order lexicographically as if they were binary.<a name="line.892"></a> |
| <span class="sourceLineNo">893</span> *<a name="line.893"></a> |
| <span class="sourceLineNo">894</span> * Since this split algorithm uses hex strings as keys, it is easy to read &amp;<a name="line.894"></a> |
| <span class="sourceLineNo">895</span> * write in the shell but takes up more space and may be non-intuitive.<a name="line.895"></a> |
| <span class="sourceLineNo">896</span> */<a name="line.896"></a> |
| <span class="sourceLineNo">897</span> public static class HexStringSplit extends NumberStringSplit {<a name="line.897"></a> |
| <span class="sourceLineNo">898</span> final static String DEFAULT_MIN_HEX = "00000000";<a name="line.898"></a> |
| <span class="sourceLineNo">899</span> final static String DEFAULT_MAX_HEX = "FFFFFFFF";<a name="line.899"></a> |
| <span class="sourceLineNo">900</span> final static int RADIX_HEX = 16;<a name="line.900"></a> |
| <span class="sourceLineNo">901</span><a name="line.901"></a> |
| <span class="sourceLineNo">902</span> public HexStringSplit() {<a name="line.902"></a> |
| <span class="sourceLineNo">903</span> super(DEFAULT_MIN_HEX, DEFAULT_MAX_HEX, RADIX_HEX);<a name="line.903"></a> |
| <span class="sourceLineNo">904</span> }<a name="line.904"></a> |
| <span class="sourceLineNo">905</span><a name="line.905"></a> |
| <span class="sourceLineNo">906</span> }<a name="line.906"></a> |
| <span class="sourceLineNo">907</span><a name="line.907"></a> |
| <span class="sourceLineNo">908</span> /**<a name="line.908"></a> |
| <span class="sourceLineNo">909</span> * The format of a DecimalStringSplit region boundary is the ASCII representation of<a name="line.909"></a> |
| <span class="sourceLineNo">910</span> * reversed sequential number, or any other uniformly distributed decimal value.<a name="line.910"></a> |
| <span class="sourceLineNo">911</span> * Row are decimal-encoded long values in the range<a name="line.911"></a> |
| <span class="sourceLineNo">912</span> * <b>"00000000" =&gt; "99999999"</b> and are left-padded with zeros to keep the<a name="line.912"></a> |
| <span class="sourceLineNo">913</span> * same order lexicographically as if they were binary.<a name="line.913"></a> |
| <span class="sourceLineNo">914</span> */<a name="line.914"></a> |
| <span class="sourceLineNo">915</span> public static class DecimalStringSplit extends NumberStringSplit {<a name="line.915"></a> |
| <span class="sourceLineNo">916</span> final static String DEFAULT_MIN_DEC = "00000000";<a name="line.916"></a> |
| <span class="sourceLineNo">917</span> final static String DEFAULT_MAX_DEC = "99999999";<a name="line.917"></a> |
| <span class="sourceLineNo">918</span> final static int RADIX_DEC = 10;<a name="line.918"></a> |
| <span class="sourceLineNo">919</span><a name="line.919"></a> |
| <span class="sourceLineNo">920</span> public DecimalStringSplit() {<a name="line.920"></a> |
| <span class="sourceLineNo">921</span> super(DEFAULT_MIN_DEC, DEFAULT_MAX_DEC, RADIX_DEC);<a name="line.921"></a> |
| <span class="sourceLineNo">922</span> }<a name="line.922"></a> |
| <span class="sourceLineNo">923</span><a name="line.923"></a> |
| <span class="sourceLineNo">924</span> }<a name="line.924"></a> |
| <span class="sourceLineNo">925</span><a name="line.925"></a> |
| <span class="sourceLineNo">926</span> public abstract static class NumberStringSplit implements SplitAlgorithm {<a name="line.926"></a> |
| <span class="sourceLineNo">927</span><a name="line.927"></a> |
| <span class="sourceLineNo">928</span> String firstRow;<a name="line.928"></a> |
| <span class="sourceLineNo">929</span> BigInteger firstRowInt;<a name="line.929"></a> |
| <span class="sourceLineNo">930</span> String lastRow;<a name="line.930"></a> |
| <span class="sourceLineNo">931</span> BigInteger lastRowInt;<a name="line.931"></a> |
| <span class="sourceLineNo">932</span> int rowComparisonLength;<a name="line.932"></a> |
| <span class="sourceLineNo">933</span> int radix;<a name="line.933"></a> |
| <span class="sourceLineNo">934</span><a name="line.934"></a> |
| <span class="sourceLineNo">935</span> NumberStringSplit(String minRow, String maxRow, int radix) {<a name="line.935"></a> |
| <span class="sourceLineNo">936</span> this.firstRow = minRow;<a name="line.936"></a> |
| <span class="sourceLineNo">937</span> this.lastRow = maxRow;<a name="line.937"></a> |
| <span class="sourceLineNo">938</span> this.radix = radix;<a name="line.938"></a> |
| <span class="sourceLineNo">939</span> this.firstRowInt = BigInteger.ZERO;<a name="line.939"></a> |
| <span class="sourceLineNo">940</span> this.lastRowInt = new BigInteger(lastRow, this.radix);<a name="line.940"></a> |
| <span class="sourceLineNo">941</span> this.rowComparisonLength = lastRow.length();<a name="line.941"></a> |
| <span class="sourceLineNo">942</span> }<a name="line.942"></a> |
| <span class="sourceLineNo">943</span><a name="line.943"></a> |
| <span class="sourceLineNo">944</span> @Override<a name="line.944"></a> |
| <span class="sourceLineNo">945</span> public byte[] split(byte[] start, byte[] end) {<a name="line.945"></a> |
| <span class="sourceLineNo">946</span> BigInteger s = convertToBigInteger(start);<a name="line.946"></a> |
| <span class="sourceLineNo">947</span> BigInteger e = convertToBigInteger(end);<a name="line.947"></a> |
| <span class="sourceLineNo">948</span> Preconditions.checkArgument(!e.equals(BigInteger.ZERO));<a name="line.948"></a> |
| <span class="sourceLineNo">949</span> return convertToByte(split2(s, e));<a name="line.949"></a> |
| <span class="sourceLineNo">950</span> }<a name="line.950"></a> |
| <span class="sourceLineNo">951</span><a name="line.951"></a> |
| <span class="sourceLineNo">952</span> @Override<a name="line.952"></a> |
| <span class="sourceLineNo">953</span> public byte[][] split(int n) {<a name="line.953"></a> |
| <span class="sourceLineNo">954</span> Preconditions.checkArgument(lastRowInt.compareTo(firstRowInt) > 0,<a name="line.954"></a> |
| <span class="sourceLineNo">955</span> "last row (%s) is configured less than first row (%s)", lastRow,<a name="line.955"></a> |
| <span class="sourceLineNo">956</span> firstRow);<a name="line.956"></a> |
| <span class="sourceLineNo">957</span> // +1 to range because the last row is inclusive<a name="line.957"></a> |
| <span class="sourceLineNo">958</span> BigInteger range = lastRowInt.subtract(firstRowInt).add(BigInteger.ONE);<a name="line.958"></a> |
| <span class="sourceLineNo">959</span> Preconditions.checkState(range.compareTo(BigInteger.valueOf(n)) >= 0,<a name="line.959"></a> |
| <span class="sourceLineNo">960</span> "split granularity (%s) is greater than the range (%s)", n, range);<a name="line.960"></a> |
| <span class="sourceLineNo">961</span><a name="line.961"></a> |
| <span class="sourceLineNo">962</span> BigInteger[] splits = new BigInteger[n - 1];<a name="line.962"></a> |
| <span class="sourceLineNo">963</span> BigInteger sizeOfEachSplit = range.divide(BigInteger.valueOf(n));<a name="line.963"></a> |
| <span class="sourceLineNo">964</span> for (int i = 1; i < n; i++) {<a name="line.964"></a> |
| <span class="sourceLineNo">965</span> // NOTE: this means the last region gets all the slop.<a name="line.965"></a> |
| <span class="sourceLineNo">966</span> // This is not a big deal if we're assuming n << MAXHEX<a name="line.966"></a> |
| <span class="sourceLineNo">967</span> splits[i - 1] = firstRowInt.add(sizeOfEachSplit.multiply(BigInteger<a name="line.967"></a> |
| <span class="sourceLineNo">968</span> .valueOf(i)));<a name="line.968"></a> |
| <span class="sourceLineNo">969</span> }<a name="line.969"></a> |
| <span class="sourceLineNo">970</span> return convertToBytes(splits);<a name="line.970"></a> |
| <span class="sourceLineNo">971</span> }<a name="line.971"></a> |
| <span class="sourceLineNo">972</span><a name="line.972"></a> |
| <span class="sourceLineNo">973</span> @Override<a name="line.973"></a> |
| <span class="sourceLineNo">974</span> public byte[][] split(byte[] start, byte[] end, int numSplits, boolean inclusive) {<a name="line.974"></a> |
| <span class="sourceLineNo">975</span> BigInteger s = convertToBigInteger(start);<a name="line.975"></a> |
| <span class="sourceLineNo">976</span> BigInteger e = convertToBigInteger(end);<a name="line.976"></a> |
| <span class="sourceLineNo">977</span><a name="line.977"></a> |
| <span class="sourceLineNo">978</span> Preconditions.checkArgument(e.compareTo(s) > 0,<a name="line.978"></a> |
| <span class="sourceLineNo">979</span> "last row (%s) is configured less than first row (%s)", rowToStr(end),<a name="line.979"></a> |
| <span class="sourceLineNo">980</span> end);<a name="line.980"></a> |
| <span class="sourceLineNo">981</span> // +1 to range because the last row is inclusive<a name="line.981"></a> |
| <span class="sourceLineNo">982</span> BigInteger range = e.subtract(s).add(BigInteger.ONE);<a name="line.982"></a> |
| <span class="sourceLineNo">983</span> Preconditions.checkState(range.compareTo(BigInteger.valueOf(numSplits)) >= 0,<a name="line.983"></a> |
| <span class="sourceLineNo">984</span> "split granularity (%s) is greater than the range (%s)", numSplits, range);<a name="line.984"></a> |
| <span class="sourceLineNo">985</span><a name="line.985"></a> |
| <span class="sourceLineNo">986</span> BigInteger[] splits = new BigInteger[numSplits - 1];<a name="line.986"></a> |
| <span class="sourceLineNo">987</span> BigInteger sizeOfEachSplit = range.divide(BigInteger.valueOf(numSplits));<a name="line.987"></a> |
| <span class="sourceLineNo">988</span> for (int i = 1; i < numSplits; i++) {<a name="line.988"></a> |
| <span class="sourceLineNo">989</span> // NOTE: this means the last region gets all the slop.<a name="line.989"></a> |
| <span class="sourceLineNo">990</span> // This is not a big deal if we're assuming n << MAXHEX<a name="line.990"></a> |
| <span class="sourceLineNo">991</span> splits[i - 1] = s.add(sizeOfEachSplit.multiply(BigInteger<a name="line.991"></a> |
| <span class="sourceLineNo">992</span> .valueOf(i)));<a name="line.992"></a> |
| <span class="sourceLineNo">993</span> }<a name="line.993"></a> |
| <span class="sourceLineNo">994</span><a name="line.994"></a> |
| <span class="sourceLineNo">995</span> if (inclusive) {<a name="line.995"></a> |
| <span class="sourceLineNo">996</span> BigInteger[] inclusiveSplitPoints = new BigInteger[numSplits + 1];<a name="line.996"></a> |
| <span class="sourceLineNo">997</span> inclusiveSplitPoints[0] = convertToBigInteger(start);<a name="line.997"></a> |
| <span class="sourceLineNo">998</span> inclusiveSplitPoints[numSplits] = convertToBigInteger(end);<a name="line.998"></a> |
| <span class="sourceLineNo">999</span> System.arraycopy(splits, 0, inclusiveSplitPoints, 1, splits.length);<a name="line.999"></a> |
| <span class="sourceLineNo">1000</span> return convertToBytes(inclusiveSplitPoints);<a name="line.1000"></a> |
| <span class="sourceLineNo">1001</span> } else {<a name="line.1001"></a> |
| <span class="sourceLineNo">1002</span> return convertToBytes(splits);<a name="line.1002"></a> |
| <span class="sourceLineNo">1003</span> }<a name="line.1003"></a> |
| <span class="sourceLineNo">1004</span> }<a name="line.1004"></a> |
| <span class="sourceLineNo">1005</span><a name="line.1005"></a> |
| <span class="sourceLineNo">1006</span> @Override<a name="line.1006"></a> |
| <span class="sourceLineNo">1007</span> public byte[] firstRow() {<a name="line.1007"></a> |
| <span class="sourceLineNo">1008</span> return convertToByte(firstRowInt);<a name="line.1008"></a> |
| <span class="sourceLineNo">1009</span> }<a name="line.1009"></a> |
| <span class="sourceLineNo">1010</span><a name="line.1010"></a> |
| <span class="sourceLineNo">1011</span> @Override<a name="line.1011"></a> |
| <span class="sourceLineNo">1012</span> public byte[] lastRow() {<a name="line.1012"></a> |
| <span class="sourceLineNo">1013</span> return convertToByte(lastRowInt);<a name="line.1013"></a> |
| <span class="sourceLineNo">1014</span> }<a name="line.1014"></a> |
| <span class="sourceLineNo">1015</span><a name="line.1015"></a> |
| <span class="sourceLineNo">1016</span> @Override<a name="line.1016"></a> |
| <span class="sourceLineNo">1017</span> public void setFirstRow(String userInput) {<a name="line.1017"></a> |
| <span class="sourceLineNo">1018</span> firstRow = userInput;<a name="line.1018"></a> |
| <span class="sourceLineNo">1019</span> firstRowInt = new BigInteger(firstRow, radix);<a name="line.1019"></a> |
| <span class="sourceLineNo">1020</span> }<a name="line.1020"></a> |
| <span class="sourceLineNo">1021</span><a name="line.1021"></a> |
| <span class="sourceLineNo">1022</span> @Override<a name="line.1022"></a> |
| <span class="sourceLineNo">1023</span> public void setLastRow(String userInput) {<a name="line.1023"></a> |
| <span class="sourceLineNo">1024</span> lastRow = userInput;<a name="line.1024"></a> |
| <span class="sourceLineNo">1025</span> lastRowInt = new BigInteger(lastRow, radix);<a name="line.1025"></a> |
| <span class="sourceLineNo">1026</span> // Precondition: lastRow > firstRow, so last's length is the greater<a name="line.1026"></a> |
| <span class="sourceLineNo">1027</span> rowComparisonLength = lastRow.length();<a name="line.1027"></a> |
| <span class="sourceLineNo">1028</span> }<a name="line.1028"></a> |
| <span class="sourceLineNo">1029</span><a name="line.1029"></a> |
| <span class="sourceLineNo">1030</span> @Override<a name="line.1030"></a> |
| <span class="sourceLineNo">1031</span> public byte[] strToRow(String in) {<a name="line.1031"></a> |
| <span class="sourceLineNo">1032</span> return convertToByte(new BigInteger(in, radix));<a name="line.1032"></a> |
| <span class="sourceLineNo">1033</span> }<a name="line.1033"></a> |
| <span class="sourceLineNo">1034</span><a name="line.1034"></a> |
| <span class="sourceLineNo">1035</span> @Override<a name="line.1035"></a> |
| <span class="sourceLineNo">1036</span> public String rowToStr(byte[] row) {<a name="line.1036"></a> |
| <span class="sourceLineNo">1037</span> return Bytes.toStringBinary(row);<a name="line.1037"></a> |
| <span class="sourceLineNo">1038</span> }<a name="line.1038"></a> |
| <span class="sourceLineNo">1039</span><a name="line.1039"></a> |
| <span class="sourceLineNo">1040</span> @Override<a name="line.1040"></a> |
| <span class="sourceLineNo">1041</span> public String separator() {<a name="line.1041"></a> |
| <span class="sourceLineNo">1042</span> return " ";<a name="line.1042"></a> |
| <span class="sourceLineNo">1043</span> }<a name="line.1043"></a> |
| <span class="sourceLineNo">1044</span><a name="line.1044"></a> |
| <span class="sourceLineNo">1045</span> @Override<a name="line.1045"></a> |
| <span class="sourceLineNo">1046</span> public void setFirstRow(byte[] userInput) {<a name="line.1046"></a> |
| <span class="sourceLineNo">1047</span> firstRow = Bytes.toString(userInput);<a name="line.1047"></a> |
| <span class="sourceLineNo">1048</span> }<a name="line.1048"></a> |
| <span class="sourceLineNo">1049</span><a name="line.1049"></a> |
| <span class="sourceLineNo">1050</span> @Override<a name="line.1050"></a> |
| <span class="sourceLineNo">1051</span> public void setLastRow(byte[] userInput) {<a name="line.1051"></a> |
| <span class="sourceLineNo">1052</span> lastRow = Bytes.toString(userInput);<a name="line.1052"></a> |
| <span class="sourceLineNo">1053</span> }<a name="line.1053"></a> |
| <span class="sourceLineNo">1054</span><a name="line.1054"></a> |
| <span class="sourceLineNo">1055</span> /**<a name="line.1055"></a> |
| <span class="sourceLineNo">1056</span> * Divide 2 numbers in half (for split algorithm)<a name="line.1056"></a> |
| <span class="sourceLineNo">1057</span> *<a name="line.1057"></a> |
| <span class="sourceLineNo">1058</span> * @param a number #1<a name="line.1058"></a> |
| <span class="sourceLineNo">1059</span> * @param b number #2<a name="line.1059"></a> |
| <span class="sourceLineNo">1060</span> * @return the midpoint of the 2 numbers<a name="line.1060"></a> |
| <span class="sourceLineNo">1061</span> */<a name="line.1061"></a> |
| <span class="sourceLineNo">1062</span> public BigInteger split2(BigInteger a, BigInteger b) {<a name="line.1062"></a> |
| <span class="sourceLineNo">1063</span> return a.add(b).divide(BigInteger.valueOf(2)).abs();<a name="line.1063"></a> |
| <span class="sourceLineNo">1064</span> }<a name="line.1064"></a> |
| <span class="sourceLineNo">1065</span><a name="line.1065"></a> |
| <span class="sourceLineNo">1066</span> /**<a name="line.1066"></a> |
| <span class="sourceLineNo">1067</span> * Returns an array of bytes corresponding to an array of BigIntegers<a name="line.1067"></a> |
| <span class="sourceLineNo">1068</span> *<a name="line.1068"></a> |
| <span class="sourceLineNo">1069</span> * @param bigIntegers numbers to convert<a name="line.1069"></a> |
| <span class="sourceLineNo">1070</span> * @return bytes corresponding to the bigIntegers<a name="line.1070"></a> |
| <span class="sourceLineNo">1071</span> */<a name="line.1071"></a> |
| <span class="sourceLineNo">1072</span> public byte[][] convertToBytes(BigInteger[] bigIntegers) {<a name="line.1072"></a> |
| <span class="sourceLineNo">1073</span> byte[][] returnBytes = new byte[bigIntegers.length][];<a name="line.1073"></a> |
| <span class="sourceLineNo">1074</span> for (int i = 0; i < bigIntegers.length; i++) {<a name="line.1074"></a> |
| <span class="sourceLineNo">1075</span> returnBytes[i] = convertToByte(bigIntegers[i]);<a name="line.1075"></a> |
| <span class="sourceLineNo">1076</span> }<a name="line.1076"></a> |
| <span class="sourceLineNo">1077</span> return returnBytes;<a name="line.1077"></a> |
| <span class="sourceLineNo">1078</span> }<a name="line.1078"></a> |
| <span class="sourceLineNo">1079</span><a name="line.1079"></a> |
| <span class="sourceLineNo">1080</span> /**<a name="line.1080"></a> |
| <span class="sourceLineNo">1081</span> * Returns the bytes corresponding to the BigInteger<a name="line.1081"></a> |
| <span class="sourceLineNo">1082</span> *<a name="line.1082"></a> |
| <span class="sourceLineNo">1083</span> * @param bigInteger number to convert<a name="line.1083"></a> |
| <span class="sourceLineNo">1084</span> * @param pad padding length<a name="line.1084"></a> |
| <span class="sourceLineNo">1085</span> * @return byte corresponding to input BigInteger<a name="line.1085"></a> |
| <span class="sourceLineNo">1086</span> */<a name="line.1086"></a> |
| <span class="sourceLineNo">1087</span> public byte[] convertToByte(BigInteger bigInteger, int pad) {<a name="line.1087"></a> |
| <span class="sourceLineNo">1088</span> String bigIntegerString = bigInteger.toString(radix);<a name="line.1088"></a> |
| <span class="sourceLineNo">1089</span> bigIntegerString = StringUtils.leftPad(bigIntegerString, pad, '0');<a name="line.1089"></a> |
| <span class="sourceLineNo">1090</span> return Bytes.toBytes(bigIntegerString);<a name="line.1090"></a> |
| <span class="sourceLineNo">1091</span> }<a name="line.1091"></a> |
| <span class="sourceLineNo">1092</span><a name="line.1092"></a> |
| <span class="sourceLineNo">1093</span> /**<a name="line.1093"></a> |
| <span class="sourceLineNo">1094</span> * Returns the bytes corresponding to the BigInteger<a name="line.1094"></a> |
| <span class="sourceLineNo">1095</span> *<a name="line.1095"></a> |
| <span class="sourceLineNo">1096</span> * @param bigInteger number to convert<a name="line.1096"></a> |
| <span class="sourceLineNo">1097</span> * @return corresponding bytes<a name="line.1097"></a> |
| <span class="sourceLineNo">1098</span> */<a name="line.1098"></a> |
| <span class="sourceLineNo">1099</span> public byte[] convertToByte(BigInteger bigInteger) {<a name="line.1099"></a> |
| <span class="sourceLineNo">1100</span> return convertToByte(bigInteger, rowComparisonLength);<a name="line.1100"></a> |
| <span class="sourceLineNo">1101</span> }<a name="line.1101"></a> |
| <span class="sourceLineNo">1102</span><a name="line.1102"></a> |
| <span class="sourceLineNo">1103</span> /**<a name="line.1103"></a> |
| <span class="sourceLineNo">1104</span> * Returns the BigInteger represented by the byte array<a name="line.1104"></a> |
| <span class="sourceLineNo">1105</span> *<a name="line.1105"></a> |
| <span class="sourceLineNo">1106</span> * @param row byte array representing row<a name="line.1106"></a> |
| <span class="sourceLineNo">1107</span> * @return the corresponding BigInteger<a name="line.1107"></a> |
| <span class="sourceLineNo">1108</span> */<a name="line.1108"></a> |
| <span class="sourceLineNo">1109</span> public BigInteger convertToBigInteger(byte[] row) {<a name="line.1109"></a> |
| <span class="sourceLineNo">1110</span> return (row.length > 0) ? new BigInteger(Bytes.toString(row), radix)<a name="line.1110"></a> |
| <span class="sourceLineNo">1111</span> : BigInteger.ZERO;<a name="line.1111"></a> |
| <span class="sourceLineNo">1112</span> }<a name="line.1112"></a> |
| <span class="sourceLineNo">1113</span><a name="line.1113"></a> |
| <span class="sourceLineNo">1114</span> @Override<a name="line.1114"></a> |
| <span class="sourceLineNo">1115</span> public String toString() {<a name="line.1115"></a> |
| <span class="sourceLineNo">1116</span> return this.getClass().getSimpleName() + " [" + rowToStr(firstRow())<a name="line.1116"></a> |
| <span class="sourceLineNo">1117</span> + "," + rowToStr(lastRow()) + "]";<a name="line.1117"></a> |
| <span class="sourceLineNo">1118</span> }<a name="line.1118"></a> |
| <span class="sourceLineNo">1119</span> }<a name="line.1119"></a> |
| <span class="sourceLineNo">1120</span><a name="line.1120"></a> |
| <span class="sourceLineNo">1121</span> /**<a name="line.1121"></a> |
| <span class="sourceLineNo">1122</span> * A SplitAlgorithm that divides the space of possible keys evenly. Useful<a name="line.1122"></a> |
| <span class="sourceLineNo">1123</span> * when the keys are approximately uniform random bytes (e.g. hashes). Rows<a name="line.1123"></a> |
| <span class="sourceLineNo">1124</span> * are raw byte values in the range <b>00 =&gt; FF</b> and are right-padded with<a name="line.1124"></a> |
| <span class="sourceLineNo">1125</span> * zeros to keep the same memcmp() order. This is the natural algorithm to use<a name="line.1125"></a> |
| <span class="sourceLineNo">1126</span> * for a byte[] environment and saves space, but is not necessarily the<a name="line.1126"></a> |
| <span class="sourceLineNo">1127</span> * easiest for readability.<a name="line.1127"></a> |
| <span class="sourceLineNo">1128</span> */<a name="line.1128"></a> |
| <span class="sourceLineNo">1129</span> public static class UniformSplit implements SplitAlgorithm {<a name="line.1129"></a> |
| <span class="sourceLineNo">1130</span> static final byte xFF = (byte) 0xFF;<a name="line.1130"></a> |
| <span class="sourceLineNo">1131</span> byte[] firstRowBytes = ArrayUtils.EMPTY_BYTE_ARRAY;<a name="line.1131"></a> |
| <span class="sourceLineNo">1132</span> byte[] lastRowBytes =<a name="line.1132"></a> |
| <span class="sourceLineNo">1133</span> new byte[] {xFF, xFF, xFF, xFF, xFF, xFF, xFF, xFF};<a name="line.1133"></a> |
| <span class="sourceLineNo">1134</span> @Override<a name="line.1134"></a> |
| <span class="sourceLineNo">1135</span> public byte[] split(byte[] start, byte[] end) {<a name="line.1135"></a> |
| <span class="sourceLineNo">1136</span> return Bytes.split(start, end, 1)[1];<a name="line.1136"></a> |
| <span class="sourceLineNo">1137</span> }<a name="line.1137"></a> |
| <span class="sourceLineNo">1138</span><a name="line.1138"></a> |
| <span class="sourceLineNo">1139</span> @Override<a name="line.1139"></a> |
| <span class="sourceLineNo">1140</span> public byte[][] split(int numRegions) {<a name="line.1140"></a> |
| <span class="sourceLineNo">1141</span> Preconditions.checkArgument(<a name="line.1141"></a> |
| <span class="sourceLineNo">1142</span> Bytes.compareTo(lastRowBytes, firstRowBytes) > 0,<a name="line.1142"></a> |
| <span class="sourceLineNo">1143</span> "last row (%s) is configured less than first row (%s)",<a name="line.1143"></a> |
| <span class="sourceLineNo">1144</span> Bytes.toStringBinary(lastRowBytes),<a name="line.1144"></a> |
| <span class="sourceLineNo">1145</span> Bytes.toStringBinary(firstRowBytes));<a name="line.1145"></a> |
| <span class="sourceLineNo">1146</span><a name="line.1146"></a> |
| <span class="sourceLineNo">1147</span> byte[][] splits = Bytes.split(firstRowBytes, lastRowBytes, true,<a name="line.1147"></a> |
| <span class="sourceLineNo">1148</span> numRegions - 1);<a name="line.1148"></a> |
| <span class="sourceLineNo">1149</span> Preconditions.checkState(splits != null,<a name="line.1149"></a> |
| <span class="sourceLineNo">1150</span> "Could not split region with given user input: " + this);<a name="line.1150"></a> |
| <span class="sourceLineNo">1151</span><a name="line.1151"></a> |
| <span class="sourceLineNo">1152</span> // remove endpoints, which are included in the splits list<a name="line.1152"></a> |
| <span class="sourceLineNo">1153</span><a name="line.1153"></a> |
| <span class="sourceLineNo">1154</span> return splits == null? null: Arrays.copyOfRange(splits, 1, splits.length - 1);<a name="line.1154"></a> |
| <span class="sourceLineNo">1155</span> }<a name="line.1155"></a> |
| <span class="sourceLineNo">1156</span><a name="line.1156"></a> |
| <span class="sourceLineNo">1157</span> @Override<a name="line.1157"></a> |
| <span class="sourceLineNo">1158</span> public byte[][] split(byte[] start, byte[] end, int numSplits, boolean inclusive) {<a name="line.1158"></a> |
| <span class="sourceLineNo">1159</span> if (Arrays.equals(start, HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.1159"></a> |
| <span class="sourceLineNo">1160</span> start = firstRowBytes;<a name="line.1160"></a> |
| <span class="sourceLineNo">1161</span> }<a name="line.1161"></a> |
| <span class="sourceLineNo">1162</span> if (Arrays.equals(end, HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.1162"></a> |
| <span class="sourceLineNo">1163</span> end = lastRowBytes;<a name="line.1163"></a> |
| <span class="sourceLineNo">1164</span> }<a name="line.1164"></a> |
| <span class="sourceLineNo">1165</span> Preconditions.checkArgument(<a name="line.1165"></a> |
| <span class="sourceLineNo">1166</span> Bytes.compareTo(end, start) > 0,<a name="line.1166"></a> |
| <span class="sourceLineNo">1167</span> "last row (%s) is configured less than first row (%s)",<a name="line.1167"></a> |
| <span class="sourceLineNo">1168</span> Bytes.toStringBinary(end),<a name="line.1168"></a> |
| <span class="sourceLineNo">1169</span> Bytes.toStringBinary(start));<a name="line.1169"></a> |
| <span class="sourceLineNo">1170</span><a name="line.1170"></a> |
| <span class="sourceLineNo">1171</span> byte[][] splits = Bytes.split(start, end, true,<a name="line.1171"></a> |
| <span class="sourceLineNo">1172</span> numSplits - 1);<a name="line.1172"></a> |
| <span class="sourceLineNo">1173</span> Preconditions.checkState(splits != null,<a name="line.1173"></a> |
| <span class="sourceLineNo">1174</span> "Could not calculate input splits with given user input: " + this);<a name="line.1174"></a> |
| <span class="sourceLineNo">1175</span> if (inclusive) {<a name="line.1175"></a> |
| <span class="sourceLineNo">1176</span> return splits;<a name="line.1176"></a> |
| <span class="sourceLineNo">1177</span> } else {<a name="line.1177"></a> |
| <span class="sourceLineNo">1178</span> // remove endpoints, which are included in the splits list<a name="line.1178"></a> |
| <span class="sourceLineNo">1179</span> return Arrays.copyOfRange(splits, 1, splits.length - 1);<a name="line.1179"></a> |
| <span class="sourceLineNo">1180</span> }<a name="line.1180"></a> |
| <span class="sourceLineNo">1181</span> }<a name="line.1181"></a> |
| <span class="sourceLineNo">1182</span><a name="line.1182"></a> |
| <span class="sourceLineNo">1183</span> @Override<a name="line.1183"></a> |
| <span class="sourceLineNo">1184</span> public byte[] firstRow() {<a name="line.1184"></a> |
| <span class="sourceLineNo">1185</span> return firstRowBytes;<a name="line.1185"></a> |
| <span class="sourceLineNo">1186</span> }<a name="line.1186"></a> |
| <span class="sourceLineNo">1187</span><a name="line.1187"></a> |
| <span class="sourceLineNo">1188</span> @Override<a name="line.1188"></a> |
| <span class="sourceLineNo">1189</span> public byte[] lastRow() {<a name="line.1189"></a> |
| <span class="sourceLineNo">1190</span> return lastRowBytes;<a name="line.1190"></a> |
| <span class="sourceLineNo">1191</span> }<a name="line.1191"></a> |
| <span class="sourceLineNo">1192</span><a name="line.1192"></a> |
| <span class="sourceLineNo">1193</span> @Override<a name="line.1193"></a> |
| <span class="sourceLineNo">1194</span> public void setFirstRow(String userInput) {<a name="line.1194"></a> |
| <span class="sourceLineNo">1195</span> firstRowBytes = Bytes.toBytesBinary(userInput);<a name="line.1195"></a> |
| <span class="sourceLineNo">1196</span> }<a name="line.1196"></a> |
| <span class="sourceLineNo">1197</span><a name="line.1197"></a> |
| <span class="sourceLineNo">1198</span> @Override<a name="line.1198"></a> |
| <span class="sourceLineNo">1199</span> public void setLastRow(String userInput) {<a name="line.1199"></a> |
| <span class="sourceLineNo">1200</span> lastRowBytes = Bytes.toBytesBinary(userInput);<a name="line.1200"></a> |
| <span class="sourceLineNo">1201</span> }<a name="line.1201"></a> |
| <span class="sourceLineNo">1202</span><a name="line.1202"></a> |
| <span class="sourceLineNo">1203</span><a name="line.1203"></a> |
| <span class="sourceLineNo">1204</span> @Override<a name="line.1204"></a> |
| <span class="sourceLineNo">1205</span> public void setFirstRow(byte[] userInput) {<a name="line.1205"></a> |
| <span class="sourceLineNo">1206</span> firstRowBytes = userInput;<a name="line.1206"></a> |
| <span class="sourceLineNo">1207</span> }<a name="line.1207"></a> |
| <span class="sourceLineNo">1208</span><a name="line.1208"></a> |
| <span class="sourceLineNo">1209</span> @Override<a name="line.1209"></a> |
| <span class="sourceLineNo">1210</span> public void setLastRow(byte[] userInput) {<a name="line.1210"></a> |
| <span class="sourceLineNo">1211</span> lastRowBytes = userInput;<a name="line.1211"></a> |
| <span class="sourceLineNo">1212</span> }<a name="line.1212"></a> |
| <span class="sourceLineNo">1213</span><a name="line.1213"></a> |
| <span class="sourceLineNo">1214</span> @Override<a name="line.1214"></a> |
| <span class="sourceLineNo">1215</span> public byte[] strToRow(String input) {<a name="line.1215"></a> |
| <span class="sourceLineNo">1216</span> return Bytes.toBytesBinary(input);<a name="line.1216"></a> |
| <span class="sourceLineNo">1217</span> }<a name="line.1217"></a> |
| <span class="sourceLineNo">1218</span><a name="line.1218"></a> |
| <span class="sourceLineNo">1219</span> @Override<a name="line.1219"></a> |
| <span class="sourceLineNo">1220</span> public String rowToStr(byte[] row) {<a name="line.1220"></a> |
| <span class="sourceLineNo">1221</span> return Bytes.toStringBinary(row);<a name="line.1221"></a> |
| <span class="sourceLineNo">1222</span> }<a name="line.1222"></a> |
| <span class="sourceLineNo">1223</span><a name="line.1223"></a> |
| <span class="sourceLineNo">1224</span> @Override<a name="line.1224"></a> |
| <span class="sourceLineNo">1225</span> public String separator() {<a name="line.1225"></a> |
| <span class="sourceLineNo">1226</span> return ",";<a name="line.1226"></a> |
| <span class="sourceLineNo">1227</span> }<a name="line.1227"></a> |
| <span class="sourceLineNo">1228</span><a name="line.1228"></a> |
| <span class="sourceLineNo">1229</span> @Override<a name="line.1229"></a> |
| <span class="sourceLineNo">1230</span> public String toString() {<a name="line.1230"></a> |
| <span class="sourceLineNo">1231</span> return this.getClass().getSimpleName() + " [" + rowToStr(firstRow())<a name="line.1231"></a> |
| <span class="sourceLineNo">1232</span> + "," + rowToStr(lastRow()) + "]";<a name="line.1232"></a> |
| <span class="sourceLineNo">1233</span> }<a name="line.1233"></a> |
| <span class="sourceLineNo">1234</span> }<a name="line.1234"></a> |
| <span class="sourceLineNo">1235</span>}<a name="line.1235"></a> |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| </pre> |
| </div> |
| </body> |
| </html> |
