hbase-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From git-site-r...@apache.org
Subject [33/51] [partial] hbase-site git commit: Published site at .
Date Thu, 11 Jan 2018 15:31:22 GMT
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/f183e80f/devapidocs/src-html/org/apache/hadoop/hbase/tool/LoadIncrementalHFiles.BulkHFileVisitor.html
----------------------------------------------------------------------
diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/tool/LoadIncrementalHFiles.BulkHFileVisitor.html b/devapidocs/src-html/org/apache/hadoop/hbase/tool/LoadIncrementalHFiles.BulkHFileVisitor.html
index 1f114e0..01e19b2 100644
--- a/devapidocs/src-html/org/apache/hadoop/hbase/tool/LoadIncrementalHFiles.BulkHFileVisitor.html
+++ b/devapidocs/src-html/org/apache/hadoop/hbase/tool/LoadIncrementalHFiles.BulkHFileVisitor.html
@@ -25,1239 +25,1263 @@
 <span class="sourceLineNo">017</span> */<a name="line.17"></a>
 <span class="sourceLineNo">018</span>package org.apache.hadoop.hbase.tool;<a name="line.18"></a>
 <span class="sourceLineNo">019</span><a name="line.19"></a>
-<span class="sourceLineNo">020</span>import static java.lang.String.format;<a name="line.20"></a>
-<span class="sourceLineNo">021</span><a name="line.21"></a>
-<span class="sourceLineNo">022</span>import java.io.FileNotFoundException;<a name="line.22"></a>
-<span class="sourceLineNo">023</span>import java.io.IOException;<a name="line.23"></a>
-<span class="sourceLineNo">024</span>import java.io.InterruptedIOException;<a name="line.24"></a>
-<span class="sourceLineNo">025</span>import java.nio.ByteBuffer;<a name="line.25"></a>
-<span class="sourceLineNo">026</span>import java.util.ArrayDeque;<a name="line.26"></a>
-<span class="sourceLineNo">027</span>import java.util.ArrayList;<a name="line.27"></a>
-<span class="sourceLineNo">028</span>import java.util.Arrays;<a name="line.28"></a>
-<span class="sourceLineNo">029</span>import java.util.Collection;<a name="line.29"></a>
-<span class="sourceLineNo">030</span>import java.util.Collections;<a name="line.30"></a>
-<span class="sourceLineNo">031</span>import java.util.Deque;<a name="line.31"></a>
-<span class="sourceLineNo">032</span>import java.util.HashMap;<a name="line.32"></a>
-<span class="sourceLineNo">033</span>import java.util.HashSet;<a name="line.33"></a>
-<span class="sourceLineNo">034</span>import java.util.List;<a name="line.34"></a>
-<span class="sourceLineNo">035</span>import java.util.Map;<a name="line.35"></a>
-<span class="sourceLineNo">036</span>import java.util.Map.Entry;<a name="line.36"></a>
-<span class="sourceLineNo">037</span>import java.util.Optional;<a name="line.37"></a>
-<span class="sourceLineNo">038</span>import java.util.Set;<a name="line.38"></a>
-<span class="sourceLineNo">039</span>import java.util.SortedMap;<a name="line.39"></a>
-<span class="sourceLineNo">040</span>import java.util.TreeMap;<a name="line.40"></a>
-<span class="sourceLineNo">041</span>import java.util.UUID;<a name="line.41"></a>
-<span class="sourceLineNo">042</span>import java.util.concurrent.Callable;<a name="line.42"></a>
-<span class="sourceLineNo">043</span>import java.util.concurrent.ExecutionException;<a name="line.43"></a>
-<span class="sourceLineNo">044</span>import java.util.concurrent.ExecutorService;<a name="line.44"></a>
-<span class="sourceLineNo">045</span>import java.util.concurrent.Future;<a name="line.45"></a>
-<span class="sourceLineNo">046</span>import java.util.concurrent.LinkedBlockingQueue;<a name="line.46"></a>
-<span class="sourceLineNo">047</span>import java.util.concurrent.ThreadPoolExecutor;<a name="line.47"></a>
-<span class="sourceLineNo">048</span>import java.util.concurrent.TimeUnit;<a name="line.48"></a>
-<span class="sourceLineNo">049</span>import java.util.concurrent.atomic.AtomicInteger;<a name="line.49"></a>
-<span class="sourceLineNo">050</span>import java.util.stream.Collectors;<a name="line.50"></a>
-<span class="sourceLineNo">051</span><a name="line.51"></a>
-<span class="sourceLineNo">052</span>import org.apache.commons.lang3.mutable.MutableInt;<a name="line.52"></a>
-<span class="sourceLineNo">053</span>import org.apache.hadoop.conf.Configuration;<a name="line.53"></a>
-<span class="sourceLineNo">054</span>import org.apache.hadoop.conf.Configured;<a name="line.54"></a>
-<span class="sourceLineNo">055</span>import org.apache.hadoop.fs.FileStatus;<a name="line.55"></a>
-<span class="sourceLineNo">056</span>import org.apache.hadoop.fs.FileSystem;<a name="line.56"></a>
-<span class="sourceLineNo">057</span>import org.apache.hadoop.fs.Path;<a name="line.57"></a>
-<span class="sourceLineNo">058</span>import org.apache.hadoop.fs.permission.FsPermission;<a name="line.58"></a>
-<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.59"></a>
-<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.HConstants;<a name="line.60"></a>
-<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.TableName;<a name="line.61"></a>
-<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.TableNotFoundException;<a name="line.62"></a>
-<span class="sourceLineNo">063</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.63"></a>
-<span class="sourceLineNo">064</span>import org.slf4j.Logger;<a name="line.64"></a>
-<span class="sourceLineNo">065</span>import org.slf4j.LoggerFactory;<a name="line.65"></a>
-<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.client.Admin;<a name="line.66"></a>
-<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.client.ClientServiceCallable;<a name="line.67"></a>
-<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.68"></a>
-<span class="sourceLineNo">069</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.69"></a>
-<span class="sourceLineNo">070</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.70"></a>
-<span class="sourceLineNo">071</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.71"></a>
-<span class="sourceLineNo">072</span>import org.apache.hadoop.hbase.client.RegionLocator;<a name="line.72"></a>
-<span class="sourceLineNo">073</span>import org.apache.hadoop.hbase.client.RpcRetryingCallerFactory;<a name="line.73"></a>
-<span class="sourceLineNo">074</span>import org.apache.hadoop.hbase.client.SecureBulkLoadClient;<a name="line.74"></a>
-<span class="sourceLineNo">075</span>import org.apache.hadoop.hbase.client.Table;<a name="line.75"></a>
-<span class="sourceLineNo">076</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.76"></a>
-<span class="sourceLineNo">077</span>import org.apache.hadoop.hbase.coprocessor.CoprocessorHost;<a name="line.77"></a>
-<span class="sourceLineNo">078</span>import org.apache.hadoop.hbase.io.HFileLink;<a name="line.78"></a>
-<span class="sourceLineNo">079</span>import org.apache.hadoop.hbase.io.HalfStoreFileReader;<a name="line.79"></a>
-<span class="sourceLineNo">080</span>import org.apache.hadoop.hbase.io.Reference;<a name="line.80"></a>
-<span class="sourceLineNo">081</span>import org.apache.hadoop.hbase.io.compress.Compression.Algorithm;<a name="line.81"></a>
-<span class="sourceLineNo">082</span>import org.apache.hadoop.hbase.io.hfile.CacheConfig;<a name="line.82"></a>
-<span class="sourceLineNo">083</span>import org.apache.hadoop.hbase.io.hfile.HFile;<a name="line.83"></a>
-<span class="sourceLineNo">084</span>import org.apache.hadoop.hbase.io.hfile.HFileContext;<a name="line.84"></a>
-<span class="sourceLineNo">085</span>import org.apache.hadoop.hbase.io.hfile.HFileContextBuilder;<a name="line.85"></a>
-<span class="sourceLineNo">086</span>import org.apache.hadoop.hbase.io.hfile.HFileDataBlockEncoder;<a name="line.86"></a>
-<span class="sourceLineNo">087</span>import org.apache.hadoop.hbase.io.hfile.HFileScanner;<a name="line.87"></a>
-<span class="sourceLineNo">088</span>import org.apache.hadoop.hbase.ipc.RpcControllerFactory;<a name="line.88"></a>
-<span class="sourceLineNo">089</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.89"></a>
-<span class="sourceLineNo">090</span>import org.apache.hadoop.hbase.regionserver.HStore;<a name="line.90"></a>
-<span class="sourceLineNo">091</span>import org.apache.hadoop.hbase.regionserver.StoreFileInfo;<a name="line.91"></a>
-<span class="sourceLineNo">092</span>import org.apache.hadoop.hbase.regionserver.StoreFileWriter;<a name="line.92"></a>
-<span class="sourceLineNo">093</span>import org.apache.hadoop.hbase.security.UserProvider;<a name="line.93"></a>
-<span class="sourceLineNo">094</span>import org.apache.hadoop.hbase.security.token.FsDelegationToken;<a name="line.94"></a>
-<span class="sourceLineNo">095</span>import org.apache.hbase.thirdparty.com.google.common.annotations.VisibleForTesting;<a name="line.95"></a>
-<span class="sourceLineNo">096</span>import org.apache.hbase.thirdparty.com.google.common.collect.HashMultimap;<a name="line.96"></a>
-<span class="sourceLineNo">097</span>import org.apache.hbase.thirdparty.com.google.common.collect.Multimap;<a name="line.97"></a>
-<span class="sourceLineNo">098</span>import org.apache.hbase.thirdparty.com.google.common.collect.Multimaps;<a name="line.98"></a>
-<span class="sourceLineNo">099</span>import org.apache.hbase.thirdparty.com.google.common.util.concurrent.ThreadFactoryBuilder;<a name="line.99"></a>
-<span class="sourceLineNo">100</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.100"></a>
-<span class="sourceLineNo">101</span>import org.apache.hadoop.hbase.util.FSHDFSUtils;<a name="line.101"></a>
-<span class="sourceLineNo">102</span>import org.apache.hadoop.hbase.util.Pair;<a name="line.102"></a>
-<span class="sourceLineNo">103</span>import org.apache.hadoop.util.Tool;<a name="line.103"></a>
-<span class="sourceLineNo">104</span>import org.apache.hadoop.util.ToolRunner;<a name="line.104"></a>
-<span class="sourceLineNo">105</span><a name="line.105"></a>
-<span class="sourceLineNo">106</span>/**<a name="line.106"></a>
-<span class="sourceLineNo">107</span> * Tool to load the output of HFileOutputFormat into an existing table.<a name="line.107"></a>
-<span class="sourceLineNo">108</span> */<a name="line.108"></a>
-<span class="sourceLineNo">109</span>@InterfaceAudience.Public<a name="line.109"></a>
-<span class="sourceLineNo">110</span>public class LoadIncrementalHFiles extends Configured implements Tool {<a name="line.110"></a>
-<span class="sourceLineNo">111</span><a name="line.111"></a>
-<span class="sourceLineNo">112</span>  private static final Logger LOG = LoggerFactory.getLogger(LoadIncrementalHFiles.class);<a name="line.112"></a>
-<span class="sourceLineNo">113</span><a name="line.113"></a>
-<span class="sourceLineNo">114</span>  public static final String NAME = "completebulkload";<a name="line.114"></a>
-<span class="sourceLineNo">115</span>  static final String RETRY_ON_IO_EXCEPTION = "hbase.bulkload.retries.retryOnIOException";<a name="line.115"></a>
-<span class="sourceLineNo">116</span>  public static final String MAX_FILES_PER_REGION_PER_FAMILY =<a name="line.116"></a>
-<span class="sourceLineNo">117</span>      "hbase.mapreduce.bulkload.max.hfiles.perRegion.perFamily";<a name="line.117"></a>
-<span class="sourceLineNo">118</span>  private static final String ASSIGN_SEQ_IDS = "hbase.mapreduce.bulkload.assign.sequenceNumbers";<a name="line.118"></a>
-<span class="sourceLineNo">119</span>  public final static String CREATE_TABLE_CONF_KEY = "create.table";<a name="line.119"></a>
-<span class="sourceLineNo">120</span>  public final static String IGNORE_UNMATCHED_CF_CONF_KEY = "ignore.unmatched.families";<a name="line.120"></a>
-<span class="sourceLineNo">121</span>  public final static String ALWAYS_COPY_FILES = "always.copy.files";<a name="line.121"></a>
-<span class="sourceLineNo">122</span><a name="line.122"></a>
-<span class="sourceLineNo">123</span>  // We use a '.' prefix which is ignored when walking directory trees<a name="line.123"></a>
-<span class="sourceLineNo">124</span>  // above. It is invalid family name.<a name="line.124"></a>
-<span class="sourceLineNo">125</span>  static final String TMP_DIR = ".tmp";<a name="line.125"></a>
-<span class="sourceLineNo">126</span><a name="line.126"></a>
-<span class="sourceLineNo">127</span>  private final int maxFilesPerRegionPerFamily;<a name="line.127"></a>
-<span class="sourceLineNo">128</span>  private final boolean assignSeqIds;<a name="line.128"></a>
-<span class="sourceLineNo">129</span><a name="line.129"></a>
-<span class="sourceLineNo">130</span>  // Source delegation token<a name="line.130"></a>
-<span class="sourceLineNo">131</span>  private final FsDelegationToken fsDelegationToken;<a name="line.131"></a>
-<span class="sourceLineNo">132</span>  private final UserProvider userProvider;<a name="line.132"></a>
-<span class="sourceLineNo">133</span>  private final int nrThreads;<a name="line.133"></a>
-<span class="sourceLineNo">134</span>  private final RpcControllerFactory rpcControllerFactory;<a name="line.134"></a>
-<span class="sourceLineNo">135</span><a name="line.135"></a>
-<span class="sourceLineNo">136</span>  private String bulkToken;<a name="line.136"></a>
-<span class="sourceLineNo">137</span><a name="line.137"></a>
-<span class="sourceLineNo">138</span>  /**<a name="line.138"></a>
-<span class="sourceLineNo">139</span>   * Represents an HFile waiting to be loaded. An queue is used in this class in order to support<a name="line.139"></a>
-<span class="sourceLineNo">140</span>   * the case where a region has split during the process of the load. When this happens, the HFile<a name="line.140"></a>
-<span class="sourceLineNo">141</span>   * is split into two physical parts across the new region boundary, and each part is added back<a name="line.141"></a>
-<span class="sourceLineNo">142</span>   * into the queue. The import process finishes when the queue is empty.<a name="line.142"></a>
-<span class="sourceLineNo">143</span>   */<a name="line.143"></a>
-<span class="sourceLineNo">144</span>  @InterfaceAudience.Public<a name="line.144"></a>
-<span class="sourceLineNo">145</span>  public static class LoadQueueItem {<a name="line.145"></a>
-<span class="sourceLineNo">146</span>    private final byte[] family;<a name="line.146"></a>
-<span class="sourceLineNo">147</span>    private final Path hfilePath;<a name="line.147"></a>
-<span class="sourceLineNo">148</span><a name="line.148"></a>
-<span class="sourceLineNo">149</span>    public LoadQueueItem(byte[] family, Path hfilePath) {<a name="line.149"></a>
-<span class="sourceLineNo">150</span>      this.family = family;<a name="line.150"></a>
-<span class="sourceLineNo">151</span>      this.hfilePath = hfilePath;<a name="line.151"></a>
-<span class="sourceLineNo">152</span>    }<a name="line.152"></a>
-<span class="sourceLineNo">153</span><a name="line.153"></a>
-<span class="sourceLineNo">154</span>    @Override<a name="line.154"></a>
-<span class="sourceLineNo">155</span>    public String toString() {<a name="line.155"></a>
-<span class="sourceLineNo">156</span>      return "family:" + Bytes.toString(family) + " path:" + hfilePath.toString();<a name="line.156"></a>
-<span class="sourceLineNo">157</span>    }<a name="line.157"></a>
-<span class="sourceLineNo">158</span><a name="line.158"></a>
-<span class="sourceLineNo">159</span>    public byte[] getFamily() {<a name="line.159"></a>
-<span class="sourceLineNo">160</span>      return family;<a name="line.160"></a>
-<span class="sourceLineNo">161</span>    }<a name="line.161"></a>
-<span class="sourceLineNo">162</span><a name="line.162"></a>
-<span class="sourceLineNo">163</span>    public Path getFilePath() {<a name="line.163"></a>
-<span class="sourceLineNo">164</span>      return hfilePath;<a name="line.164"></a>
-<span class="sourceLineNo">165</span>    }<a name="line.165"></a>
-<span class="sourceLineNo">166</span>  }<a name="line.166"></a>
-<span class="sourceLineNo">167</span><a name="line.167"></a>
-<span class="sourceLineNo">168</span>  public LoadIncrementalHFiles(Configuration conf) {<a name="line.168"></a>
-<span class="sourceLineNo">169</span>    // make a copy, just to be sure we're not overriding someone else's config<a name="line.169"></a>
-<span class="sourceLineNo">170</span>    super(HBaseConfiguration.create(conf));<a name="line.170"></a>
-<span class="sourceLineNo">171</span>    conf = getConf();<a name="line.171"></a>
-<span class="sourceLineNo">172</span>    // disable blockcache for tool invocation, see HBASE-10500<a name="line.172"></a>
-<span class="sourceLineNo">173</span>    conf.setFloat(HConstants.HFILE_BLOCK_CACHE_SIZE_KEY, 0);<a name="line.173"></a>
-<span class="sourceLineNo">174</span>    userProvider = UserProvider.instantiate(conf);<a name="line.174"></a>
-<span class="sourceLineNo">175</span>    fsDelegationToken = new FsDelegationToken(userProvider, "renewer");<a name="line.175"></a>
-<span class="sourceLineNo">176</span>    assignSeqIds = conf.getBoolean(ASSIGN_SEQ_IDS, true);<a name="line.176"></a>
-<span class="sourceLineNo">177</span>    maxFilesPerRegionPerFamily = conf.getInt(MAX_FILES_PER_REGION_PER_FAMILY, 32);<a name="line.177"></a>
-<span class="sourceLineNo">178</span>    nrThreads = conf.getInt("hbase.loadincremental.threads.max",<a name="line.178"></a>
-<span class="sourceLineNo">179</span>      Runtime.getRuntime().availableProcessors());<a name="line.179"></a>
-<span class="sourceLineNo">180</span>    rpcControllerFactory = new RpcControllerFactory(conf);<a name="line.180"></a>
-<span class="sourceLineNo">181</span>  }<a name="line.181"></a>
-<span class="sourceLineNo">182</span><a name="line.182"></a>
-<span class="sourceLineNo">183</span>  private void usage() {<a name="line.183"></a>
-<span class="sourceLineNo">184</span>    System.err.println("usage: " + NAME + " /path/to/hfileoutputformat-output tablename" + "\n -D" +<a name="line.184"></a>
-<span class="sourceLineNo">185</span>        CREATE_TABLE_CONF_KEY + "=no - can be used to avoid creation of table by this tool\n" +<a name="line.185"></a>
-<span class="sourceLineNo">186</span>        "  Note: if you set this to 'no', then the target table must already exist in HBase\n -D" +<a name="line.186"></a>
-<span class="sourceLineNo">187</span>        IGNORE_UNMATCHED_CF_CONF_KEY + "=yes - can be used to ignore unmatched column families\n" +<a name="line.187"></a>
-<span class="sourceLineNo">188</span>        "\n");<a name="line.188"></a>
-<span class="sourceLineNo">189</span>  }<a name="line.189"></a>
-<span class="sourceLineNo">190</span><a name="line.190"></a>
-<span class="sourceLineNo">191</span>  /**<a name="line.191"></a>
-<span class="sourceLineNo">192</span>   * Prepare a collection of {@link LoadQueueItem} from list of source hfiles contained in the<a name="line.192"></a>
-<span class="sourceLineNo">193</span>   * passed directory and validates whether the prepared queue has all the valid table column<a name="line.193"></a>
-<span class="sourceLineNo">194</span>   * families in it.<a name="line.194"></a>
-<span class="sourceLineNo">195</span>   * @param hfilesDir directory containing list of hfiles to be loaded into the table<a name="line.195"></a>
-<span class="sourceLineNo">196</span>   * @param table table to which hfiles should be loaded<a name="line.196"></a>
-<span class="sourceLineNo">197</span>   * @param queue queue which needs to be loaded into the table<a name="line.197"></a>
-<span class="sourceLineNo">198</span>   * @param validateHFile if true hfiles will be validated for its format<a name="line.198"></a>
-<span class="sourceLineNo">199</span>   * @throws IOException If any I/O or network error occurred<a name="line.199"></a>
-<span class="sourceLineNo">200</span>   */<a name="line.200"></a>
-<span class="sourceLineNo">201</span>  public void prepareHFileQueue(Path hfilesDir, Table table, Deque&lt;LoadQueueItem&gt; queue,<a name="line.201"></a>
-<span class="sourceLineNo">202</span>      boolean validateHFile) throws IOException {<a name="line.202"></a>
-<span class="sourceLineNo">203</span>    prepareHFileQueue(hfilesDir, table, queue, validateHFile, false);<a name="line.203"></a>
-<span class="sourceLineNo">204</span>  }<a name="line.204"></a>
-<span class="sourceLineNo">205</span><a name="line.205"></a>
-<span class="sourceLineNo">206</span>  /**<a name="line.206"></a>
-<span class="sourceLineNo">207</span>   * Prepare a collection of {@link LoadQueueItem} from list of source hfiles contained in the<a name="line.207"></a>
-<span class="sourceLineNo">208</span>   * passed directory and validates whether the prepared queue has all the valid table column<a name="line.208"></a>
-<span class="sourceLineNo">209</span>   * families in it.<a name="line.209"></a>
-<span class="sourceLineNo">210</span>   * @param hfilesDir directory containing list of hfiles to be loaded into the table<a name="line.210"></a>
-<span class="sourceLineNo">211</span>   * @param table table to which hfiles should be loaded<a name="line.211"></a>
-<span class="sourceLineNo">212</span>   * @param queue queue which needs to be loaded into the table<a name="line.212"></a>
-<span class="sourceLineNo">213</span>   * @param validateHFile if true hfiles will be validated for its format<a name="line.213"></a>
-<span class="sourceLineNo">214</span>   * @param silence true to ignore unmatched column families<a name="line.214"></a>
-<span class="sourceLineNo">215</span>   * @throws IOException If any I/O or network error occurred<a name="line.215"></a>
-<span class="sourceLineNo">216</span>   */<a name="line.216"></a>
-<span class="sourceLineNo">217</span>  public void prepareHFileQueue(Path hfilesDir, Table table, Deque&lt;LoadQueueItem&gt; queue,<a name="line.217"></a>
-<span class="sourceLineNo">218</span>      boolean validateHFile, boolean silence) throws IOException {<a name="line.218"></a>
-<span class="sourceLineNo">219</span>    discoverLoadQueue(queue, hfilesDir, validateHFile);<a name="line.219"></a>
-<span class="sourceLineNo">220</span>    validateFamiliesInHFiles(table, queue, silence);<a name="line.220"></a>
-<span class="sourceLineNo">221</span>  }<a name="line.221"></a>
-<span class="sourceLineNo">222</span><a name="line.222"></a>
-<span class="sourceLineNo">223</span>  /**<a name="line.223"></a>
-<span class="sourceLineNo">224</span>   * Prepare a collection of {@link LoadQueueItem} from list of source hfiles contained in the<a name="line.224"></a>
-<span class="sourceLineNo">225</span>   * passed directory and validates whether the prepared queue has all the valid table column<a name="line.225"></a>
-<span class="sourceLineNo">226</span>   * families in it.<a name="line.226"></a>
-<span class="sourceLineNo">227</span>   * @param map map of family to List of hfiles<a name="line.227"></a>
-<span class="sourceLineNo">228</span>   * @param table table to which hfiles should be loaded<a name="line.228"></a>
-<span class="sourceLineNo">229</span>   * @param queue queue which needs to be loaded into the table<a name="line.229"></a>
-<span class="sourceLineNo">230</span>   * @param silence true to ignore unmatched column families<a name="line.230"></a>
-<span class="sourceLineNo">231</span>   * @throws IOException If any I/O or network error occurred<a name="line.231"></a>
-<span class="sourceLineNo">232</span>   */<a name="line.232"></a>
-<span class="sourceLineNo">233</span>  public void prepareHFileQueue(Map&lt;byte[], List&lt;Path&gt;&gt; map, Table table,<a name="line.233"></a>
-<span class="sourceLineNo">234</span>      Deque&lt;LoadQueueItem&gt; queue, boolean silence) throws IOException {<a name="line.234"></a>
-<span class="sourceLineNo">235</span>    populateLoadQueue(queue, map);<a name="line.235"></a>
-<span class="sourceLineNo">236</span>    validateFamiliesInHFiles(table, queue, silence);<a name="line.236"></a>
-<span class="sourceLineNo">237</span>  }<a name="line.237"></a>
-<span class="sourceLineNo">238</span><a name="line.238"></a>
-<span class="sourceLineNo">239</span>  /**<a name="line.239"></a>
-<span class="sourceLineNo">240</span>   * Perform a bulk load of the given directory into the given pre-existing table. This method is<a name="line.240"></a>
-<span class="sourceLineNo">241</span>   * not threadsafe.<a name="line.241"></a>
-<span class="sourceLineNo">242</span>   * @param hfofDir the directory that was provided as the output path of a job using<a name="line.242"></a>
-<span class="sourceLineNo">243</span>   *          HFileOutputFormat<a name="line.243"></a>
-<span class="sourceLineNo">244</span>   * @param admin the Admin<a name="line.244"></a>
-<span class="sourceLineNo">245</span>   * @param table the table to load into<a name="line.245"></a>
-<span class="sourceLineNo">246</span>   * @param regionLocator region locator<a name="line.246"></a>
-<span class="sourceLineNo">247</span>   * @throws TableNotFoundException if table does not yet exist<a name="line.247"></a>
-<span class="sourceLineNo">248</span>   */<a name="line.248"></a>
-<span class="sourceLineNo">249</span>  public Map&lt;LoadQueueItem, ByteBuffer&gt; doBulkLoad(Path hfofDir, final Admin admin, Table table,<a name="line.249"></a>
-<span class="sourceLineNo">250</span>      RegionLocator regionLocator) throws TableNotFoundException, IOException {<a name="line.250"></a>
-<span class="sourceLineNo">251</span>    return doBulkLoad(hfofDir, admin, table, regionLocator, false, false);<a name="line.251"></a>
-<span class="sourceLineNo">252</span>  }<a name="line.252"></a>
-<span class="sourceLineNo">253</span><a name="line.253"></a>
-<span class="sourceLineNo">254</span>  /**<a name="line.254"></a>
-<span class="sourceLineNo">255</span>   * Perform a bulk load of the given directory into the given pre-existing table. This method is<a name="line.255"></a>
-<span class="sourceLineNo">256</span>   * not threadsafe.<a name="line.256"></a>
-<span class="sourceLineNo">257</span>   * @param map map of family to List of hfiles<a name="line.257"></a>
-<span class="sourceLineNo">258</span>   * @param admin the Admin<a name="line.258"></a>
-<span class="sourceLineNo">259</span>   * @param table the table to load into<a name="line.259"></a>
-<span class="sourceLineNo">260</span>   * @param regionLocator region locator<a name="line.260"></a>
-<span class="sourceLineNo">261</span>   * @param silence true to ignore unmatched column families<a name="line.261"></a>
-<span class="sourceLineNo">262</span>   * @param copyFile always copy hfiles if true<a name="line.262"></a>
-<span class="sourceLineNo">263</span>   * @throws TableNotFoundException if table does not yet exist<a name="line.263"></a>
-<span class="sourceLineNo">264</span>   */<a name="line.264"></a>
-<span class="sourceLineNo">265</span>  public Map&lt;LoadQueueItem, ByteBuffer&gt; doBulkLoad(Map&lt;byte[], List&lt;Path&gt;&gt; map, final Admin admin,<a name="line.265"></a>
-<span class="sourceLineNo">266</span>      Table table, RegionLocator regionLocator, boolean silence, boolean copyFile)<a name="line.266"></a>
-<span class="sourceLineNo">267</span>      throws TableNotFoundException, IOException {<a name="line.267"></a>
-<span class="sourceLineNo">268</span>    if (!admin.isTableAvailable(regionLocator.getName())) {<a name="line.268"></a>
-<span class="sourceLineNo">269</span>      throw new TableNotFoundException("Table " + table.getName() + " is not currently available.");<a name="line.269"></a>
-<span class="sourceLineNo">270</span>    }<a name="line.270"></a>
-<span class="sourceLineNo">271</span>    // LQI queue does not need to be threadsafe -- all operations on this queue<a name="line.271"></a>
-<span class="sourceLineNo">272</span>    // happen in this thread<a name="line.272"></a>
-<span class="sourceLineNo">273</span>    Deque&lt;LoadQueueItem&gt; queue = new ArrayDeque&lt;&gt;();<a name="line.273"></a>
-<span class="sourceLineNo">274</span>    ExecutorService pool = null;<a name="line.274"></a>
-<span class="sourceLineNo">275</span>    SecureBulkLoadClient secureClient = null;<a name="line.275"></a>
-<span class="sourceLineNo">276</span>    try {<a name="line.276"></a>
-<span class="sourceLineNo">277</span>      prepareHFileQueue(map, table, queue, silence);<a name="line.277"></a>
-<span class="sourceLineNo">278</span>      if (queue.isEmpty()) {<a name="line.278"></a>
-<span class="sourceLineNo">279</span>        LOG.warn("Bulk load operation did not get any files to load");<a name="line.279"></a>
-<span class="sourceLineNo">280</span>        return Collections.emptyMap();<a name="line.280"></a>
-<span class="sourceLineNo">281</span>      }<a name="line.281"></a>
-<span class="sourceLineNo">282</span>      pool = createExecutorService();<a name="line.282"></a>
-<span class="sourceLineNo">283</span>      secureClient = new SecureBulkLoadClient(table.getConfiguration(), table);<a name="line.283"></a>
-<span class="sourceLineNo">284</span>      return performBulkLoad(admin, table, regionLocator, queue, pool, secureClient, copyFile);<a name="line.284"></a>
-<span class="sourceLineNo">285</span>    } finally {<a name="line.285"></a>
-<span class="sourceLineNo">286</span>      cleanup(admin, queue, pool, secureClient);<a name="line.286"></a>
-<span class="sourceLineNo">287</span>    }<a name="line.287"></a>
-<span class="sourceLineNo">288</span>  }<a name="line.288"></a>
-<span class="sourceLineNo">289</span><a name="line.289"></a>
-<span class="sourceLineNo">290</span>  /**<a name="line.290"></a>
-<span class="sourceLineNo">291</span>   * Perform a bulk load of the given directory into the given pre-existing table. This method is<a name="line.291"></a>
-<span class="sourceLineNo">292</span>   * not threadsafe.<a name="line.292"></a>
-<span class="sourceLineNo">293</span>   * @param hfofDir the directory that was provided as the output path of a job using<a name="line.293"></a>
-<span class="sourceLineNo">294</span>   *          HFileOutputFormat<a name="line.294"></a>
-<span class="sourceLineNo">295</span>   * @param admin the Admin<a name="line.295"></a>
-<span class="sourceLineNo">296</span>   * @param table the table to load into<a name="line.296"></a>
-<span class="sourceLineNo">297</span>   * @param regionLocator region locator<a name="line.297"></a>
-<span class="sourceLineNo">298</span>   * @param silence true to ignore unmatched column families<a name="line.298"></a>
-<span class="sourceLineNo">299</span>   * @param copyFile always copy hfiles if true<a name="line.299"></a>
-<span class="sourceLineNo">300</span>   * @throws TableNotFoundException if table does not yet exist<a name="line.300"></a>
-<span class="sourceLineNo">301</span>   */<a name="line.301"></a>
-<span class="sourceLineNo">302</span>  public Map&lt;LoadQueueItem, ByteBuffer&gt; doBulkLoad(Path hfofDir, final Admin admin, Table table,<a name="line.302"></a>
-<span class="sourceLineNo">303</span>      RegionLocator regionLocator, boolean silence, boolean copyFile)<a name="line.303"></a>
-<span class="sourceLineNo">304</span>      throws TableNotFoundException, IOException {<a name="line.304"></a>
-<span class="sourceLineNo">305</span>    if (!admin.isTableAvailable(regionLocator.getName())) {<a name="line.305"></a>
-<span class="sourceLineNo">306</span>      throw new TableNotFoundException("Table " + table.getName() + " is not currently available.");<a name="line.306"></a>
-<span class="sourceLineNo">307</span>    }<a name="line.307"></a>
-<span class="sourceLineNo">308</span><a name="line.308"></a>
-<span class="sourceLineNo">309</span>    /*<a name="line.309"></a>
-<span class="sourceLineNo">310</span>     * Checking hfile format is a time-consuming operation, we should have an option to skip this<a name="line.310"></a>
-<span class="sourceLineNo">311</span>     * step when bulkloading millions of HFiles. See HBASE-13985.<a name="line.311"></a>
-<span class="sourceLineNo">312</span>     */<a name="line.312"></a>
-<span class="sourceLineNo">313</span>    boolean validateHFile = getConf().getBoolean("hbase.loadincremental.validate.hfile", true);<a name="line.313"></a>
-<span class="sourceLineNo">314</span>    if (!validateHFile) {<a name="line.314"></a>
-<span class="sourceLineNo">315</span>      LOG.warn("You are skipping HFiles validation, it might cause some data loss if files " +<a name="line.315"></a>
-<span class="sourceLineNo">316</span>          "are not correct. If you fail to read data from your table after using this " +<a name="line.316"></a>
-<span class="sourceLineNo">317</span>          "option, consider removing the files and bulkload again without this option. " +<a name="line.317"></a>
-<span class="sourceLineNo">318</span>          "See HBASE-13985");<a name="line.318"></a>
-<span class="sourceLineNo">319</span>    }<a name="line.319"></a>
-<span class="sourceLineNo">320</span>    // LQI queue does not need to be threadsafe -- all operations on this queue<a name="line.320"></a>
-<span class="sourceLineNo">321</span>    // happen in this thread<a name="line.321"></a>
-<span class="sourceLineNo">322</span>    Deque&lt;LoadQueueItem&gt; queue = new ArrayDeque&lt;&gt;();<a name="line.322"></a>
-<span class="sourceLineNo">323</span>    ExecutorService pool = null;<a name="line.323"></a>
-<span class="sourceLineNo">324</span>    SecureBulkLoadClient secureClient = null;<a name="line.324"></a>
-<span class="sourceLineNo">325</span>    try {<a name="line.325"></a>
-<span class="sourceLineNo">326</span>      prepareHFileQueue(hfofDir, table, queue, validateHFile, silence);<a name="line.326"></a>
-<span class="sourceLineNo">327</span><a name="line.327"></a>
-<span class="sourceLineNo">328</span>      if (queue.isEmpty()) {<a name="line.328"></a>
-<span class="sourceLineNo">329</span>        LOG.warn(<a name="line.329"></a>
-<span class="sourceLineNo">330</span>          "Bulk load operation did not find any files to load in " + "directory " + hfofDir != null<a name="line.330"></a>
-<span class="sourceLineNo">331</span>              ? hfofDir.toUri().toString()<a name="line.331"></a>
-<span class="sourceLineNo">332</span>              : "" + ".  Does it contain files in " +<a name="line.332"></a>
-<span class="sourceLineNo">333</span>                  "subdirectories that correspond to column family names?");<a name="line.333"></a>
-<span class="sourceLineNo">334</span>        return Collections.emptyMap();<a name="line.334"></a>
-<span class="sourceLineNo">335</span>      }<a name="line.335"></a>
-<span class="sourceLineNo">336</span>      pool = createExecutorService();<a name="line.336"></a>
-<span class="sourceLineNo">337</span>      secureClient = new SecureBulkLoadClient(table.getConfiguration(), table);<a name="line.337"></a>
-<span class="sourceLineNo">338</span>      return performBulkLoad(admin, table, regionLocator, queue, pool, secureClient, copyFile);<a name="line.338"></a>
-<span class="sourceLineNo">339</span>    } finally {<a name="line.339"></a>
-<span class="sourceLineNo">340</span>      cleanup(admin, queue, pool, secureClient);<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>   * Used by the replication sink to load the hfiles from the source cluster. It does the following,<a name="line.345"></a>
-<span class="sourceLineNo">346</span>   * &lt;ol&gt;<a name="line.346"></a>
-<span class="sourceLineNo">347</span>   * &lt;li&gt;LoadIncrementalHFiles#groupOrSplitPhase(Table, ExecutorService, Deque, Pair)}&lt;/li&gt;<a name="line.347"></a>
-<span class="sourceLineNo">348</span>   * &lt;li&gt;LoadIncrementalHFiles#bulkLoadPhase(Table, Connection, ExecutorService, Deque, Multimap)<a name="line.348"></a>
-<span class="sourceLineNo">349</span>   * &lt;/li&gt;<a name="line.349"></a>
-<span class="sourceLineNo">350</span>   * &lt;/ol&gt;<a name="line.350"></a>
-<span class="sourceLineNo">351</span>   * @param table Table to which these hfiles should be loaded to<a name="line.351"></a>
-<span class="sourceLineNo">352</span>   * @param conn Connection to use<a name="line.352"></a>
-<span class="sourceLineNo">353</span>   * @param queue {@link LoadQueueItem} has hfiles yet to be loaded<a name="line.353"></a>
-<span class="sourceLineNo">354</span>   * @param startEndKeys starting and ending row keys of the region<a name="line.354"></a>
-<span class="sourceLineNo">355</span>   */<a name="line.355"></a>
-<span class="sourceLineNo">356</span>  public void loadHFileQueue(Table table, Connection conn, Deque&lt;LoadQueueItem&gt; queue,<a name="line.356"></a>
-<span class="sourceLineNo">357</span>      Pair&lt;byte[][], byte[][]&gt; startEndKeys) throws IOException {<a name="line.357"></a>
-<span class="sourceLineNo">358</span>    loadHFileQueue(table, conn, queue, startEndKeys, false);<a name="line.358"></a>
-<span class="sourceLineNo">359</span>  }<a name="line.359"></a>
-<span class="sourceLineNo">360</span><a name="line.360"></a>
-<span class="sourceLineNo">361</span>  /**<a name="line.361"></a>
-<span class="sourceLineNo">362</span>   * Used by the replication sink to load the hfiles from the source cluster. It does the following,<a name="line.362"></a>
-<span class="sourceLineNo">363</span>   * &lt;ol&gt;<a name="line.363"></a>
-<span class="sourceLineNo">364</span>   * &lt;li&gt;LoadIncrementalHFiles#groupOrSplitPhase(Table, ExecutorService, Deque, Pair)}&lt;/li&gt;<a name="line.364"></a>
-<span class="sourceLineNo">365</span>   * &lt;li&gt;LoadIncrementalHFiles#bulkLoadPhase(Table, Connection, ExecutorService, Deque, Multimap)<a name="line.365"></a>
-<span class="sourceLineNo">366</span>   * &lt;/li&gt;<a name="line.366"></a>
-<span class="sourceLineNo">367</span>   * &lt;/ol&gt;<a name="line.367"></a>
-<span class="sourceLineNo">368</span>   * @param table Table to which these hfiles should be loaded to<a name="line.368"></a>
-<span class="sourceLineNo">369</span>   * @param conn Connection to use<a name="line.369"></a>
-<span class="sourceLineNo">370</span>   * @param queue {@link LoadQueueItem} has hfiles yet to be loaded<a name="line.370"></a>
-<span class="sourceLineNo">371</span>   * @param startEndKeys starting and ending row keys of the region<a name="line.371"></a>
-<span class="sourceLineNo">372</span>   */<a name="line.372"></a>
-<span class="sourceLineNo">373</span>  public void loadHFileQueue(Table table, Connection conn, Deque&lt;LoadQueueItem&gt; queue,<a name="line.373"></a>
-<span class="sourceLineNo">374</span>      Pair&lt;byte[][], byte[][]&gt; startEndKeys, boolean copyFile) throws IOException {<a name="line.374"></a>
-<span class="sourceLineNo">375</span>    ExecutorService pool = null;<a name="line.375"></a>
-<span class="sourceLineNo">376</span>    try {<a name="line.376"></a>
-<span class="sourceLineNo">377</span>      pool = createExecutorService();<a name="line.377"></a>
-<span class="sourceLineNo">378</span>      Multimap&lt;ByteBuffer, LoadQueueItem&gt; regionGroups =<a name="line.378"></a>
-<span class="sourceLineNo">379</span>          groupOrSplitPhase(table, pool, queue, startEndKeys).getFirst();<a name="line.379"></a>
-<span class="sourceLineNo">380</span>      bulkLoadPhase(table, conn, pool, queue, regionGroups, copyFile, null);<a name="line.380"></a>
-<span class="sourceLineNo">381</span>    } finally {<a name="line.381"></a>
-<span class="sourceLineNo">382</span>      if (pool != null) {<a name="line.382"></a>
-<span class="sourceLineNo">383</span>        pool.shutdown();<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><a name="line.387"></a>
-<span class="sourceLineNo">388</span>  private Map&lt;LoadQueueItem, ByteBuffer&gt; performBulkLoad(Admin admin, Table table,<a name="line.388"></a>
-<span class="sourceLineNo">389</span>      RegionLocator regionLocator, Deque&lt;LoadQueueItem&gt; queue, ExecutorService pool,<a name="line.389"></a>
-<span class="sourceLineNo">390</span>      SecureBulkLoadClient secureClient, boolean copyFile) throws IOException {<a name="line.390"></a>
-<span class="sourceLineNo">391</span>    int count = 0;<a name="line.391"></a>
-<span class="sourceLineNo">392</span><a name="line.392"></a>
-<span class="sourceLineNo">393</span>    if (isSecureBulkLoadEndpointAvailable()) {<a name="line.393"></a>
-<span class="sourceLineNo">394</span>      LOG.warn("SecureBulkLoadEndpoint is deprecated. It will be removed in future releases.");<a name="line.394"></a>
-<span class="sourceLineNo">395</span>      LOG.warn("Secure bulk load has been integrated into HBase core.");<a name="line.395"></a>
-<span class="sourceLineNo">396</span>    }<a name="line.396"></a>
-<span class="sourceLineNo">397</span><a name="line.397"></a>
-<span class="sourceLineNo">398</span>    fsDelegationToken.acquireDelegationToken(queue.peek().getFilePath().getFileSystem(getConf()));<a name="line.398"></a>
-<span class="sourceLineNo">399</span>    bulkToken = secureClient.prepareBulkLoad(admin.getConnection());<a name="line.399"></a>
-<span class="sourceLineNo">400</span>    Pair&lt;Multimap&lt;ByteBuffer, LoadQueueItem&gt;, Set&lt;String&gt;&gt; pair = null;<a name="line.400"></a>
-<span class="sourceLineNo">401</span><a name="line.401"></a>
-<span class="sourceLineNo">402</span>    Map&lt;LoadQueueItem, ByteBuffer&gt; item2RegionMap = new HashMap&lt;&gt;();<a name="line.402"></a>
-<span class="sourceLineNo">403</span>    // Assumes that region splits can happen while this occurs.<a name="line.403"></a>
-<span class="sourceLineNo">404</span>    while (!queue.isEmpty()) {<a name="line.404"></a>
-<span class="sourceLineNo">405</span>      // need to reload split keys each iteration.<a name="line.405"></a>
-<span class="sourceLineNo">406</span>      final Pair&lt;byte[][], byte[][]&gt; startEndKeys = regionLocator.getStartEndKeys();<a name="line.406"></a>
-<span class="sourceLineNo">407</span>      if (count != 0) {<a name="line.407"></a>
-<span class="sourceLineNo">408</span>        LOG.info("Split occurred while grouping HFiles, retry attempt " + +count + " with " +<a name="line.408"></a>
-<span class="sourceLineNo">409</span>            queue.size() + " files remaining to group or split");<a name="line.409"></a>
-<span class="sourceLineNo">410</span>      }<a name="line.410"></a>
-<span class="sourceLineNo">411</span><a name="line.411"></a>
-<span class="sourceLineNo">412</span>      int maxRetries = getConf().getInt(HConstants.BULKLOAD_MAX_RETRIES_NUMBER, 10);<a name="line.412"></a>
-<span class="sourceLineNo">413</span>      maxRetries = Math.max(maxRetries, startEndKeys.getFirst().length + 1);<a name="line.413"></a>
-<span class="sourceLineNo">414</span>      if (maxRetries != 0 &amp;&amp; count &gt;= maxRetries) {<a name="line.414"></a>
-<span class="sourceLineNo">415</span>        throw new IOException(<a name="line.415"></a>
-<span class="sourceLineNo">416</span>            "Retry attempted " + count + " times without completing, bailing out");<a name="line.416"></a>
-<span class="sourceLineNo">417</span>      }<a name="line.417"></a>
-<span class="sourceLineNo">418</span>      count++;<a name="line.418"></a>
-<span class="sourceLineNo">419</span><a name="line.419"></a>
-<span class="sourceLineNo">420</span>      // Using ByteBuffer for byte[] equality semantics<a name="line.420"></a>
-<span class="sourceLineNo">421</span>      pair = groupOrSplitPhase(table, pool, queue, startEndKeys);<a name="line.421"></a>
-<span class="sourceLineNo">422</span>      Multimap&lt;ByteBuffer, LoadQueueItem&gt; regionGroups = pair.getFirst();<a name="line.422"></a>
-<span class="sourceLineNo">423</span><a name="line.423"></a>
-<span class="sourceLineNo">424</span>      if (!checkHFilesCountPerRegionPerFamily(regionGroups)) {<a name="line.424"></a>
-<span class="sourceLineNo">425</span>        // Error is logged inside checkHFilesCountPerRegionPerFamily.<a name="line.425"></a>
-<span class="sourceLineNo">426</span>        throw new IOException("Trying to load more than " + maxFilesPerRegionPerFamily +<a name="line.426"></a>
-<span class="sourceLineNo">427</span>            " hfiles to one family of one region");<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>      bulkLoadPhase(table, admin.getConnection(), pool, queue, regionGroups, copyFile,<a name="line.430"></a>
-<span class="sourceLineNo">431</span>        item2RegionMap);<a name="line.431"></a>
+<span class="sourceLineNo">020</span>import java.io.FileNotFoundException;<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.io.InterruptedIOException;<a name="line.22"></a>
+<span class="sourceLineNo">023</span>import java.nio.ByteBuffer;<a name="line.23"></a>
+<span class="sourceLineNo">024</span>import java.util.ArrayDeque;<a name="line.24"></a>
+<span class="sourceLineNo">025</span>import java.util.ArrayList;<a name="line.25"></a>
+<span class="sourceLineNo">026</span>import java.util.Arrays;<a name="line.26"></a>
+<span class="sourceLineNo">027</span>import java.util.Collection;<a name="line.27"></a>
+<span class="sourceLineNo">028</span>import java.util.Collections;<a name="line.28"></a>
+<span class="sourceLineNo">029</span>import java.util.Deque;<a name="line.29"></a>
+<span class="sourceLineNo">030</span>import java.util.HashMap;<a name="line.30"></a>
+<span class="sourceLineNo">031</span>import java.util.HashSet;<a name="line.31"></a>
+<span class="sourceLineNo">032</span>import java.util.List;<a name="line.32"></a>
+<span class="sourceLineNo">033</span>import java.util.Map;<a name="line.33"></a>
+<span class="sourceLineNo">034</span>import java.util.Map.Entry;<a name="line.34"></a>
+<span class="sourceLineNo">035</span>import java.util.Optional;<a name="line.35"></a>
+<span class="sourceLineNo">036</span>import java.util.Set;<a name="line.36"></a>
+<span class="sourceLineNo">037</span>import java.util.SortedMap;<a name="line.37"></a>
+<span class="sourceLineNo">038</span>import java.util.TreeMap;<a name="line.38"></a>
+<span class="sourceLineNo">039</span>import java.util.UUID;<a name="line.39"></a>
+<span class="sourceLineNo">040</span>import java.util.concurrent.Callable;<a name="line.40"></a>
+<span class="sourceLineNo">041</span>import java.util.concurrent.ExecutionException;<a name="line.41"></a>
+<span class="sourceLineNo">042</span>import java.util.concurrent.ExecutorService;<a name="line.42"></a>
+<span class="sourceLineNo">043</span>import java.util.concurrent.Future;<a name="line.43"></a>
+<span class="sourceLineNo">044</span>import java.util.concurrent.LinkedBlockingQueue;<a name="line.44"></a>
+<span class="sourceLineNo">045</span>import java.util.concurrent.ThreadPoolExecutor;<a name="line.45"></a>
+<span class="sourceLineNo">046</span>import java.util.concurrent.TimeUnit;<a name="line.46"></a>
+<span class="sourceLineNo">047</span>import java.util.concurrent.atomic.AtomicInteger;<a name="line.47"></a>
+<span class="sourceLineNo">048</span>import java.util.stream.Collectors;<a name="line.48"></a>
+<span class="sourceLineNo">049</span>import static java.lang.String.format;<a name="line.49"></a>
+<span class="sourceLineNo">050</span>import org.apache.commons.lang3.mutable.MutableInt;<a name="line.50"></a>
+<span class="sourceLineNo">051</span>import org.apache.hadoop.conf.Configuration;<a name="line.51"></a>
+<span class="sourceLineNo">052</span>import org.apache.hadoop.conf.Configured;<a name="line.52"></a>
+<span class="sourceLineNo">053</span>import org.apache.hadoop.fs.FileStatus;<a name="line.53"></a>
+<span class="sourceLineNo">054</span>import org.apache.hadoop.fs.FileSystem;<a name="line.54"></a>
+<span class="sourceLineNo">055</span>import org.apache.hadoop.fs.Path;<a name="line.55"></a>
+<span class="sourceLineNo">056</span>import org.apache.hadoop.fs.permission.FsPermission;<a name="line.56"></a>
+<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.57"></a>
+<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.HConstants;<a name="line.58"></a>
+<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.TableName;<a name="line.59"></a>
+<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.TableNotFoundException;<a name="line.60"></a>
+<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.client.Admin;<a name="line.61"></a>
+<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.client.ClientServiceCallable;<a name="line.62"></a>
+<span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.63"></a>
+<span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.64"></a>
+<span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.65"></a>
+<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.66"></a>
+<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.client.RegionLocator;<a name="line.67"></a>
+<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.client.RpcRetryingCallerFactory;<a name="line.68"></a>
+<span class="sourceLineNo">069</span>import org.apache.hadoop.hbase.client.SecureBulkLoadClient;<a name="line.69"></a>
+<span class="sourceLineNo">070</span>import org.apache.hadoop.hbase.client.Table;<a name="line.70"></a>
+<span class="sourceLineNo">071</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.71"></a>
+<span class="sourceLineNo">072</span>import org.apache.hadoop.hbase.coprocessor.CoprocessorHost;<a name="line.72"></a>
+<span class="sourceLineNo">073</span>import org.apache.hadoop.hbase.io.HFileLink;<a name="line.73"></a>
+<span class="sourceLineNo">074</span>import org.apache.hadoop.hbase.io.HalfStoreFileReader;<a name="line.74"></a>
+<span class="sourceLineNo">075</span>import org.apache.hadoop.hbase.io.Reference;<a name="line.75"></a>
+<span class="sourceLineNo">076</span>import org.apache.hadoop.hbase.io.compress.Compression.Algorithm;<a name="line.76"></a>
+<span class="sourceLineNo">077</span>import org.apache.hadoop.hbase.io.hfile.CacheConfig;<a name="line.77"></a>
+<span class="sourceLineNo">078</span>import org.apache.hadoop.hbase.io.hfile.HFile;<a name="line.78"></a>
+<span class="sourceLineNo">079</span>import org.apache.hadoop.hbase.io.hfile.HFileContext;<a name="line.79"></a>
+<span class="sourceLineNo">080</span>import org.apache.hadoop.hbase.io.hfile.HFileContextBuilder;<a name="line.80"></a>
+<span class="sourceLineNo">081</span>import org.apache.hadoop.hbase.io.hfile.HFileDataBlockEncoder;<a name="line.81"></a>
+<span class="sourceLineNo">082</span>import org.apache.hadoop.hbase.io.hfile.HFileScanner;<a name="line.82"></a>
+<span class="sourceLineNo">083</span>import org.apache.hadoop.hbase.ipc.RpcControllerFactory;<a name="line.83"></a>
+<span class="sourceLineNo">084</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.84"></a>
+<span class="sourceLineNo">085</span>import org.apache.hadoop.hbase.regionserver.HStore;<a name="line.85"></a>
+<span class="sourceLineNo">086</span>import org.apache.hadoop.hbase.regionserver.StoreFileInfo;<a name="line.86"></a>
+<span class="sourceLineNo">087</span>import org.apache.hadoop.hbase.regionserver.StoreFileWriter;<a name="line.87"></a>
+<span class="sourceLineNo">088</span>import org.apache.hadoop.hbase.security.UserProvider;<a name="line.88"></a>
+<span class="sourceLineNo">089</span>import org.apache.hadoop.hbase.security.token.FsDelegationToken;<a name="line.89"></a>
+<span class="sourceLineNo">090</span>import org.apache.hbase.thirdparty.com.google.common.annotations.VisibleForTesting;<a name="line.90"></a>
+<span class="sourceLineNo">091</span>import org.apache.hbase.thirdparty.com.google.common.collect.HashMultimap;<a name="line.91"></a>
+<span class="sourceLineNo">092</span>import org.apache.hbase.thirdparty.com.google.common.collect.Lists;<a name="line.92"></a>
+<span class="sourceLineNo">093</span>import org.apache.hbase.thirdparty.com.google.common.collect.Maps;<a name="line.93"></a>
+<span class="sourceLineNo">094</span>import org.apache.hbase.thirdparty.com.google.common.collect.Multimap;<a name="line.94"></a>
+<span class="sourceLineNo">095</span>import org.apache.hbase.thirdparty.com.google.common.collect.Multimaps;<a name="line.95"></a>
+<span class="sourceLineNo">096</span>import org.apache.hbase.thirdparty.com.google.common.util.concurrent.ThreadFactoryBuilder;<a name="line.96"></a>
+<span class="sourceLineNo">097</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.97"></a>
+<span class="sourceLineNo">098</span>import org.apache.hadoop.hbase.util.FSHDFSUtils;<a name="line.98"></a>
+<span class="sourceLineNo">099</span>import org.apache.hadoop.hbase.util.FSVisitor;<a name="line.99"></a>
+<span class="sourceLineNo">100</span>import org.apache.hadoop.hbase.util.Pair;<a name="line.100"></a>
+<span class="sourceLineNo">101</span>import org.apache.hadoop.util.Tool;<a name="line.101"></a>
+<span class="sourceLineNo">102</span>import org.apache.hadoop.util.ToolRunner;<a name="line.102"></a>
+<span class="sourceLineNo">103</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.103"></a>
+<span class="sourceLineNo">104</span>import org.slf4j.Logger;<a name="line.104"></a>
+<span class="sourceLineNo">105</span>import org.slf4j.LoggerFactory;<a name="line.105"></a>
+<span class="sourceLineNo">106</span><a name="line.106"></a>
+<span class="sourceLineNo">107</span>/**<a name="line.107"></a>
+<span class="sourceLineNo">108</span> * Tool to load the output of HFileOutputFormat into an existing table.<a name="line.108"></a>
+<span class="sourceLineNo">109</span> */<a name="line.109"></a>
+<span class="sourceLineNo">110</span>@InterfaceAudience.Public<a name="line.110"></a>
+<span class="sourceLineNo">111</span>public class LoadIncrementalHFiles extends Configured implements Tool {<a name="line.111"></a>
+<span class="sourceLineNo">112</span><a name="line.112"></a>
+<span class="sourceLineNo">113</span>  private static final Logger LOG = LoggerFactory.getLogger(LoadIncrementalHFiles.class);<a name="line.113"></a>
+<span class="sourceLineNo">114</span><a name="line.114"></a>
+<span class="sourceLineNo">115</span>  public static final String NAME = "completebulkload";<a name="line.115"></a>
+<span class="sourceLineNo">116</span>  static final String RETRY_ON_IO_EXCEPTION = "hbase.bulkload.retries.retryOnIOException";<a name="line.116"></a>
+<span class="sourceLineNo">117</span>  public static final String MAX_FILES_PER_REGION_PER_FAMILY =<a name="line.117"></a>
+<span class="sourceLineNo">118</span>      "hbase.mapreduce.bulkload.max.hfiles.perRegion.perFamily";<a name="line.118"></a>
+<span class="sourceLineNo">119</span>  private static final String ASSIGN_SEQ_IDS = "hbase.mapreduce.bulkload.assign.sequenceNumbers";<a name="line.119"></a>
+<span class="sourceLineNo">120</span>  public final static String CREATE_TABLE_CONF_KEY = "create.table";<a name="line.120"></a>
+<span class="sourceLineNo">121</span>  public final static String IGNORE_UNMATCHED_CF_CONF_KEY = "ignore.unmatched.families";<a name="line.121"></a>
+<span class="sourceLineNo">122</span>  public final static String ALWAYS_COPY_FILES = "always.copy.files";<a name="line.122"></a>
+<span class="sourceLineNo">123</span><a name="line.123"></a>
+<span class="sourceLineNo">124</span>  // We use a '.' prefix which is ignored when walking directory trees<a name="line.124"></a>
+<span class="sourceLineNo">125</span>  // above. It is invalid family name.<a name="line.125"></a>
+<span class="sourceLineNo">126</span>  static final String TMP_DIR = ".tmp";<a name="line.126"></a>
+<span class="sourceLineNo">127</span><a name="line.127"></a>
+<span class="sourceLineNo">128</span>  private final int maxFilesPerRegionPerFamily;<a name="line.128"></a>
+<span class="sourceLineNo">129</span>  private final boolean assignSeqIds;<a name="line.129"></a>
+<span class="sourceLineNo">130</span><a name="line.130"></a>
+<span class="sourceLineNo">131</span>  // Source delegation token<a name="line.131"></a>
+<span class="sourceLineNo">132</span>  private final FsDelegationToken fsDelegationToken;<a name="line.132"></a>
+<span class="sourceLineNo">133</span>  private final UserProvider userProvider;<a name="line.133"></a>
+<span class="sourceLineNo">134</span>  private final int nrThreads;<a name="line.134"></a>
+<span class="sourceLineNo">135</span>  private final RpcControllerFactory rpcControllerFactory;<a name="line.135"></a>
+<span class="sourceLineNo">136</span><a name="line.136"></a>
+<span class="sourceLineNo">137</span>  private String bulkToken;<a name="line.137"></a>
+<span class="sourceLineNo">138</span><a name="line.138"></a>
+<span class="sourceLineNo">139</span>  /**<a name="line.139"></a>
+<span class="sourceLineNo">140</span>   * Represents an HFile waiting to be loaded. An queue is used in this class in order to support<a name="line.140"></a>
+<span class="sourceLineNo">141</span>   * the case where a region has split during the process of the load. When this happens, the HFile<a name="line.141"></a>
+<span class="sourceLineNo">142</span>   * is split into two physical parts across the new region boundary, and each part is added back<a name="line.142"></a>
+<span class="sourceLineNo">143</span>   * into the queue. The import process finishes when the queue is empty.<a name="line.143"></a>
+<span class="sourceLineNo">144</span>   */<a name="line.144"></a>
+<span class="sourceLineNo">145</span>  @InterfaceAudience.Public<a name="line.145"></a>
+<span class="sourceLineNo">146</span>  public static class LoadQueueItem {<a name="line.146"></a>
+<span class="sourceLineNo">147</span>    private final byte[] family;<a name="line.147"></a>
+<span class="sourceLineNo">148</span>    private final Path hfilePath;<a name="line.148"></a>
+<span class="sourceLineNo">149</span><a name="line.149"></a>
+<span class="sourceLineNo">150</span>    public LoadQueueItem(byte[] family, Path hfilePath) {<a name="line.150"></a>
+<span class="sourceLineNo">151</span>      this.family = family;<a name="line.151"></a>
+<span class="sourceLineNo">152</span>      this.hfilePath = hfilePath;<a name="line.152"></a>
+<span class="sourceLineNo">153</span>    }<a name="line.153"></a>
+<span class="sourceLineNo">154</span><a name="line.154"></a>
+<span class="sourceLineNo">155</span>    @Override<a name="line.155"></a>
+<span class="sourceLineNo">156</span>    public String toString() {<a name="line.156"></a>
+<span class="sourceLineNo">157</span>      return "family:" + Bytes.toString(family) + " path:" + hfilePath.toString();<a name="line.157"></a>
+<span class="sourceLineNo">158</span>    }<a name="line.158"></a>
+<span class="sourceLineNo">159</span><a name="line.159"></a>
+<span class="sourceLineNo">160</span>    public byte[] getFamily() {<a name="line.160"></a>
+<span class="sourceLineNo">161</span>      return family;<a name="line.161"></a>
+<span class="sourceLineNo">162</span>    }<a name="line.162"></a>
+<span class="sourceLineNo">163</span><a name="line.163"></a>
+<span class="sourceLineNo">164</span>    public Path getFilePath() {<a name="line.164"></a>
+<span class="sourceLineNo">165</span>      return hfilePath;<a name="line.165"></a>
+<span class="sourceLineNo">166</span>    }<a name="line.166"></a>
+<span class="sourceLineNo">167</span>  }<a name="line.167"></a>
+<span class="sourceLineNo">168</span><a name="line.168"></a>
+<span class="sourceLineNo">169</span>  public LoadIncrementalHFiles(Configuration conf) {<a name="line.169"></a>
+<span class="sourceLineNo">170</span>    // make a copy, just to be sure we're not overriding someone else's config<a name="line.170"></a>
+<span class="sourceLineNo">171</span>    super(HBaseConfiguration.create(conf));<a name="line.171"></a>
+<span class="sourceLineNo">172</span>    conf = getConf();<a name="line.172"></a>
+<span class="sourceLineNo">173</span>    // disable blockcache for tool invocation, see HBASE-10500<a name="line.173"></a>
+<span class="sourceLineNo">174</span>    conf.setFloat(HConstants.HFILE_BLOCK_CACHE_SIZE_KEY, 0);<a name="line.174"></a>
+<span class="sourceLineNo">175</span>    userProvider = UserProvider.instantiate(conf);<a name="line.175"></a>
+<span class="sourceLineNo">176</span>    fsDelegationToken = new FsDelegationToken(userProvider, "renewer");<a name="line.176"></a>
+<span class="sourceLineNo">177</span>    assignSeqIds = conf.getBoolean(ASSIGN_SEQ_IDS, true);<a name="line.177"></a>
+<span class="sourceLineNo">178</span>    maxFilesPerRegionPerFamily = conf.getInt(MAX_FILES_PER_REGION_PER_FAMILY, 32);<a name="line.178"></a>
+<span class="sourceLineNo">179</span>    nrThreads = conf.getInt("hbase.loadincremental.threads.max",<a name="line.179"></a>
+<span class="sourceLineNo">180</span>      Runtime.getRuntime().availableProcessors());<a name="line.180"></a>
+<span class="sourceLineNo">181</span>    rpcControllerFactory = new RpcControllerFactory(conf);<a name="line.181"></a>
+<span class="sourceLineNo">182</span>  }<a name="line.182"></a>
+<span class="sourceLineNo">183</span><a name="line.183"></a>
+<span class="sourceLineNo">184</span>  private void usage() {<a name="line.184"></a>
+<span class="sourceLineNo">185</span>    System.err.println("usage: " + NAME + " /path/to/hfileoutputformat-output tablename -loadTable"<a name="line.185"></a>
+<span class="sourceLineNo">186</span>        + "\n -D" + CREATE_TABLE_CONF_KEY + "=no - can be used to avoid creation of table by "<a name="line.186"></a>
+<span class="sourceLineNo">187</span>        + "this tool\n  Note: if you set this to 'no', then the target table must already exist "<a name="line.187"></a>
+<span class="sourceLineNo">188</span>        + "in HBase\n -loadTable implies your baseDirectory to store file has a depth of 3 ,you"<a name="line.188"></a>
+<span class="sourceLineNo">189</span>        + " must have an existing table\n-D" + IGNORE_UNMATCHED_CF_CONF_KEY + "=yes - can be used "<a name="line.189"></a>
+<span class="sourceLineNo">190</span>        + "to ignore unmatched column families\n" +<a name="line.190"></a>
+<span class="sourceLineNo">191</span>        "\n");<a name="line.191"></a>
+<span class="sourceLineNo">192</span>  }<a name="line.192"></a>
+<span class="sourceLineNo">193</span><a name="line.193"></a>
+<span class="sourceLineNo">194</span>  /**<a name="line.194"></a>
+<span class="sourceLineNo">195</span>   * Prepare a collection of {@link LoadQueueItem} from list of source hfiles contained in the<a name="line.195"></a>
+<span class="sourceLineNo">196</span>   * passed directory and validates whether the prepared queue has all the valid table column<a name="line.196"></a>
+<span class="sourceLineNo">197</span>   * families in it.<a name="line.197"></a>
+<span class="sourceLineNo">198</span>   * @param hfilesDir directory containing list of hfiles to be loaded into the table<a name="line.198"></a>
+<span class="sourceLineNo">199</span>   * @param table table to which hfiles should be loaded<a name="line.199"></a>
+<span class="sourceLineNo">200</span>   * @param queue queue which needs to be loaded into the table<a name="line.200"></a>
+<span class="sourceLineNo">201</span>   * @param validateHFile if true hfiles will be validated for its format<a name="line.201"></a>
+<span class="sourceLineNo">202</span>   * @throws IOException If any I/O or network error occurred<a name="line.202"></a>
+<span class="sourceLineNo">203</span>   */<a name="line.203"></a>
+<span class="sourceLineNo">204</span>  public void prepareHFileQueue(Path hfilesDir, Table table, Deque&lt;LoadQueueItem&gt; queue,<a name="line.204"></a>
+<span class="sourceLineNo">205</span>      boolean validateHFile) throws IOException {<a name="line.205"></a>
+<span class="sourceLineNo">206</span>    prepareHFileQueue(hfilesDir, table, queue, validateHFile, false);<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>  /**<a name="line.209"></a>
+<span class="sourceLineNo">210</span>   * Prepare a collection of {@link LoadQueueItem} from list of source hfiles contained in the<a name="line.210"></a>
+<span class="sourceLineNo">211</span>   * passed directory and validates whether the prepared queue has all the valid table column<a name="line.211"></a>
+<span class="sourceLineNo">212</span>   * families in it.<a name="line.212"></a>
+<span class="sourceLineNo">213</span>   * @param hfilesDir directory containing list of hfiles to be loaded into the table<a name="line.213"></a>
+<span class="sourceLineNo">214</span>   * @param table table to which hfiles should be loaded<a name="line.214"></a>
+<span class="sourceLineNo">215</span>   * @param queue queue which needs to be loaded into the table<a name="line.215"></a>
+<span class="sourceLineNo">216</span>   * @param validateHFile if true hfiles will be validated for its format<a name="line.216"></a>
+<span class="sourceLineNo">217</span>   * @param silence true to ignore unmatched column families<a name="line.217"></a>
+<span class="sourceLineNo">218</span>   * @throws IOException If any I/O or network error occurred<a name="line.218"></a>
+<span class="sourceLineNo">219</span>   */<a name="line.219"></a>
+<span class="sourceLineNo">220</span>  public void prepareHFileQueue(Path hfilesDir, Table table, Deque&lt;LoadQueueItem&gt; queue,<a name="line.220"></a>
+<span class="sourceLineNo">221</span>      boolean validateHFile, boolean silence) throws IOException {<a name="line.221"></a>
+<span class="sourceLineNo">222</span>    discoverLoadQueue(queue, hfilesDir, validateHFile);<a name="line.222"></a>
+<span class="sourceLineNo">223</span>    validateFamiliesInHFiles(table, queue, silence);<a name="line.223"></a>
+<span class="sourceLineNo">224</span>  }<a name="line.224"></a>
+<span class="sourceLineNo">225</span><a name="line.225"></a>
+<span class="sourceLineNo">226</span>  /**<a name="line.226"></a>
+<span class="sourceLineNo">227</span>   * Prepare a collection of {@link LoadQueueItem} from list of source hfiles contained in the<a name="line.227"></a>
+<span class="sourceLineNo">228</span>   * passed directory and validates whether the prepared queue has all the valid table column<a name="line.228"></a>
+<span class="sourceLineNo">229</span>   * families in it.<a name="line.229"></a>
+<span class="sourceLineNo">230</span>   * @param map map of family to List of hfiles<a name="line.230"></a>
+<span class="sourceLineNo">231</span>   * @param table table to which hfiles should be loaded<a name="line.231"></a>
+<span class="sourceLineNo">232</span>   * @param queue queue which needs to be loaded into the table<a name="line.232"></a>
+<span class="sourceLineNo">233</span>   * @param silence true to ignore unmatched column families<a name="line.233"></a>
+<span class="sourceLineNo">234</span>   * @throws IOException If any I/O or network error occurred<a name="line.234"></a>
+<span class="sourceLineNo">235</span>   */<a name="line.235"></a>
+<span class="sourceLineNo">236</span>  public void prepareHFileQueue(Map&lt;byte[], List&lt;Path&gt;&gt; map, Table table,<a name="line.236"></a>
+<span class="sourceLineNo">237</span>      Deque&lt;LoadQueueItem&gt; queue, boolean silence) throws IOException {<a name="line.237"></a>
+<span class="sourceLineNo">238</span>    populateLoadQueue(queue, map);<a name="line.238"></a>
+<span class="sourceLineNo">239</span>    validateFamiliesInHFiles(table, queue, silence);<a name="line.239"></a>
+<span class="sourceLineNo">240</span>  }<a name="line.240"></a>
+<span class="sourceLineNo">241</span><a name="line.241"></a>
+<span class="sourceLineNo">242</span>  /**<a name="line.242"></a>
+<span class="sourceLineNo">243</span>   * Perform a bulk load of the given directory into the given pre-existing table. This method is<a name="line.243"></a>
+<span class="sourceLineNo">244</span>   * not threadsafe.<a name="line.244"></a>
+<span class="sourceLineNo">245</span>   * @param hfofDir the directory that was provided as the output path of a job using<a name="line.245"></a>
+<span class="sourceLineNo">246</span>   *          HFileOutputFormat<a name="line.246"></a>
+<span class="sourceLineNo">247</span>   * @param admin the Admin<a name="line.247"></a>
+<span class="sourceLineNo">248</span>   * @param table the table to load into<a name="line.248"></a>
+<span class="sourceLineNo">249</span>   * @param regionLocator region locator<a name="line.249"></a>
+<span class="sourceLineNo">250</span>   * @throws TableNotFoundException if table does not yet exist<a name="line.250"></a>
+<span class="sourceLineNo">251</span>   */<a name="line.251"></a>
+<span class="sourceLineNo">252</span>  public Map&lt;LoadQueueItem, ByteBuffer&gt; doBulkLoad(Path hfofDir, final Admin admin, Table table,<a name="line.252"></a>
+<span class="sourceLineNo">253</span>      RegionLocator regionLocator) throws TableNotFoundException, IOException {<a name="line.253"></a>
+<span class="sourceLineNo">254</span>    return doBulkLoad(hfofDir, admin, table, regionLocator, false, false);<a name="line.254"></a>
+<span class="sourceLineNo">255</span>  }<a name="line.255"></a>
+<span class="sourceLineNo">256</span><a name="line.256"></a>
+<span class="sourceLineNo">257</span>  /**<a name="line.257"></a>
+<span class="sourceLineNo">258</span>   * Perform a bulk load of the given directory into the given pre-existing table. This method is<a name="line.258"></a>
+<span class="sourceLineNo">259</span>   * not threadsafe.<a name="line.259"></a>
+<span class="sourceLineNo">260</span>   * @param map map of family to List of hfiles<a name="line.260"></a>
+<span class="sourceLineNo">261</span>   * @param admin the Admin<a name="line.261"></a>
+<span class="sourceLineNo">262</span>   * @param table the table to load into<a name="line.262"></a>
+<span class="sourceLineNo">263</span>   * @param regionLocator region locator<a name="line.263"></a>
+<span class="sourceLineNo">264</span>   * @param silence true to ignore unmatched column families<a name="line.264"></a>
+<span class="sourceLineNo">265</span>   * @param copyFile always copy hfiles if true<a name="line.265"></a>
+<span class="sourceLineNo">266</span>   * @throws TableNotFoundException if table does not yet exist<a name="line.266"></a>
+<span class="sourceLineNo">267</span>   */<a name="line.267"></a>
+<span class="sourceLineNo">268</span>  public Map&lt;LoadQueueItem, ByteBuffer&gt; doBulkLoad(Map&lt;byte[], List&lt;Path&gt;&gt; map, final Admin admin,<a name="line.268"></a>
+<span class="sourceLineNo">269</span>      Table table, RegionLocator regionLocator, boolean silence, boolean copyFile)<a name="line.269"></a>
+<span class="sourceLineNo">270</span>      throws TableNotFoundException, IOException {<a name="line.270"></a>
+<span class="sourceLineNo">271</span>    if (!admin.isTableAvailable(regionLocator.getName())) {<a name="line.271"></a>
+<span class="sourceLineNo">272</span>      throw new TableNotFoundException("Table " + table.getName() + " is not currently available.");<a name="line.272"></a>
+<span class="sourceLineNo">273</span>    }<a name="line.273"></a>
+<span class="sourceLineNo">274</span>    // LQI queue does not need to be threadsafe -- all operations on this queue<a name="line.274"></a>
+<span class="sourceLineNo">275</span>    // happen in this thread<a name="line.275"></a>
+<span class="sourceLineNo">276</span>    Deque&lt;LoadQueueItem&gt; queue = new ArrayDeque&lt;&gt;();<a name="line.276"></a>
+<span class="sourceLineNo">277</span>    ExecutorService pool = null;<a name="line.277"></a>
+<span class="sourceLineNo">278</span>    SecureBulkLoadClient secureClient = null;<a name="line.278"></a>
+<span class="sourceLineNo">279</span>    try {<a name="line.279"></a>
+<span class="sourceLineNo">280</span>      prepareHFileQueue(map, table, queue, silence);<a name="line.280"></a>
+<span class="sourceLineNo">281</span>      if (queue.isEmpty()) {<a name="line.281"></a>
+<span class="sourceLineNo">282</span>        LOG.warn("Bulk load operation did not get any files to load");<a name="line.282"></a>
+<span class="sourceLineNo">283</span>        return Collections.emptyMap();<a name="line.283"></a>
+<span class="sourceLineNo">284</span>      }<a name="line.284"></a>
+<span class="sourceLineNo">285</span>      pool = createExecutorService();<a name="line.285"></a>
+<span class="sourceLineNo">286</span>      secureClient = new SecureBulkLoadClient(table.getConfiguration(), table);<a name="line.286"></a>
+<span class="sourceLineNo">287</span>      return performBulkLoad(admin, table, regionLocator, queue, pool, secureClient, copyFile);<a name="line.287"></a>
+<span class="sourceLineNo">288</span>    } finally {<a name="line.288"></a>
+<span class="sourceLineNo">289</span>      cleanup(admin, queue, pool, secureClient);<a name="line.289"></a>
+<span class="sourceLineNo">290</span>    }<a name="line.290"></a>
+<span class="sourceLineNo">291</span>  }<a name="line.291"></a>
+<span class="sourceLineNo">292</span><a name="line.292"></a>
+<span class="sourceLineNo">293</span>  /**<a name="line.293"></a>
+<span class="sourceLineNo">294</span>   * Perform a bulk load of the given directory into the given pre-existing table. This method is<a name="line.294"></a>
+<span class="sourceLineNo">295</span>   * not threadsafe.<a name="line.295"></a>
+<span class="sourceLineNo">296</span>   * @param hfofDir the directory that was provided as the output path of a job using<a name="line.296"></a>
+<span class="sourceLineNo">297</span>   *          HFileOutputFormat<a name="line.297"></a>
+<span class="sourceLineNo">298</span>   * @param admin the Admin<a name="line.298"></a>
+<span class="sourceLineNo">299</span>   * @param table the table to load into<a name="line.299"></a>
+<span class="sourceLineNo">300</span>   * @param regionLocator region locator<a name="line.300"></a>
+<span class="sourceLineNo">301</span>   * @param silence true to ignore unmatched column families<a name="line.301"></a>
+<span class="sourceLineNo">302</span>   * @param copyFile always copy hfiles if true<a name="line.302"></a>
+<span class="sourceLineNo">303</span>   * @throws TableNotFoundException if table does not yet exist<a name="line.303"></a>
+<span class="sourceLineNo">304</span>   */<a name="line.304"></a>
+<span class="sourceLineNo">305</span>  public Map&lt;LoadQueueItem, ByteBuffer&gt; doBulkLoad(Path hfofDir, final Admin admin, Table table,<a name="line.305"></a>
+<span class="sourceLineNo">306</span>      RegionLocator regionLocator, boolean silence, boolean copyFile)<a name="line.306"></a>
+<span class="sourceLineNo">307</span>      throws TableNotFoundException, IOException {<a name="line.307"></a>
+<span class="sourceLineNo">308</span>    if (!admin.isTableAvailable(regionLocator.getName())) {<a name="line.308"></a>
+<span class="sourceLineNo">309</span>      throw new TableNotFoundException("Table " + table.getName() + " is not currently available.");<a name="line.309"></a>
+<span class="sourceLineNo">310</span>    }<a name="line.310"></a>
+<span class="sourceLineNo">311</span><a name="line.311"></a>
+<span class="sourceLineNo">312</span>    /*<a name="line.312"></a>
+<span class="sourceLineNo">313</span>     * Checking hfile format is a time-consuming operation, we should have an option to skip this<a name="line.313"></a>
+<span class="sourceLineNo">314</span>     * step when bulkloading millions of HFiles. See HBASE-13985.<a name="line.314"></a>
+<span class="sourceLineNo">315</span>     */<a name="line.315"></a>
+<span class="sourceLineNo">316</span>    boolean validateHFile = getConf().getBoolean("hbase.loadincremental.validate.hfile", true);<a name="line.316"></a>
+<span class="sourceLineNo">317</span>    if (!validateHFile) {<a name="line.317"></a>
+<span class="sourceLineNo">318</span>      LOG.warn("You are skipping HFiles validation, it might cause some data loss if files " +<a name="line.318"></a>
+<span class="sourceLineNo">319</span>          "are not correct. If you fail to read data from your table after using this " +<a name="line.319"></a>
+<span class="sourceLineNo">320</span>          "option, consider removing the files and bulkload again without this option. " +<a name="line.320"></a>
+<span class="sourceLineNo">321</span>          "See HBASE-13985");<a name="line.321"></a>
+<span class="sourceLineNo">322</span>    }<a name="line.322"></a>
+<span class="sourceLineNo">323</span>    // LQI queue does not need to be threadsafe -- all operations on this queue<a name="line.323"></a>
+<span class="sourceLineNo">324</span>    // happen in this thread<a name="line.324"></a>
+<span class="sourceLineNo">325</span>    Deque&lt;LoadQueueItem&gt; queue = new ArrayDeque&lt;&gt;();<a name="line.325"></a>
+<span class="sourceLineNo">326</span>    ExecutorService pool = null;<a name="line.326"></a>
+<span class="sourceLineNo">327</span>    SecureBulkLoadClient secureClient = null;<a name="line.327"></a>
+<span class="sourceLineNo">328</span>    try {<a name="line.328"></a>
+<span class="sourceLineNo">329</span>      prepareHFileQueue(hfofDir, table, queue, validateHFile, silence);<a name="line.329"></a>
+<span class="sourceLineNo">330</span><a name="line.330"></a>
+<span class="sourceLineNo">331</span>      if (queue.isEmpty()) {<a name="line.331"></a>
+<span class="sourceLineNo">332</span>        LOG.warn(<a name="line.332"></a>
+<span class="sourceLineNo">333</span>          "Bulk load operation did not find any files to load in " + "directory " + hfofDir != null<a name="line.333"></a>
+<span class="sourceLineNo">334</span>              ? hfofDir.toUri().toString()<a name="line.334"></a>
+<span class="sourceLineNo">335</span>              : "" + ".  Does it contain files in " +<a name="line.335"></a>
+<span class="sourceLineNo">336</span>                  "subdirectories that correspond to column family names?");<a name="line.336"></a>
+<span class="sourceLineNo">337</span>        return Collections.emptyMap();<a name="line.337"></a>
+<span class="sourceLineNo">338</span>      }<a name="line.338"></a>
+<span class="sourceLineNo">339</span>      pool = createExecutorService();<a name="line.339"></a>
+<span class="sourceLineNo">340</span>      secureClient = new SecureBulkLoadClient(table.getConfiguration(), table);<a name="line.340"></a>
+<span class="sourceLineNo">341</span>      return performBulkLoad(admin, table, regionLocator, queue, pool, secureClient, copyFile);<a name="line.341"></a>
+<span class="sourceLineNo">342</span>    } finally {<a name="line.342"></a>
+<span class="sourceLineNo">343</span>      cleanup(admin, queue, pool, secureClient);<a name="line.343"></a>
+<span class="sourceLineNo">344</span>    }<a name="line.344"></a>
+<span class="sourceLineNo">345</span>  }<a name="line.345"></a>
+<span class="sourceLineNo">346</span><a name="line.346"></a>
+<span class="sourceLineNo">347</span>  /**<a name="line.347"></a>
+<span class="sourceLineNo">348</span>   * Used by the replication sink to load the hfiles from the source cluster. It does the following,<a name="line.348"></a>
+<span class="sourceLineNo">349</span>   * &lt;ol&gt;<a name="line.349"></a>
+<span class="sourceLineNo">350</span>   * &lt;li&gt;LoadIncrementalHFiles#groupOrSplitPhase(Table, ExecutorService, Deque, Pair)}&lt;/li&gt;<a name="line.350"></a>
+<span class="sourceLineNo">351</span>   * &lt;li&gt;LoadIncrementalHFiles#bulkLoadPhase(Table, Connection, ExecutorService, Deque, Multimap)<a name="line.351"></a>
+<span class="sourceLineNo">352</span>   * &lt;/li&gt;<a name="line.352"></a>
+<span class="sourceLineNo">353</span>   * &lt;/ol&gt;<a name="line.353"></a>
+<span class="sourceLineNo">354</span>   * @param table Table to which these hfiles should be loaded to<a name="line.354"></a>
+<span class="sourceLineNo">355</span>   * @param conn Connection to use<a name="line.355"></a>
+<span class="sourceLineNo">356</span>   * @param queue {@link LoadQueueItem} has hfiles yet to be loaded<a name="line.356"></a>
+<span class="sourceLineNo">357</span>   * @param startEndKeys starting and ending row keys of the region<a name="line.357"></a>
+<span class="sourceLineNo">358</span>   */<a name="line.358"></a>
+<span class="sourceLineNo">359</span>  public void loadHFileQueue(Table table, Connection conn, Deque&lt;LoadQueueItem&gt; queue,<a name="line.359"></a>
+<span class="sourceLineNo">360</span>      Pair&lt;byte[][], byte[][]&gt; startEndKeys) throws IOException {<a name="line.360"></a>
+<span class="sourceLineNo">361</span>    loadHFileQueue(table, conn, queue, startEndKeys, false);<a name="line.361"></a>
+<span class="sourceLineNo">362</span>  }<a name="line.362"></a>
+<span class="sourceLineNo">363</span><a name="line.363"></a>
+<span class="sourceLineNo">364</span>  /**<a name="line.364"></a>
+<span class="sourceLineNo">365</span>   * Used by the replication sink to load the hfiles from the source cluster. It does the following,<a name="line.365"></a>
+<span class="sourceLineNo">366</span>   * &lt;ol&gt;<a name="line.366"></a>
+<span class="sourceLineNo">367</span>   * &lt;li&gt;LoadIncrementalHFiles#groupOrSplitPhase(Table, ExecutorService, Deque, Pair)}&lt;/li&gt;<a name="line.367"></a>
+<span class="sourceLineNo">368</span>   * &lt;li&gt;LoadIncrementalHFiles#bulkLoadPhase(Table, Connection, ExecutorService, Deque, Multimap)<a name="line.368"></a>
+<span class="sourceLineNo">369</span>   * &lt;/li&gt;<a name="line.369"></a>
+<span class="sourceLineNo">370</span>   * &lt;/ol&gt;<a name="line.370"></a>
+<span class="sourceLineNo">371</span>   * @param table Table to which these hfiles should be loaded to<a name="line.371"></a>
+<span class="sourceLineNo">372</span>   * @param conn Connection to use<a name="line.372"></a>
+<span class="sourceLineNo">373</span>   * @param queue {@link LoadQueueItem} has hfiles yet to be loaded<a name="line.373"></a>
+<span class="sourceLineNo">374</span>   * @param startEndKeys starting and ending row keys of the region<a name="line.374"></a>
+<span class="sourceLineNo">375</span>   */<a name="line.375"></a>
+<span class="sourceLineNo">376</span>  public void loadHFileQueue(Table table, Connection conn, Deque&lt;LoadQueueItem&gt; queue,<a name="line.376"></a>
+<span class="sourceLineNo">377</span>      Pair&lt;byte[][], byte[][]&gt; startEndKeys, boolean copyFile) throws IOException {<a name="line.377"></a>
+<span class="sourceLineNo">378</span>    ExecutorService pool = null;<a name="line.378"></a>
+<span class="sourceLineNo">379</span>    try {<a name="line.379"></a>
+<span class="sourceLineNo">380</span>      pool = createExecutorService();<a name="line.380"></a>
+<span class="sourceLineNo">381</span>      Multimap&lt;ByteBuffer, LoadQueueItem&gt; regionGroups =<a name="line.381"></a>
+<span class="sourceLineNo">382</span>          groupOrSplitPhase(table, pool, queue, startEndKeys).getFirst();<a name="line.382"></a>
+<span class="sourceLineNo">383</span>      bulkLoadPhase(table, conn, pool, queue, regionGroups, copyFile, null);<a name="line.383"></a>
+<span class="sourceLineNo">384</span>    } finally {<a name="line.384"></a>
+<span class="sourceLineNo">385</span>      if (pool != null) {<a name="line.385"></a>
+<span class="sourceLineNo">386</span>        pool.shutdown();<a name="line.386"></a>
+<span class="sourceLineNo">387</span>      }<a name="line.387"></a>
+<span class="sourceLineNo">388</span>    }<a name="line.388"></a>
+<span class="sourceLineNo">389</span>  }<a name="line.389"></a>
+<span class="sourceLineNo">390</span><a name="line.390"></a>
+<span class="sourceLineNo">391</span>  private Map&lt;LoadQueueItem, ByteBuffer&gt; performBulkLoad(Admin admin, Table table,<a name="line.391"></a>
+<span class="sourceLineNo">392</span>      RegionLocator regionLocator, Deque&lt;LoadQueueItem&gt; queue, ExecutorService pool,<a name="line.392"></a>
+<span class="sourceLineNo">393</span>      SecureBulkLoadClient secureClient, boolean copyFile) throws IOException {<a name="line.393"></a>
+<span class="sourceLineNo">394</span>    int count = 0;<a name="line.394"></a>
+<span class="sourceLineNo">395</span><a name="line.395"></a>
+<span class="sourceLineNo">396</span>    if (isSecureBulkLoadEndpointAvailable()) {<a name="line.396"></a>
+<span class="sourceLineNo">397</span>      LOG.warn("SecureBulkLoadEndpoint is deprecated. It will be removed in future releases.");<a name="line.397"></a>
+<span class="sourceLineNo">398</span>      LOG.warn("Secure bulk load has been integrated into HBase core.");<a name="line.398"></a>
+<span class="sourceLineNo">399</span>    }<a name="line.399"></a>
+<span class="sourceLineNo">400</span><a name="line.400"></a>
+<span class="sourceLineNo">401</span>    fsDelegationToken.acquireDelegationToken(queue.peek().getFilePath().getFileSystem(getConf()));<a name="line.401"></a>
+<span class="sourceLineNo">402</span>    bulkToken = secureClient.prepareBulkLoad(admin.getConnection());<a name="line.402"></a>
+<span class="sourceLineNo">403</span>    Pair&lt;Multimap&lt;ByteBuffer, LoadQueueItem&gt;, Set&lt;String&gt;&gt; pair = null;<a name="line.403"></a>
+<span class="sourceLineNo">404</span><a name="line.404"></a>
+<span class="sourceLineNo">405</span>    Map&lt;LoadQueueItem, ByteBuffer&gt; item2RegionMap = new HashMap&lt;&gt;();<a name="line.405"></a>
+<span class="sourceLineNo">406</span>    // Assumes that region splits can happen while this occurs.<a name="line.406"></a>
+<span class="sourceLineNo">407</span>    while (!queue.isEmpty()) {<a name="line.407"></a>
+<span class="sourceLineNo">408</span>      // need to reload split keys each iteration.<a name="line.408"></a>
+<span class="sourceLineNo">409</span>      final Pair&lt;byte[][], byte[][]&gt; startEndKeys = regionLocator.getStartEndKeys();<a name="line.409"></a>
+<span class="sourceLineNo">410</span>      if (count != 0) {<a name="line.410"></a>
+<span class="sourceLineNo">411</span>        LOG.info("Split occurred while grouping HFiles, retry attempt " + +count + " with " +<a name="line.411"></a>
+<span class="sourceLineNo">412</span>            queue.size() + " files remaining to group or split");<a name="line.412"></a>
+<span class="sourceLineNo">413</span>      }<a name="line.413"></a>
+<span class="sourceLineNo">414</span><a name="line.414"></a>
+<span class="sourceLineNo">415</span>      int maxRetries = getConf().getInt(HConstants.BULKLOAD_MAX_RETRIES_NUMBER, 10);<a name="line.415"></a>
+<span class="sourceLineNo">416</span>      maxRetries = Math.max(maxRetries, startEndKeys.getFirst().length + 1);<a name="line.416"></a>
+<span class="sourceLineNo">417</span>      if (maxRetries != 0 &amp;&amp; count &gt;= maxRetries) {<a name="line.417"></a>
+<span class="sourceLineNo">418</span>        throw new IOException(<a name="line.418"></a>
+<span class="sourceLineNo">419</span>            "Retry attempted " + count + " times without completing, bailing out");<a name="line.419"></a>
+<span class="sourceLineNo">420</span>      }<a name="line.420"></a>
+<span class="sourceLineNo">421</span>      count++;<a name="line.421"></a>
+<span class="sourceLineNo">422</span><a name="line.422"></a>
+<span class="sourceLineNo">423</span>      // Using ByteBuffer for byte[] equality semantics<a name="line.423"></a>
+<span class="sourceLineNo">424</span>      pair = groupOrSplitPhase(table, pool, queue, startEndKeys);<a name="line.424"></a>
+<span class="sourceLineNo">425</span>      Multimap&lt;ByteBuffer, LoadQueueItem&gt; regionGroups = pair.getFirst();<a name="line.425"></a>
+<span class="sourceLineNo">426</span><a name="line.426"></a>
+<span class="sourceLineNo">427</span>      if (!checkHFilesCountPerRegionPerFamily(regionGroups)) {<a name="line.427"></a>
+<span class="sourceLineNo">428</span>        // Error is logged inside checkHFilesCountPerRegionPerFamily.<a name="line.428"></a>
+<span class="sourceLineNo">429</span>        throw new IOException("Trying to load more than " + maxFilesPerRegionPerFamily +<a name="line.429"></a>
+<span class="sourceLineNo">430</span>            " hfiles to one family of one region");<a name="line.430"></a>
+<span class="sourceLineNo">431</span>      }<a name="line.431"></a>
 <span class="sourceLineNo">432</span><a name="line.432"></a>
-<span class="sourceLineNo">433</span>      // NOTE: The next iteration's split / group could happen in parallel to<a name="line.433"></a>
-<span class="sourceLineNo">434</span>      // atomic bulkloads assuming that there are splits and no merges, and<a name="line.434"></a>
-<span class="sourceLineNo">435</span>      // that we can atomically pull out the groups we want to retry.<a name="line.435"></a>
-<span class="sourceLineNo">436</span>    }<a name="line.436"></a>
-<span class="sourceLineNo">437</span><a

<TRUNCATED>

Mime
View raw message