| <!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> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.2"></a> |
| <span class="sourceLineNo">003</span> * or more contributor license agreements. See the NOTICE file<a name="line.3"></a> |
| <span class="sourceLineNo">004</span> * distributed with this work for additional information<a name="line.4"></a> |
| <span class="sourceLineNo">005</span> * regarding copyright ownership. The ASF licenses this file<a name="line.5"></a> |
| <span class="sourceLineNo">006</span> * to you under the Apache License, Version 2.0 (the<a name="line.6"></a> |
| <span class="sourceLineNo">007</span> * "License"); you may not use this file except in compliance<a name="line.7"></a> |
| <span class="sourceLineNo">008</span> * with the License. You may obtain a copy of the License at<a name="line.8"></a> |
| <span class="sourceLineNo">009</span> *<a name="line.9"></a> |
| <span class="sourceLineNo">010</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.10"></a> |
| <span class="sourceLineNo">011</span> *<a name="line.11"></a> |
| <span class="sourceLineNo">012</span> * Unless required by applicable law or agreed to in writing, software<a name="line.12"></a> |
| <span class="sourceLineNo">013</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.13"></a> |
| <span class="sourceLineNo">014</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.14"></a> |
| <span class="sourceLineNo">015</span> * See the License for the specific language governing permissions and<a name="line.15"></a> |
| <span class="sourceLineNo">016</span> * limitations under the License.<a name="line.16"></a> |
| <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.CompletableFuture;<a name="line.43"></a> |
| <span class="sourceLineNo">044</span>import java.util.concurrent.ExecutionException;<a name="line.44"></a> |
| <span class="sourceLineNo">045</span>import java.util.concurrent.ExecutorService;<a name="line.45"></a> |
| <span class="sourceLineNo">046</span>import java.util.concurrent.Future;<a name="line.46"></a> |
| <span class="sourceLineNo">047</span>import java.util.concurrent.LinkedBlockingQueue;<a name="line.47"></a> |
| <span class="sourceLineNo">048</span>import java.util.concurrent.ThreadPoolExecutor;<a name="line.48"></a> |
| <span class="sourceLineNo">049</span>import java.util.concurrent.TimeUnit;<a name="line.49"></a> |
| <span class="sourceLineNo">050</span>import java.util.concurrent.atomic.AtomicInteger;<a name="line.50"></a> |
| <span class="sourceLineNo">051</span>import java.util.stream.Collectors;<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.HBaseInterfaceAudience;<a name="line.60"></a> |
| <span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.HConstants;<a name="line.61"></a> |
| <span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.TableName;<a name="line.62"></a> |
| <span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.TableNotFoundException;<a name="line.63"></a> |
| <span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.client.AsyncAdmin;<a name="line.64"></a> |
| <span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.client.AsyncClusterConnection;<a name="line.65"></a> |
| <span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.client.ClusterConnectionFactory;<a name="line.66"></a> |
| <span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.67"></a> |
| <span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.68"></a> |
| <span class="sourceLineNo">069</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.69"></a> |
| <span class="sourceLineNo">070</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.70"></a> |
| <span class="sourceLineNo">071</span>import org.apache.hadoop.hbase.io.HFileLink;<a name="line.71"></a> |
| <span class="sourceLineNo">072</span>import org.apache.hadoop.hbase.io.HalfStoreFileReader;<a name="line.72"></a> |
| <span class="sourceLineNo">073</span>import org.apache.hadoop.hbase.io.Reference;<a name="line.73"></a> |
| <span class="sourceLineNo">074</span>import org.apache.hadoop.hbase.io.compress.Compression.Algorithm;<a name="line.74"></a> |
| <span class="sourceLineNo">075</span>import org.apache.hadoop.hbase.io.hfile.CacheConfig;<a name="line.75"></a> |
| <span class="sourceLineNo">076</span>import org.apache.hadoop.hbase.io.hfile.HFile;<a name="line.76"></a> |
| <span class="sourceLineNo">077</span>import org.apache.hadoop.hbase.io.hfile.HFileContext;<a name="line.77"></a> |
| <span class="sourceLineNo">078</span>import org.apache.hadoop.hbase.io.hfile.HFileContextBuilder;<a name="line.78"></a> |
| <span class="sourceLineNo">079</span>import org.apache.hadoop.hbase.io.hfile.HFileDataBlockEncoder;<a name="line.79"></a> |
| <span class="sourceLineNo">080</span>import org.apache.hadoop.hbase.io.hfile.HFileScanner;<a name="line.80"></a> |
| <span class="sourceLineNo">081</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.81"></a> |
| <span class="sourceLineNo">082</span>import org.apache.hadoop.hbase.regionserver.HStore;<a name="line.82"></a> |
| <span class="sourceLineNo">083</span>import org.apache.hadoop.hbase.regionserver.StoreFileInfo;<a name="line.83"></a> |
| <span class="sourceLineNo">084</span>import org.apache.hadoop.hbase.regionserver.StoreFileWriter;<a name="line.84"></a> |
| <span class="sourceLineNo">085</span>import org.apache.hadoop.hbase.security.UserProvider;<a name="line.85"></a> |
| <span class="sourceLineNo">086</span>import org.apache.hadoop.hbase.security.token.FsDelegationToken;<a name="line.86"></a> |
| <span class="sourceLineNo">087</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.87"></a> |
| <span class="sourceLineNo">088</span>import org.apache.hadoop.hbase.util.FSVisitor;<a name="line.88"></a> |
| <span class="sourceLineNo">089</span>import org.apache.hadoop.hbase.util.FutureUtils;<a name="line.89"></a> |
| <span class="sourceLineNo">090</span>import org.apache.hadoop.hbase.util.Pair;<a name="line.90"></a> |
| <span class="sourceLineNo">091</span>import org.apache.hadoop.util.Tool;<a name="line.91"></a> |
| <span class="sourceLineNo">092</span>import org.apache.hadoop.util.ToolRunner;<a name="line.92"></a> |
| <span class="sourceLineNo">093</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.93"></a> |
| <span class="sourceLineNo">094</span>import org.slf4j.Logger;<a name="line.94"></a> |
| <span class="sourceLineNo">095</span>import org.slf4j.LoggerFactory;<a name="line.95"></a> |
| <span class="sourceLineNo">096</span><a name="line.96"></a> |
| <span class="sourceLineNo">097</span>import org.apache.hbase.thirdparty.com.google.common.annotations.VisibleForTesting;<a name="line.97"></a> |
| <span class="sourceLineNo">098</span>import org.apache.hbase.thirdparty.com.google.common.collect.HashMultimap;<a name="line.98"></a> |
| <span class="sourceLineNo">099</span>import org.apache.hbase.thirdparty.com.google.common.collect.Lists;<a name="line.99"></a> |
| <span class="sourceLineNo">100</span>import org.apache.hbase.thirdparty.com.google.common.collect.Maps;<a name="line.100"></a> |
| <span class="sourceLineNo">101</span>import org.apache.hbase.thirdparty.com.google.common.collect.Multimap;<a name="line.101"></a> |
| <span class="sourceLineNo">102</span>import org.apache.hbase.thirdparty.com.google.common.collect.Multimaps;<a name="line.102"></a> |
| <span class="sourceLineNo">103</span>import org.apache.hbase.thirdparty.com.google.common.util.concurrent.ThreadFactoryBuilder;<a name="line.103"></a> |
| <span class="sourceLineNo">104</span><a name="line.104"></a> |
| <span class="sourceLineNo">105</span>/**<a name="line.105"></a> |
| <span class="sourceLineNo">106</span> * The implementation for {@link BulkLoadHFiles}, and also can be executed from command line as a<a name="line.106"></a> |
| <span class="sourceLineNo">107</span> * tool.<a name="line.107"></a> |
| <span class="sourceLineNo">108</span> */<a name="line.108"></a> |
| <span class="sourceLineNo">109</span>@InterfaceAudience.LimitedPrivate(HBaseInterfaceAudience.TOOLS)<a name="line.109"></a> |
| <span class="sourceLineNo">110</span>public class BulkLoadHFilesTool extends Configured implements BulkLoadHFiles, 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(BulkLoadHFilesTool.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><a name="line.115"></a> |
| <span class="sourceLineNo">116</span> // We use a '.' prefix which is ignored when walking directory trees<a name="line.116"></a> |
| <span class="sourceLineNo">117</span> // above. It is invalid family name.<a name="line.117"></a> |
| <span class="sourceLineNo">118</span> static final String TMP_DIR = ".tmp";<a name="line.118"></a> |
| <span class="sourceLineNo">119</span><a name="line.119"></a> |
| <span class="sourceLineNo">120</span> private final int maxFilesPerRegionPerFamily;<a name="line.120"></a> |
| <span class="sourceLineNo">121</span> private final boolean assignSeqIds;<a name="line.121"></a> |
| <span class="sourceLineNo">122</span><a name="line.122"></a> |
| <span class="sourceLineNo">123</span> // Source delegation token<a name="line.123"></a> |
| <span class="sourceLineNo">124</span> private final FsDelegationToken fsDelegationToken;<a name="line.124"></a> |
| <span class="sourceLineNo">125</span> private final UserProvider userProvider;<a name="line.125"></a> |
| <span class="sourceLineNo">126</span> private final int nrThreads;<a name="line.126"></a> |
| <span class="sourceLineNo">127</span> private final AtomicInteger numRetries = new AtomicInteger(0);<a name="line.127"></a> |
| <span class="sourceLineNo">128</span> private String bulkToken;<a name="line.128"></a> |
| <span class="sourceLineNo">129</span><a name="line.129"></a> |
| <span class="sourceLineNo">130</span> public BulkLoadHFilesTool(Configuration conf) {<a name="line.130"></a> |
| <span class="sourceLineNo">131</span> // make a copy, just to be sure we're not overriding someone else's config<a name="line.131"></a> |
| <span class="sourceLineNo">132</span> super(new Configuration(conf));<a name="line.132"></a> |
| <span class="sourceLineNo">133</span> // disable blockcache for tool invocation, see HBASE-10500<a name="line.133"></a> |
| <span class="sourceLineNo">134</span> conf.setFloat(HConstants.HFILE_BLOCK_CACHE_SIZE_KEY, 0);<a name="line.134"></a> |
| <span class="sourceLineNo">135</span> userProvider = UserProvider.instantiate(conf);<a name="line.135"></a> |
| <span class="sourceLineNo">136</span> fsDelegationToken = new FsDelegationToken(userProvider, "renewer");<a name="line.136"></a> |
| <span class="sourceLineNo">137</span> assignSeqIds = conf.getBoolean(ASSIGN_SEQ_IDS, true);<a name="line.137"></a> |
| <span class="sourceLineNo">138</span> maxFilesPerRegionPerFamily = conf.getInt(MAX_FILES_PER_REGION_PER_FAMILY, 32);<a name="line.138"></a> |
| <span class="sourceLineNo">139</span> nrThreads =<a name="line.139"></a> |
| <span class="sourceLineNo">140</span> conf.getInt("hbase.loadincremental.threads.max", Runtime.getRuntime().availableProcessors());<a name="line.140"></a> |
| <span class="sourceLineNo">141</span> }<a name="line.141"></a> |
| <span class="sourceLineNo">142</span><a name="line.142"></a> |
| <span class="sourceLineNo">143</span> // Initialize a thread pool<a name="line.143"></a> |
| <span class="sourceLineNo">144</span> private ExecutorService createExecutorService() {<a name="line.144"></a> |
| <span class="sourceLineNo">145</span> ThreadPoolExecutor pool = new ThreadPoolExecutor(nrThreads, nrThreads, 60, TimeUnit.SECONDS,<a name="line.145"></a> |
| <span class="sourceLineNo">146</span> new LinkedBlockingQueue<>(),<a name="line.146"></a> |
| <span class="sourceLineNo">147</span> new ThreadFactoryBuilder().setNameFormat("BulkLoadHFilesTool-%1$d").build());<a name="line.147"></a> |
| <span class="sourceLineNo">148</span> pool.allowCoreThreadTimeOut(true);<a name="line.148"></a> |
| <span class="sourceLineNo">149</span> return pool;<a name="line.149"></a> |
| <span class="sourceLineNo">150</span> }<a name="line.150"></a> |
| <span class="sourceLineNo">151</span><a name="line.151"></a> |
| <span class="sourceLineNo">152</span> private boolean isCreateTable() {<a name="line.152"></a> |
| <span class="sourceLineNo">153</span> return "yes".equalsIgnoreCase(getConf().get(CREATE_TABLE_CONF_KEY, "yes"));<a name="line.153"></a> |
| <span class="sourceLineNo">154</span> }<a name="line.154"></a> |
| <span class="sourceLineNo">155</span><a name="line.155"></a> |
| <span class="sourceLineNo">156</span> private boolean isSilence() {<a name="line.156"></a> |
| <span class="sourceLineNo">157</span> return "yes".equalsIgnoreCase(getConf().get(IGNORE_UNMATCHED_CF_CONF_KEY, ""));<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> private boolean isAlwaysCopyFiles() {<a name="line.160"></a> |
| <span class="sourceLineNo">161</span> return getConf().getBoolean(ALWAYS_COPY_FILES, false);<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> private static boolean shouldCopyHFileMetaKey(byte[] key) {<a name="line.164"></a> |
| <span class="sourceLineNo">165</span> // skip encoding to keep hfile meta consistent with data block info, see HBASE-15085<a name="line.165"></a> |
| <span class="sourceLineNo">166</span> if (Bytes.equals(key, HFileDataBlockEncoder.DATA_BLOCK_ENCODING)) {<a name="line.166"></a> |
| <span class="sourceLineNo">167</span> return false;<a name="line.167"></a> |
| <span class="sourceLineNo">168</span> }<a name="line.168"></a> |
| <span class="sourceLineNo">169</span><a name="line.169"></a> |
| <span class="sourceLineNo">170</span> return !HFile.isReservedFileInfoKey(key);<a name="line.170"></a> |
| <span class="sourceLineNo">171</span> }<a name="line.171"></a> |
| <span class="sourceLineNo">172</span><a name="line.172"></a> |
| <span class="sourceLineNo">173</span> /**<a name="line.173"></a> |
| <span class="sourceLineNo">174</span> * Checks whether there is any invalid family name in HFiles to be bulk loaded.<a name="line.174"></a> |
| <span class="sourceLineNo">175</span> */<a name="line.175"></a> |
| <span class="sourceLineNo">176</span> private static void validateFamiliesInHFiles(TableDescriptor tableDesc,<a name="line.176"></a> |
| <span class="sourceLineNo">177</span> Deque<LoadQueueItem> queue, boolean silence) throws IOException {<a name="line.177"></a> |
| <span class="sourceLineNo">178</span> Set<String> familyNames = Arrays.asList(tableDesc.getColumnFamilies()).stream()<a name="line.178"></a> |
| <span class="sourceLineNo">179</span> .map(f -> f.getNameAsString()).collect(Collectors.toSet());<a name="line.179"></a> |
| <span class="sourceLineNo">180</span> List<String> unmatchedFamilies = queue.stream().map(item -> Bytes.toString(item.getFamily()))<a name="line.180"></a> |
| <span class="sourceLineNo">181</span> .filter(fn -> !familyNames.contains(fn)).distinct().collect(Collectors.toList());<a name="line.181"></a> |
| <span class="sourceLineNo">182</span> if (unmatchedFamilies.size() > 0) {<a name="line.182"></a> |
| <span class="sourceLineNo">183</span> String msg =<a name="line.183"></a> |
| <span class="sourceLineNo">184</span> "Unmatched family names found: unmatched family names in HFiles to be bulkloaded: " +<a name="line.184"></a> |
| <span class="sourceLineNo">185</span> unmatchedFamilies + "; valid family names of table " + tableDesc.getTableName() +<a name="line.185"></a> |
| <span class="sourceLineNo">186</span> " are: " + familyNames;<a name="line.186"></a> |
| <span class="sourceLineNo">187</span> LOG.error(msg);<a name="line.187"></a> |
| <span class="sourceLineNo">188</span> if (!silence) {<a name="line.188"></a> |
| <span class="sourceLineNo">189</span> throw new IOException(msg);<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> }<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> * Populate the Queue with given HFiles<a name="line.195"></a> |
| <span class="sourceLineNo">196</span> */<a name="line.196"></a> |
| <span class="sourceLineNo">197</span> private static void populateLoadQueue(Deque<LoadQueueItem> ret, Map<byte[], List<Path>> map) {<a name="line.197"></a> |
| <span class="sourceLineNo">198</span> map.forEach((k, v) -> v.stream().map(p -> new LoadQueueItem(k, p)).forEachOrdered(ret::add));<a name="line.198"></a> |
| <span class="sourceLineNo">199</span> }<a name="line.199"></a> |
| <span class="sourceLineNo">200</span><a name="line.200"></a> |
| <span class="sourceLineNo">201</span> private interface BulkHFileVisitor<TFamily> {<a name="line.201"></a> |
| <span class="sourceLineNo">202</span><a name="line.202"></a> |
| <span class="sourceLineNo">203</span> TFamily bulkFamily(byte[] familyName) throws IOException;<a name="line.203"></a> |
| <span class="sourceLineNo">204</span><a name="line.204"></a> |
| <span class="sourceLineNo">205</span> void bulkHFile(TFamily family, FileStatus hfileStatus) throws IOException;<a name="line.205"></a> |
| <span class="sourceLineNo">206</span> }<a name="line.206"></a> |
| <span class="sourceLineNo">207</span><a name="line.207"></a> |
| <span class="sourceLineNo">208</span> /**<a name="line.208"></a> |
| <span class="sourceLineNo">209</span> * Iterate over the bulkDir hfiles. Skip reference, HFileLink, files starting with "_". Check and<a name="line.209"></a> |
| <span class="sourceLineNo">210</span> * skip non-valid hfiles by default, or skip this validation by setting<a name="line.210"></a> |
| <span class="sourceLineNo">211</span> * 'hbase.loadincremental.validate.hfile' to false.<a name="line.211"></a> |
| <span class="sourceLineNo">212</span> */<a name="line.212"></a> |
| <span class="sourceLineNo">213</span> private static <TFamily> void visitBulkHFiles(FileSystem fs, Path bulkDir,<a name="line.213"></a> |
| <span class="sourceLineNo">214</span> BulkHFileVisitor<TFamily> visitor, boolean validateHFile) throws IOException {<a name="line.214"></a> |
| <span class="sourceLineNo">215</span> FileStatus[] familyDirStatuses = fs.listStatus(bulkDir);<a name="line.215"></a> |
| <span class="sourceLineNo">216</span> for (FileStatus familyStat : familyDirStatuses) {<a name="line.216"></a> |
| <span class="sourceLineNo">217</span> if (!familyStat.isDirectory()) {<a name="line.217"></a> |
| <span class="sourceLineNo">218</span> LOG.warn("Skipping non-directory " + familyStat.getPath());<a name="line.218"></a> |
| <span class="sourceLineNo">219</span> continue;<a name="line.219"></a> |
| <span class="sourceLineNo">220</span> }<a name="line.220"></a> |
| <span class="sourceLineNo">221</span> Path familyDir = familyStat.getPath();<a name="line.221"></a> |
| <span class="sourceLineNo">222</span> byte[] familyName = Bytes.toBytes(familyDir.getName());<a name="line.222"></a> |
| <span class="sourceLineNo">223</span> // Skip invalid family<a name="line.223"></a> |
| <span class="sourceLineNo">224</span> try {<a name="line.224"></a> |
| <span class="sourceLineNo">225</span> ColumnFamilyDescriptorBuilder.isLegalColumnFamilyName(familyName);<a name="line.225"></a> |
| <span class="sourceLineNo">226</span> } catch (IllegalArgumentException e) {<a name="line.226"></a> |
| <span class="sourceLineNo">227</span> LOG.warn("Skipping invalid " + familyStat.getPath());<a name="line.227"></a> |
| <span class="sourceLineNo">228</span> continue;<a name="line.228"></a> |
| <span class="sourceLineNo">229</span> }<a name="line.229"></a> |
| <span class="sourceLineNo">230</span> TFamily family = visitor.bulkFamily(familyName);<a name="line.230"></a> |
| <span class="sourceLineNo">231</span><a name="line.231"></a> |
| <span class="sourceLineNo">232</span> FileStatus[] hfileStatuses = fs.listStatus(familyDir);<a name="line.232"></a> |
| <span class="sourceLineNo">233</span> for (FileStatus hfileStatus : hfileStatuses) {<a name="line.233"></a> |
| <span class="sourceLineNo">234</span> if (!fs.isFile(hfileStatus.getPath())) {<a name="line.234"></a> |
| <span class="sourceLineNo">235</span> LOG.warn("Skipping non-file " + hfileStatus);<a name="line.235"></a> |
| <span class="sourceLineNo">236</span> continue;<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> Path hfile = hfileStatus.getPath();<a name="line.239"></a> |
| <span class="sourceLineNo">240</span> // Skip "_", reference, HFileLink<a name="line.240"></a> |
| <span class="sourceLineNo">241</span> String fileName = hfile.getName();<a name="line.241"></a> |
| <span class="sourceLineNo">242</span> if (fileName.startsWith("_")) {<a name="line.242"></a> |
| <span class="sourceLineNo">243</span> continue;<a name="line.243"></a> |
| <span class="sourceLineNo">244</span> }<a name="line.244"></a> |
| <span class="sourceLineNo">245</span> if (StoreFileInfo.isReference(fileName)) {<a name="line.245"></a> |
| <span class="sourceLineNo">246</span> LOG.warn("Skipping reference " + fileName);<a name="line.246"></a> |
| <span class="sourceLineNo">247</span> continue;<a name="line.247"></a> |
| <span class="sourceLineNo">248</span> }<a name="line.248"></a> |
| <span class="sourceLineNo">249</span> if (HFileLink.isHFileLink(fileName)) {<a name="line.249"></a> |
| <span class="sourceLineNo">250</span> LOG.warn("Skipping HFileLink " + fileName);<a name="line.250"></a> |
| <span class="sourceLineNo">251</span> continue;<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> // Validate HFile Format if needed<a name="line.254"></a> |
| <span class="sourceLineNo">255</span> if (validateHFile) {<a name="line.255"></a> |
| <span class="sourceLineNo">256</span> try {<a name="line.256"></a> |
| <span class="sourceLineNo">257</span> if (!HFile.isHFileFormat(fs, hfile)) {<a name="line.257"></a> |
| <span class="sourceLineNo">258</span> LOG.warn("the file " + hfile + " doesn't seems to be an hfile. skipping");<a name="line.258"></a> |
| <span class="sourceLineNo">259</span> continue;<a name="line.259"></a> |
| <span class="sourceLineNo">260</span> }<a name="line.260"></a> |
| <span class="sourceLineNo">261</span> } catch (FileNotFoundException e) {<a name="line.261"></a> |
| <span class="sourceLineNo">262</span> LOG.warn("the file " + hfile + " was removed");<a name="line.262"></a> |
| <span class="sourceLineNo">263</span> continue;<a name="line.263"></a> |
| <span class="sourceLineNo">264</span> }<a name="line.264"></a> |
| <span class="sourceLineNo">265</span> }<a name="line.265"></a> |
| <span class="sourceLineNo">266</span><a name="line.266"></a> |
| <span class="sourceLineNo">267</span> visitor.bulkHFile(family, hfileStatus);<a name="line.267"></a> |
| <span class="sourceLineNo">268</span> }<a name="line.268"></a> |
| <span class="sourceLineNo">269</span> }<a name="line.269"></a> |
| <span class="sourceLineNo">270</span> }<a name="line.270"></a> |
| <span class="sourceLineNo">271</span><a name="line.271"></a> |
| <span class="sourceLineNo">272</span> /**<a name="line.272"></a> |
| <span class="sourceLineNo">273</span> * Walk the given directory for all HFiles, and return a Queue containing all such files.<a name="line.273"></a> |
| <span class="sourceLineNo">274</span> */<a name="line.274"></a> |
| <span class="sourceLineNo">275</span> private static void discoverLoadQueue(Configuration conf, Deque<LoadQueueItem> ret, Path hfofDir,<a name="line.275"></a> |
| <span class="sourceLineNo">276</span> boolean validateHFile) throws IOException {<a name="line.276"></a> |
| <span class="sourceLineNo">277</span> visitBulkHFiles(hfofDir.getFileSystem(conf), hfofDir, new BulkHFileVisitor<byte[]>() {<a name="line.277"></a> |
| <span class="sourceLineNo">278</span> @Override<a name="line.278"></a> |
| <span class="sourceLineNo">279</span> public byte[] bulkFamily(final byte[] familyName) {<a name="line.279"></a> |
| <span class="sourceLineNo">280</span> return familyName;<a name="line.280"></a> |
| <span class="sourceLineNo">281</span> }<a name="line.281"></a> |
| <span class="sourceLineNo">282</span><a name="line.282"></a> |
| <span class="sourceLineNo">283</span> @Override<a name="line.283"></a> |
| <span class="sourceLineNo">284</span> public void bulkHFile(final byte[] family, final FileStatus hfile) throws IOException {<a name="line.284"></a> |
| <span class="sourceLineNo">285</span> long length = hfile.getLen();<a name="line.285"></a> |
| <span class="sourceLineNo">286</span> if (length > conf.getLong(HConstants.HREGION_MAX_FILESIZE,<a name="line.286"></a> |
| <span class="sourceLineNo">287</span> HConstants.DEFAULT_MAX_FILE_SIZE)) {<a name="line.287"></a> |
| <span class="sourceLineNo">288</span> LOG.warn("Trying to bulk load hfile " + hfile.getPath() + " with size: " + length +<a name="line.288"></a> |
| <span class="sourceLineNo">289</span> " bytes can be problematic as it may lead to oversplitting.");<a name="line.289"></a> |
| <span class="sourceLineNo">290</span> }<a name="line.290"></a> |
| <span class="sourceLineNo">291</span> ret.add(new LoadQueueItem(family, hfile.getPath()));<a name="line.291"></a> |
| <span class="sourceLineNo">292</span> }<a name="line.292"></a> |
| <span class="sourceLineNo">293</span> }, validateHFile);<a name="line.293"></a> |
| <span class="sourceLineNo">294</span> }<a name="line.294"></a> |
| <span class="sourceLineNo">295</span><a name="line.295"></a> |
| <span class="sourceLineNo">296</span> /**<a name="line.296"></a> |
| <span class="sourceLineNo">297</span> * Prepare a collection of {@code LoadQueueItem} from list of source hfiles contained in the<a name="line.297"></a> |
| <span class="sourceLineNo">298</span> * passed directory and validates whether the prepared queue has all the valid table column<a name="line.298"></a> |
| <span class="sourceLineNo">299</span> * families in it.<a name="line.299"></a> |
| <span class="sourceLineNo">300</span> * @param map map of family to List of hfiles<a name="line.300"></a> |
| <span class="sourceLineNo">301</span> * @param tableName table to which hfiles should be loaded<a name="line.301"></a> |
| <span class="sourceLineNo">302</span> * @param queue queue which needs to be loaded into the table<a name="line.302"></a> |
| <span class="sourceLineNo">303</span> * @param silence true to ignore unmatched column families<a name="line.303"></a> |
| <span class="sourceLineNo">304</span> * @throws IOException If any I/O or network error occurred<a name="line.304"></a> |
| <span class="sourceLineNo">305</span> */<a name="line.305"></a> |
| <span class="sourceLineNo">306</span> public static void prepareHFileQueue(AsyncClusterConnection conn, TableName tableName,<a name="line.306"></a> |
| <span class="sourceLineNo">307</span> Map<byte[], List<Path>> map, Deque<LoadQueueItem> queue, boolean silence) throws IOException {<a name="line.307"></a> |
| <span class="sourceLineNo">308</span> populateLoadQueue(queue, map);<a name="line.308"></a> |
| <span class="sourceLineNo">309</span> validateFamiliesInHFiles(FutureUtils.get(conn.getAdmin().getDescriptor(tableName)), queue,<a name="line.309"></a> |
| <span class="sourceLineNo">310</span> silence);<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> /**<a name="line.313"></a> |
| <span class="sourceLineNo">314</span> * Prepare a collection of {@code LoadQueueItem} from list of source hfiles contained in the<a name="line.314"></a> |
| <span class="sourceLineNo">315</span> * passed directory and validates whether the prepared queue has all the valid table column<a name="line.315"></a> |
| <span class="sourceLineNo">316</span> * families in it.<a name="line.316"></a> |
| <span class="sourceLineNo">317</span> * @param hfilesDir directory containing list of hfiles to be loaded into the table<a name="line.317"></a> |
| <span class="sourceLineNo">318</span> * @param queue queue which needs to be loaded into the table<a name="line.318"></a> |
| <span class="sourceLineNo">319</span> * @param validateHFile if true hfiles will be validated for its format<a name="line.319"></a> |
| <span class="sourceLineNo">320</span> * @param silence true to ignore unmatched column families<a name="line.320"></a> |
| <span class="sourceLineNo">321</span> * @throws IOException If any I/O or network error occurred<a name="line.321"></a> |
| <span class="sourceLineNo">322</span> */<a name="line.322"></a> |
| <span class="sourceLineNo">323</span> public static void prepareHFileQueue(Configuration conf, AsyncClusterConnection conn,<a name="line.323"></a> |
| <span class="sourceLineNo">324</span> TableName tableName, Path hfilesDir, Deque<LoadQueueItem> queue, boolean validateHFile,<a name="line.324"></a> |
| <span class="sourceLineNo">325</span> boolean silence) throws IOException {<a name="line.325"></a> |
| <span class="sourceLineNo">326</span> discoverLoadQueue(conf, queue, hfilesDir, validateHFile);<a name="line.326"></a> |
| <span class="sourceLineNo">327</span> validateFamiliesInHFiles(FutureUtils.get(conn.getAdmin().getDescriptor(tableName)), queue,<a name="line.327"></a> |
| <span class="sourceLineNo">328</span> silence);<a name="line.328"></a> |
| <span class="sourceLineNo">329</span> }<a name="line.329"></a> |
| <span class="sourceLineNo">330</span><a name="line.330"></a> |
| <span class="sourceLineNo">331</span> /**<a name="line.331"></a> |
| <span class="sourceLineNo">332</span> * Used by the replication sink to load the hfiles from the source cluster. It does the following,<a name="line.332"></a> |
| <span class="sourceLineNo">333</span> * <ol><a name="line.333"></a> |
| <span class="sourceLineNo">334</span> * <li>{@link #groupOrSplitPhase(AsyncClusterConnection, TableName, ExecutorService, Deque, List)}<a name="line.334"></a> |
| <span class="sourceLineNo">335</span> * </li><a name="line.335"></a> |
| <span class="sourceLineNo">336</span> * <li>{@link #bulkLoadPhase(AsyncClusterConnection, TableName, Deque, Multimap, boolean, Map)}<a name="line.336"></a> |
| <span class="sourceLineNo">337</span> * </li><a name="line.337"></a> |
| <span class="sourceLineNo">338</span> * </ol><a name="line.338"></a> |
| <span class="sourceLineNo">339</span> * @param conn Connection to use<a name="line.339"></a> |
| <span class="sourceLineNo">340</span> * @param tableName Table to which these hfiles should be loaded to<a name="line.340"></a> |
| <span class="sourceLineNo">341</span> * @param queue {@code LoadQueueItem} has hfiles yet to be loaded<a name="line.341"></a> |
| <span class="sourceLineNo">342</span> */<a name="line.342"></a> |
| <span class="sourceLineNo">343</span> public void loadHFileQueue(AsyncClusterConnection conn, TableName tableName,<a name="line.343"></a> |
| <span class="sourceLineNo">344</span> Deque<LoadQueueItem> queue, boolean copyFiles) throws IOException {<a name="line.344"></a> |
| <span class="sourceLineNo">345</span> ExecutorService pool = createExecutorService();<a name="line.345"></a> |
| <span class="sourceLineNo">346</span> try {<a name="line.346"></a> |
| <span class="sourceLineNo">347</span> Multimap<ByteBuffer, LoadQueueItem> regionGroups = groupOrSplitPhase(conn, tableName, pool,<a name="line.347"></a> |
| <span class="sourceLineNo">348</span> queue, FutureUtils.get(conn.getRegionLocator(tableName).getStartEndKeys())).getFirst();<a name="line.348"></a> |
| <span class="sourceLineNo">349</span> bulkLoadPhase(conn, tableName, queue, regionGroups, copyFiles, null);<a name="line.349"></a> |
| <span class="sourceLineNo">350</span> } finally {<a name="line.350"></a> |
| <span class="sourceLineNo">351</span> pool.shutdown();<a name="line.351"></a> |
| <span class="sourceLineNo">352</span> }<a name="line.352"></a> |
| <span class="sourceLineNo">353</span> }<a name="line.353"></a> |
| <span class="sourceLineNo">354</span><a name="line.354"></a> |
| <span class="sourceLineNo">355</span> /**<a name="line.355"></a> |
| <span class="sourceLineNo">356</span> * This takes the LQI's grouped by likely regions and attempts to bulk load them. Any failures are<a name="line.356"></a> |
| <span class="sourceLineNo">357</span> * re-queued for another pass with the groupOrSplitPhase.<a name="line.357"></a> |
| <span class="sourceLineNo">358</span> * <p/><a name="line.358"></a> |
| <span class="sourceLineNo">359</span> * protected for testing.<a name="line.359"></a> |
| <span class="sourceLineNo">360</span> */<a name="line.360"></a> |
| <span class="sourceLineNo">361</span> @VisibleForTesting<a name="line.361"></a> |
| <span class="sourceLineNo">362</span> protected void bulkLoadPhase(AsyncClusterConnection conn, TableName tableName,<a name="line.362"></a> |
| <span class="sourceLineNo">363</span> Deque<LoadQueueItem> queue, Multimap<ByteBuffer, LoadQueueItem> regionGroups,<a name="line.363"></a> |
| <span class="sourceLineNo">364</span> boolean copyFiles, Map<LoadQueueItem, ByteBuffer> item2RegionMap) throws IOException {<a name="line.364"></a> |
| <span class="sourceLineNo">365</span> // atomically bulk load the groups.<a name="line.365"></a> |
| <span class="sourceLineNo">366</span> List<Future<Collection<LoadQueueItem>>> loadingFutures = new ArrayList<>();<a name="line.366"></a> |
| <span class="sourceLineNo">367</span> for (Entry<ByteBuffer, ? extends Collection<LoadQueueItem>> entry : regionGroups.asMap()<a name="line.367"></a> |
| <span class="sourceLineNo">368</span> .entrySet()) {<a name="line.368"></a> |
| <span class="sourceLineNo">369</span> byte[] first = entry.getKey().array();<a name="line.369"></a> |
| <span class="sourceLineNo">370</span> final Collection<LoadQueueItem> lqis = entry.getValue();<a name="line.370"></a> |
| <span class="sourceLineNo">371</span> List<Pair<byte[], String>> familyPaths =<a name="line.371"></a> |
| <span class="sourceLineNo">372</span> lqis.stream().map(lqi -> Pair.newPair(lqi.getFamily(), lqi.getFilePath().toString()))<a name="line.372"></a> |
| <span class="sourceLineNo">373</span> .collect(Collectors.toList());<a name="line.373"></a> |
| <span class="sourceLineNo">374</span> CompletableFuture<Collection<LoadQueueItem>> future = new CompletableFuture<>();<a name="line.374"></a> |
| <span class="sourceLineNo">375</span> FutureUtils.addListener(conn.bulkLoad(tableName, familyPaths, first, assignSeqIds,<a name="line.375"></a> |
| <span class="sourceLineNo">376</span> fsDelegationToken.getUserToken(), bulkToken, copyFiles), (loaded, error) -> {<a name="line.376"></a> |
| <span class="sourceLineNo">377</span> if (error != null) {<a name="line.377"></a> |
| <span class="sourceLineNo">378</span> LOG.error("Encountered unrecoverable error from region server", error);<a name="line.378"></a> |
| <span class="sourceLineNo">379</span> if (getConf().getBoolean(RETRY_ON_IO_EXCEPTION, false) &&<a name="line.379"></a> |
| <span class="sourceLineNo">380</span> numRetries.get() < getConf().getInt(HConstants.HBASE_CLIENT_RETRIES_NUMBER,<a name="line.380"></a> |
| <span class="sourceLineNo">381</span> HConstants.DEFAULT_HBASE_CLIENT_RETRIES_NUMBER)) {<a name="line.381"></a> |
| <span class="sourceLineNo">382</span> LOG.warn("Will attempt to retry loading failed HFiles. Retry #" +<a name="line.382"></a> |
| <span class="sourceLineNo">383</span> numRetries.incrementAndGet());<a name="line.383"></a> |
| <span class="sourceLineNo">384</span> // return lqi's to retry<a name="line.384"></a> |
| <span class="sourceLineNo">385</span> future.complete(lqis);<a name="line.385"></a> |
| <span class="sourceLineNo">386</span> } else {<a name="line.386"></a> |
| <span class="sourceLineNo">387</span> LOG.error(RETRY_ON_IO_EXCEPTION +<a name="line.387"></a> |
| <span class="sourceLineNo">388</span> " is disabled or we have reached retry limit. Unable to recover");<a name="line.388"></a> |
| <span class="sourceLineNo">389</span> future.completeExceptionally(error);<a name="line.389"></a> |
| <span class="sourceLineNo">390</span> }<a name="line.390"></a> |
| <span class="sourceLineNo">391</span> } else {<a name="line.391"></a> |
| <span class="sourceLineNo">392</span> if (loaded) {<a name="line.392"></a> |
| <span class="sourceLineNo">393</span> future.complete(Collections.emptyList());<a name="line.393"></a> |
| <span class="sourceLineNo">394</span> } else {<a name="line.394"></a> |
| <span class="sourceLineNo">395</span> LOG.warn("Attempt to bulk load region containing " + Bytes.toStringBinary(first) +<a name="line.395"></a> |
| <span class="sourceLineNo">396</span> " into table " + tableName + " with files " + lqis +<a name="line.396"></a> |
| <span class="sourceLineNo">397</span> " failed. This is recoverable and they will be retried.");<a name="line.397"></a> |
| <span class="sourceLineNo">398</span> // return lqi's to retry<a name="line.398"></a> |
| <span class="sourceLineNo">399</span> future.complete(lqis);<a name="line.399"></a> |
| <span class="sourceLineNo">400</span> }<a name="line.400"></a> |
| <span class="sourceLineNo">401</span> }<a name="line.401"></a> |
| <span class="sourceLineNo">402</span> });<a name="line.402"></a> |
| <span class="sourceLineNo">403</span> loadingFutures.add(future);<a name="line.403"></a> |
| <span class="sourceLineNo">404</span> if (item2RegionMap != null) {<a name="line.404"></a> |
| <span class="sourceLineNo">405</span> for (LoadQueueItem lqi : lqis) {<a name="line.405"></a> |
| <span class="sourceLineNo">406</span> item2RegionMap.put(lqi, entry.getKey());<a name="line.406"></a> |
| <span class="sourceLineNo">407</span> }<a name="line.407"></a> |
| <span class="sourceLineNo">408</span> }<a name="line.408"></a> |
| <span class="sourceLineNo">409</span> }<a name="line.409"></a> |
| <span class="sourceLineNo">410</span><a name="line.410"></a> |
| <span class="sourceLineNo">411</span> // get all the results.<a name="line.411"></a> |
| <span class="sourceLineNo">412</span> for (Future<Collection<LoadQueueItem>> future : loadingFutures) {<a name="line.412"></a> |
| <span class="sourceLineNo">413</span> try {<a name="line.413"></a> |
| <span class="sourceLineNo">414</span> Collection<LoadQueueItem> toRetry = future.get();<a name="line.414"></a> |
| <span class="sourceLineNo">415</span><a name="line.415"></a> |
| <span class="sourceLineNo">416</span> for (LoadQueueItem lqi : toRetry) {<a name="line.416"></a> |
| <span class="sourceLineNo">417</span> item2RegionMap.remove(lqi);<a name="line.417"></a> |
| <span class="sourceLineNo">418</span> }<a name="line.418"></a> |
| <span class="sourceLineNo">419</span> // LQIs that are requeued to be regrouped.<a name="line.419"></a> |
| <span class="sourceLineNo">420</span> queue.addAll(toRetry);<a name="line.420"></a> |
| <span class="sourceLineNo">421</span> } catch (ExecutionException e1) {<a name="line.421"></a> |
| <span class="sourceLineNo">422</span> Throwable t = e1.getCause();<a name="line.422"></a> |
| <span class="sourceLineNo">423</span> if (t instanceof IOException) {<a name="line.423"></a> |
| <span class="sourceLineNo">424</span> // At this point something unrecoverable has happened.<a name="line.424"></a> |
| <span class="sourceLineNo">425</span> // TODO Implement bulk load recovery<a name="line.425"></a> |
| <span class="sourceLineNo">426</span> throw new IOException("BulkLoad encountered an unrecoverable problem", t);<a name="line.426"></a> |
| <span class="sourceLineNo">427</span> }<a name="line.427"></a> |
| <span class="sourceLineNo">428</span> LOG.error("Unexpected execution exception during bulk load", e1);<a name="line.428"></a> |
| <span class="sourceLineNo">429</span> throw new IllegalStateException(t);<a name="line.429"></a> |
| <span class="sourceLineNo">430</span> } catch (InterruptedException e1) {<a name="line.430"></a> |
| <span class="sourceLineNo">431</span> LOG.error("Unexpected interrupted exception during bulk load", e1);<a name="line.431"></a> |
| <span class="sourceLineNo">432</span> throw (InterruptedIOException) new InterruptedIOException().initCause(e1);<a name="line.432"></a> |
| <span class="sourceLineNo">433</span> }<a name="line.433"></a> |
| <span class="sourceLineNo">434</span> }<a name="line.434"></a> |
| <span class="sourceLineNo">435</span> }<a name="line.435"></a> |
| <span class="sourceLineNo">436</span><a name="line.436"></a> |
| <span class="sourceLineNo">437</span> private boolean checkHFilesCountPerRegionPerFamily(<a name="line.437"></a> |
| <span class="sourceLineNo">438</span> final Multimap<ByteBuffer, LoadQueueItem> regionGroups) {<a name="line.438"></a> |
| <span class="sourceLineNo">439</span> for (Map.Entry<ByteBuffer, Collection<LoadQueueItem>> e : regionGroups.asMap().entrySet()) {<a name="line.439"></a> |
| <span class="sourceLineNo">440</span> Map<byte[], MutableInt> filesMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.440"></a> |
| <span class="sourceLineNo">441</span> for (LoadQueueItem lqi : e.getValue()) {<a name="line.441"></a> |
| <span class="sourceLineNo">442</span> MutableInt count = filesMap.computeIfAbsent(lqi.getFamily(), k -> new MutableInt());<a name="line.442"></a> |
| <span class="sourceLineNo">443</span> count.increment();<a name="line.443"></a> |
| <span class="sourceLineNo">444</span> if (count.intValue() > maxFilesPerRegionPerFamily) {<a name="line.444"></a> |
| <span class="sourceLineNo">445</span> LOG.error("Trying to load more than " + maxFilesPerRegionPerFamily +<a name="line.445"></a> |
| <span class="sourceLineNo">446</span> " hfiles to family " + Bytes.toStringBinary(lqi.getFamily()) +<a name="line.446"></a> |
| <span class="sourceLineNo">447</span> " of region with start key " + Bytes.toStringBinary(e.getKey()));<a name="line.447"></a> |
| <span class="sourceLineNo">448</span> return false;<a name="line.448"></a> |
| <span class="sourceLineNo">449</span> }<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> return true;<a name="line.452"></a> |
| <span class="sourceLineNo">453</span> }<a name="line.453"></a> |
| <span class="sourceLineNo">454</span><a name="line.454"></a> |
| <span class="sourceLineNo">455</span> /**<a name="line.455"></a> |
| <span class="sourceLineNo">456</span> * @param conn the HBase cluster connection<a name="line.456"></a> |
| <span class="sourceLineNo">457</span> * @param tableName the table name of the table to load into<a name="line.457"></a> |
| <span class="sourceLineNo">458</span> * @param pool the ExecutorService<a name="line.458"></a> |
| <span class="sourceLineNo">459</span> * @param queue the queue for LoadQueueItem<a name="line.459"></a> |
| <span class="sourceLineNo">460</span> * @param startEndKeys start and end keys<a name="line.460"></a> |
| <span class="sourceLineNo">461</span> * @return A map that groups LQI by likely bulk load region targets and Set of missing hfiles.<a name="line.461"></a> |
| <span class="sourceLineNo">462</span> */<a name="line.462"></a> |
| <span class="sourceLineNo">463</span> private Pair<Multimap<ByteBuffer, LoadQueueItem>, Set<String>> groupOrSplitPhase(<a name="line.463"></a> |
| <span class="sourceLineNo">464</span> AsyncClusterConnection conn, TableName tableName, ExecutorService pool,<a name="line.464"></a> |
| <span class="sourceLineNo">465</span> Deque<LoadQueueItem> queue, List<Pair<byte[], byte[]>> startEndKeys) throws IOException {<a name="line.465"></a> |
| <span class="sourceLineNo">466</span> // <region start key, LQI> need synchronized only within this scope of this<a name="line.466"></a> |
| <span class="sourceLineNo">467</span> // phase because of the puts that happen in futures.<a name="line.467"></a> |
| <span class="sourceLineNo">468</span> Multimap<ByteBuffer, LoadQueueItem> rgs = HashMultimap.create();<a name="line.468"></a> |
| <span class="sourceLineNo">469</span> final Multimap<ByteBuffer, LoadQueueItem> regionGroups = Multimaps.synchronizedMultimap(rgs);<a name="line.469"></a> |
| <span class="sourceLineNo">470</span> Set<String> missingHFiles = new HashSet<>();<a name="line.470"></a> |
| <span class="sourceLineNo">471</span> Pair<Multimap<ByteBuffer, LoadQueueItem>, Set<String>> pair =<a name="line.471"></a> |
| <span class="sourceLineNo">472</span> new Pair<>(regionGroups, missingHFiles);<a name="line.472"></a> |
| <span class="sourceLineNo">473</span><a name="line.473"></a> |
| <span class="sourceLineNo">474</span> // drain LQIs and figure out bulk load groups<a name="line.474"></a> |
| <span class="sourceLineNo">475</span> Set<Future<Pair<List<LoadQueueItem>, String>>> splittingFutures = new HashSet<>();<a name="line.475"></a> |
| <span class="sourceLineNo">476</span> while (!queue.isEmpty()) {<a name="line.476"></a> |
| <span class="sourceLineNo">477</span> final LoadQueueItem item = queue.remove();<a name="line.477"></a> |
| <span class="sourceLineNo">478</span><a name="line.478"></a> |
| <span class="sourceLineNo">479</span> final Callable<Pair<List<LoadQueueItem>, String>> call =<a name="line.479"></a> |
| <span class="sourceLineNo">480</span> new Callable<Pair<List<LoadQueueItem>, String>>() {<a name="line.480"></a> |
| <span class="sourceLineNo">481</span> @Override<a name="line.481"></a> |
| <span class="sourceLineNo">482</span> public Pair<List<LoadQueueItem>, String> call() throws Exception {<a name="line.482"></a> |
| <span class="sourceLineNo">483</span> Pair<List<LoadQueueItem>, String> splits =<a name="line.483"></a> |
| <span class="sourceLineNo">484</span> groupOrSplit(conn, tableName, regionGroups, item, startEndKeys);<a name="line.484"></a> |
| <span class="sourceLineNo">485</span> return splits;<a name="line.485"></a> |
| <span class="sourceLineNo">486</span> }<a name="line.486"></a> |
| <span class="sourceLineNo">487</span> };<a name="line.487"></a> |
| <span class="sourceLineNo">488</span> splittingFutures.add(pool.submit(call));<a name="line.488"></a> |
| <span class="sourceLineNo">489</span> }<a name="line.489"></a> |
| <span class="sourceLineNo">490</span> // get all the results. All grouping and splitting must finish before<a name="line.490"></a> |
| <span class="sourceLineNo">491</span> // we can attempt the atomic loads.<a name="line.491"></a> |
| <span class="sourceLineNo">492</span> for (Future<Pair<List<LoadQueueItem>, String>> lqis : splittingFutures) {<a name="line.492"></a> |
| <span class="sourceLineNo">493</span> try {<a name="line.493"></a> |
| <span class="sourceLineNo">494</span> Pair<List<LoadQueueItem>, String> splits = lqis.get();<a name="line.494"></a> |
| <span class="sourceLineNo">495</span> if (splits != null) {<a name="line.495"></a> |
| <span class="sourceLineNo">496</span> if (splits.getFirst() != null) {<a name="line.496"></a> |
| <span class="sourceLineNo">497</span> queue.addAll(splits.getFirst());<a name="line.497"></a> |
| <span class="sourceLineNo">498</span> } else {<a name="line.498"></a> |
| <span class="sourceLineNo">499</span> missingHFiles.add(splits.getSecond());<a name="line.499"></a> |
| <span class="sourceLineNo">500</span> }<a name="line.500"></a> |
| <span class="sourceLineNo">501</span> }<a name="line.501"></a> |
| <span class="sourceLineNo">502</span> } catch (ExecutionException e1) {<a name="line.502"></a> |
| <span class="sourceLineNo">503</span> Throwable t = e1.getCause();<a name="line.503"></a> |
| <span class="sourceLineNo">504</span> if (t instanceof IOException) {<a name="line.504"></a> |
| <span class="sourceLineNo">505</span> LOG.error("IOException during splitting", e1);<a name="line.505"></a> |
| <span class="sourceLineNo">506</span> throw (IOException) t; // would have been thrown if not parallelized,<a name="line.506"></a> |
| <span class="sourceLineNo">507</span> }<a name="line.507"></a> |
| <span class="sourceLineNo">508</span> LOG.error("Unexpected execution exception during splitting", e1);<a name="line.508"></a> |
| <span class="sourceLineNo">509</span> throw new IllegalStateException(t);<a name="line.509"></a> |
| <span class="sourceLineNo">510</span> } catch (InterruptedException e1) {<a name="line.510"></a> |
| <span class="sourceLineNo">511</span> LOG.error("Unexpected interrupted exception during splitting", e1);<a name="line.511"></a> |
| <span class="sourceLineNo">512</span> throw (InterruptedIOException) new InterruptedIOException().initCause(e1);<a name="line.512"></a> |
| <span class="sourceLineNo">513</span> }<a name="line.513"></a> |
| <span class="sourceLineNo">514</span> }<a name="line.514"></a> |
| <span class="sourceLineNo">515</span> return pair;<a name="line.515"></a> |
| <span class="sourceLineNo">516</span> }<a name="line.516"></a> |
| <span class="sourceLineNo">517</span><a name="line.517"></a> |
| <span class="sourceLineNo">518</span> // unique file name for the table<a name="line.518"></a> |
| <span class="sourceLineNo">519</span> private String getUniqueName() {<a name="line.519"></a> |
| <span class="sourceLineNo">520</span> return UUID.randomUUID().toString().replaceAll("-", "");<a name="line.520"></a> |
| <span class="sourceLineNo">521</span> }<a name="line.521"></a> |
| <span class="sourceLineNo">522</span><a name="line.522"></a> |
| <span class="sourceLineNo">523</span> private List<LoadQueueItem> splitStoreFile(LoadQueueItem item, TableDescriptor tableDesc,<a name="line.523"></a> |
| <span class="sourceLineNo">524</span> byte[] startKey, byte[] splitKey) throws IOException {<a name="line.524"></a> |
| <span class="sourceLineNo">525</span> Path hfilePath = item.getFilePath();<a name="line.525"></a> |
| <span class="sourceLineNo">526</span> byte[] family = item.getFamily();<a name="line.526"></a> |
| <span class="sourceLineNo">527</span> Path tmpDir = hfilePath.getParent();<a name="line.527"></a> |
| <span class="sourceLineNo">528</span> if (!tmpDir.getName().equals(TMP_DIR)) {<a name="line.528"></a> |
| <span class="sourceLineNo">529</span> tmpDir = new Path(tmpDir, TMP_DIR);<a name="line.529"></a> |
| <span class="sourceLineNo">530</span> }<a name="line.530"></a> |
| <span class="sourceLineNo">531</span><a name="line.531"></a> |
| <span class="sourceLineNo">532</span> LOG.info("HFile at " + hfilePath + " no longer fits inside a single " + "region. Splitting...");<a name="line.532"></a> |
| <span class="sourceLineNo">533</span><a name="line.533"></a> |
| <span class="sourceLineNo">534</span> String uniqueName = getUniqueName();<a name="line.534"></a> |
| <span class="sourceLineNo">535</span> ColumnFamilyDescriptor familyDesc = tableDesc.getColumnFamily(family);<a name="line.535"></a> |
| <span class="sourceLineNo">536</span><a name="line.536"></a> |
| <span class="sourceLineNo">537</span> Path botOut = new Path(tmpDir, uniqueName + ".bottom");<a name="line.537"></a> |
| <span class="sourceLineNo">538</span> Path topOut = new Path(tmpDir, uniqueName + ".top");<a name="line.538"></a> |
| <span class="sourceLineNo">539</span> splitStoreFile(getConf(), hfilePath, familyDesc, splitKey, botOut, topOut);<a name="line.539"></a> |
| <span class="sourceLineNo">540</span><a name="line.540"></a> |
| <span class="sourceLineNo">541</span> FileSystem fs = tmpDir.getFileSystem(getConf());<a name="line.541"></a> |
| <span class="sourceLineNo">542</span> fs.setPermission(tmpDir, FsPermission.valueOf("-rwxrwxrwx"));<a name="line.542"></a> |
| <span class="sourceLineNo">543</span> fs.setPermission(botOut, FsPermission.valueOf("-rwxrwxrwx"));<a name="line.543"></a> |
| <span class="sourceLineNo">544</span> fs.setPermission(topOut, FsPermission.valueOf("-rwxrwxrwx"));<a name="line.544"></a> |
| <span class="sourceLineNo">545</span><a name="line.545"></a> |
| <span class="sourceLineNo">546</span> // Add these back at the *front* of the queue, so there's a lower<a name="line.546"></a> |
| <span class="sourceLineNo">547</span> // chance that the region will just split again before we get there.<a name="line.547"></a> |
| <span class="sourceLineNo">548</span> List<LoadQueueItem> lqis = new ArrayList<>(2);<a name="line.548"></a> |
| <span class="sourceLineNo">549</span> lqis.add(new LoadQueueItem(family, botOut));<a name="line.549"></a> |
| <span class="sourceLineNo">550</span> lqis.add(new LoadQueueItem(family, topOut));<a name="line.550"></a> |
| <span class="sourceLineNo">551</span><a name="line.551"></a> |
| <span class="sourceLineNo">552</span> // If the current item is already the result of previous splits,<a name="line.552"></a> |
| <span class="sourceLineNo">553</span> // we don't need it anymore. Clean up to save space.<a name="line.553"></a> |
| <span class="sourceLineNo">554</span> // It is not part of the original input files.<a name="line.554"></a> |
| <span class="sourceLineNo">555</span> try {<a name="line.555"></a> |
| <span class="sourceLineNo">556</span> if (tmpDir.getName().equals(TMP_DIR)) {<a name="line.556"></a> |
| <span class="sourceLineNo">557</span> fs.delete(hfilePath, false);<a name="line.557"></a> |
| <span class="sourceLineNo">558</span> }<a name="line.558"></a> |
| <span class="sourceLineNo">559</span> } catch (IOException e) {<a name="line.559"></a> |
| <span class="sourceLineNo">560</span> LOG.warn("Unable to delete temporary split file " + hfilePath);<a name="line.560"></a> |
| <span class="sourceLineNo">561</span> }<a name="line.561"></a> |
| <span class="sourceLineNo">562</span> LOG.info("Successfully split into new HFiles " + botOut + " and " + topOut);<a name="line.562"></a> |
| <span class="sourceLineNo">563</span> return lqis;<a name="line.563"></a> |
| <span class="sourceLineNo">564</span> }<a name="line.564"></a> |
| <span class="sourceLineNo">565</span><a name="line.565"></a> |
| <span class="sourceLineNo">566</span> /**<a name="line.566"></a> |
| <span class="sourceLineNo">567</span> * Attempt to assign the given load queue item into its target region group. If the hfile boundary<a name="line.567"></a> |
| <span class="sourceLineNo">568</span> * no longer fits into a region, physically splits the hfile such that the new bottom half will<a name="line.568"></a> |
| <span class="sourceLineNo">569</span> * fit and returns the list of LQI's corresponding to the resultant hfiles.<a name="line.569"></a> |
| <span class="sourceLineNo">570</span> * <p/><a name="line.570"></a> |
| <span class="sourceLineNo">571</span> * protected for testing<a name="line.571"></a> |
| <span class="sourceLineNo">572</span> * @throws IOException if an IO failure is encountered<a name="line.572"></a> |
| <span class="sourceLineNo">573</span> */<a name="line.573"></a> |
| <span class="sourceLineNo">574</span> @VisibleForTesting<a name="line.574"></a> |
| <span class="sourceLineNo">575</span> protected Pair<List<LoadQueueItem>, String> groupOrSplit(AsyncClusterConnection conn,<a name="line.575"></a> |
| <span class="sourceLineNo">576</span> TableName tableName, Multimap<ByteBuffer, LoadQueueItem> regionGroups, LoadQueueItem item,<a name="line.576"></a> |
| <span class="sourceLineNo">577</span> List<Pair<byte[], byte[]>> startEndKeys) throws IOException {<a name="line.577"></a> |
| <span class="sourceLineNo">578</span> Path hfilePath = item.getFilePath();<a name="line.578"></a> |
| <span class="sourceLineNo">579</span> Optional<byte[]> first, last;<a name="line.579"></a> |
| <span class="sourceLineNo">580</span> try (HFile.Reader hfr = HFile.createReader(hfilePath.getFileSystem(getConf()), hfilePath,<a name="line.580"></a> |
| <span class="sourceLineNo">581</span> CacheConfig.DISABLED, true, getConf())) {<a name="line.581"></a> |
| <span class="sourceLineNo">582</span> hfr.loadFileInfo();<a name="line.582"></a> |
| <span class="sourceLineNo">583</span> first = hfr.getFirstRowKey();<a name="line.583"></a> |
| <span class="sourceLineNo">584</span> last = hfr.getLastRowKey();<a name="line.584"></a> |
| <span class="sourceLineNo">585</span> } catch (FileNotFoundException fnfe) {<a name="line.585"></a> |
| <span class="sourceLineNo">586</span> LOG.debug("encountered", fnfe);<a name="line.586"></a> |
| <span class="sourceLineNo">587</span> return new Pair<>(null, hfilePath.getName());<a name="line.587"></a> |
| <span class="sourceLineNo">588</span> }<a name="line.588"></a> |
| <span class="sourceLineNo">589</span><a name="line.589"></a> |
| <span class="sourceLineNo">590</span> LOG.info("Trying to load hfile=" + hfilePath + " first=" + first.map(Bytes::toStringBinary) +<a name="line.590"></a> |
| <span class="sourceLineNo">591</span> " last=" + last.map(Bytes::toStringBinary));<a name="line.591"></a> |
| <span class="sourceLineNo">592</span> if (!first.isPresent() || !last.isPresent()) {<a name="line.592"></a> |
| <span class="sourceLineNo">593</span> assert !first.isPresent() && !last.isPresent();<a name="line.593"></a> |
| <span class="sourceLineNo">594</span> // TODO what if this is due to a bad HFile?<a name="line.594"></a> |
| <span class="sourceLineNo">595</span> LOG.info("hfile " + hfilePath + " has no entries, skipping");<a name="line.595"></a> |
| <span class="sourceLineNo">596</span> return null;<a name="line.596"></a> |
| <span class="sourceLineNo">597</span> }<a name="line.597"></a> |
| <span class="sourceLineNo">598</span> if (Bytes.compareTo(first.get(), last.get()) > 0) {<a name="line.598"></a> |
| <span class="sourceLineNo">599</span> throw new IllegalArgumentException("Invalid range: " + Bytes.toStringBinary(first.get()) +<a name="line.599"></a> |
| <span class="sourceLineNo">600</span> " > " + Bytes.toStringBinary(last.get()));<a name="line.600"></a> |
| <span class="sourceLineNo">601</span> }<a name="line.601"></a> |
| <span class="sourceLineNo">602</span> int idx =<a name="line.602"></a> |
| <span class="sourceLineNo">603</span> Collections.binarySearch(startEndKeys, Pair.newPair(first.get(), HConstants.EMPTY_END_ROW),<a name="line.603"></a> |
| <span class="sourceLineNo">604</span> (p1, p2) -> Bytes.compareTo(p1.getFirst(), p2.getFirst()));<a name="line.604"></a> |
| <span class="sourceLineNo">605</span> if (idx < 0) {<a name="line.605"></a> |
| <span class="sourceLineNo">606</span> // not on boundary, returns -(insertion index). Calculate region it<a name="line.606"></a> |
| <span class="sourceLineNo">607</span> // would be in.<a name="line.607"></a> |
| <span class="sourceLineNo">608</span> idx = -(idx + 1) - 1;<a name="line.608"></a> |
| <span class="sourceLineNo">609</span> }<a name="line.609"></a> |
| <span class="sourceLineNo">610</span> int indexForCallable = idx;<a name="line.610"></a> |
| <span class="sourceLineNo">611</span><a name="line.611"></a> |
| <span class="sourceLineNo">612</span> /**<a name="line.612"></a> |
| <span class="sourceLineNo">613</span> * we can consider there is a region hole in following conditions. 1) if idx < 0,then first<a name="line.613"></a> |
| <span class="sourceLineNo">614</span> * region info is lost. 2) if the endkey of a region is not equal to the startkey of the next<a name="line.614"></a> |
| <span class="sourceLineNo">615</span> * region. 3) if the endkey of the last region is not empty.<a name="line.615"></a> |
| <span class="sourceLineNo">616</span> */<a name="line.616"></a> |
| <span class="sourceLineNo">617</span> if (indexForCallable < 0) {<a name="line.617"></a> |
| <span class="sourceLineNo">618</span> throw new IOException("The first region info for table " + tableName +<a name="line.618"></a> |
| <span class="sourceLineNo">619</span> " can't be found in hbase:meta.Please use hbck tool to fix it first.");<a name="line.619"></a> |
| <span class="sourceLineNo">620</span> } else if ((indexForCallable == startEndKeys.size() - 1) &&<a name="line.620"></a> |
| <span class="sourceLineNo">621</span> !Bytes.equals(startEndKeys.get(indexForCallable).getSecond(), HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.621"></a> |
| <span class="sourceLineNo">622</span> throw new IOException("The last region info for table " + tableName +<a name="line.622"></a> |
| <span class="sourceLineNo">623</span> " can't be found in hbase:meta.Please use hbck tool to fix it first.");<a name="line.623"></a> |
| <span class="sourceLineNo">624</span> } else if (indexForCallable + 1 < startEndKeys.size() &&<a name="line.624"></a> |
| <span class="sourceLineNo">625</span> !(Bytes.compareTo(startEndKeys.get(indexForCallable).getSecond(),<a name="line.625"></a> |
| <span class="sourceLineNo">626</span> startEndKeys.get(indexForCallable + 1).getFirst()) == 0)) {<a name="line.626"></a> |
| <span class="sourceLineNo">627</span> throw new IOException("The endkey of one region for table " + tableName +<a name="line.627"></a> |
| <span class="sourceLineNo">628</span> " is not equal to the startkey of the next region in hbase:meta." +<a name="line.628"></a> |
| <span class="sourceLineNo">629</span> "Please use hbck tool to fix it first.");<a name="line.629"></a> |
| <span class="sourceLineNo">630</span> }<a name="line.630"></a> |
| <span class="sourceLineNo">631</span><a name="line.631"></a> |
| <span class="sourceLineNo">632</span> boolean lastKeyInRange = Bytes.compareTo(last.get(), startEndKeys.get(idx).getSecond()) < 0 ||<a name="line.632"></a> |
| <span class="sourceLineNo">633</span> Bytes.equals(startEndKeys.get(idx).getSecond(), HConstants.EMPTY_BYTE_ARRAY);<a name="line.633"></a> |
| <span class="sourceLineNo">634</span> if (!lastKeyInRange) {<a name="line.634"></a> |
| <span class="sourceLineNo">635</span> Pair<byte[], byte[]> startEndKey = startEndKeys.get(indexForCallable);<a name="line.635"></a> |
| <span class="sourceLineNo">636</span> List<LoadQueueItem> lqis =<a name="line.636"></a> |
| <span class="sourceLineNo">637</span> splitStoreFile(item, FutureUtils.get(conn.getAdmin().getDescriptor(tableName)),<a name="line.637"></a> |
| <span class="sourceLineNo">638</span> startEndKey.getFirst(), startEndKey.getSecond());<a name="line.638"></a> |
| <span class="sourceLineNo">639</span> return new Pair<>(lqis, null);<a name="line.639"></a> |
| <span class="sourceLineNo">640</span> }<a name="line.640"></a> |
| <span class="sourceLineNo">641</span><a name="line.641"></a> |
| <span class="sourceLineNo">642</span> // group regions.<a name="line.642"></a> |
| <span class="sourceLineNo">643</span> regionGroups.put(ByteBuffer.wrap(startEndKeys.get(idx).getFirst()), item);<a name="line.643"></a> |
| <span class="sourceLineNo">644</span> return null;<a name="line.644"></a> |
| <span class="sourceLineNo">645</span> }<a name="line.645"></a> |
| <span class="sourceLineNo">646</span><a name="line.646"></a> |
| <span class="sourceLineNo">647</span> /**<a name="line.647"></a> |
| <span class="sourceLineNo">648</span> * Split a storefile into a top and bottom half, maintaining the metadata, recreating bloom<a name="line.648"></a> |
| <span class="sourceLineNo">649</span> * filters, etc.<a name="line.649"></a> |
| <span class="sourceLineNo">650</span> */<a name="line.650"></a> |
| <span class="sourceLineNo">651</span> @VisibleForTesting<a name="line.651"></a> |
| <span class="sourceLineNo">652</span> static void splitStoreFile(Configuration conf, Path inFile, ColumnFamilyDescriptor familyDesc,<a name="line.652"></a> |
| <span class="sourceLineNo">653</span> byte[] splitKey, Path bottomOut, Path topOut) throws IOException {<a name="line.653"></a> |
| <span class="sourceLineNo">654</span> // Open reader with no block cache, and not in-memory<a name="line.654"></a> |
| <span class="sourceLineNo">655</span> Reference topReference = Reference.createTopReference(splitKey);<a name="line.655"></a> |
| <span class="sourceLineNo">656</span> Reference bottomReference = Reference.createBottomReference(splitKey);<a name="line.656"></a> |
| <span class="sourceLineNo">657</span><a name="line.657"></a> |
| <span class="sourceLineNo">658</span> copyHFileHalf(conf, inFile, topOut, topReference, familyDesc);<a name="line.658"></a> |
| <span class="sourceLineNo">659</span> copyHFileHalf(conf, inFile, bottomOut, bottomReference, familyDesc);<a name="line.659"></a> |
| <span class="sourceLineNo">660</span> }<a name="line.660"></a> |
| <span class="sourceLineNo">661</span><a name="line.661"></a> |
| <span class="sourceLineNo">662</span> /**<a name="line.662"></a> |
| <span class="sourceLineNo">663</span> * Copy half of an HFile into a new HFile.<a name="line.663"></a> |
| <span class="sourceLineNo">664</span> */<a name="line.664"></a> |
| <span class="sourceLineNo">665</span> private static void copyHFileHalf(Configuration conf, Path inFile, Path outFile,<a name="line.665"></a> |
| <span class="sourceLineNo">666</span> Reference reference, ColumnFamilyDescriptor familyDescriptor) throws IOException {<a name="line.666"></a> |
| <span class="sourceLineNo">667</span> FileSystem fs = inFile.getFileSystem(conf);<a name="line.667"></a> |
| <span class="sourceLineNo">668</span> CacheConfig cacheConf = CacheConfig.DISABLED;<a name="line.668"></a> |
| <span class="sourceLineNo">669</span> HalfStoreFileReader halfReader = null;<a name="line.669"></a> |
| <span class="sourceLineNo">670</span> StoreFileWriter halfWriter = null;<a name="line.670"></a> |
| <span class="sourceLineNo">671</span> try {<a name="line.671"></a> |
| <span class="sourceLineNo">672</span> halfReader = new HalfStoreFileReader(fs, inFile, cacheConf, reference, true,<a name="line.672"></a> |
| <span class="sourceLineNo">673</span> new AtomicInteger(0), true, conf);<a name="line.673"></a> |
| <span class="sourceLineNo">674</span> Map<byte[], byte[]> fileInfo = halfReader.loadFileInfo();<a name="line.674"></a> |
| <span class="sourceLineNo">675</span><a name="line.675"></a> |
| <span class="sourceLineNo">676</span> int blocksize = familyDescriptor.getBlocksize();<a name="line.676"></a> |
| <span class="sourceLineNo">677</span> Algorithm compression = familyDescriptor.getCompressionType();<a name="line.677"></a> |
| <span class="sourceLineNo">678</span> BloomType bloomFilterType = familyDescriptor.getBloomFilterType();<a name="line.678"></a> |
| <span class="sourceLineNo">679</span> HFileContext hFileContext = new HFileContextBuilder().withCompression(compression)<a name="line.679"></a> |
| <span class="sourceLineNo">680</span> .withChecksumType(HStore.getChecksumType(conf))<a name="line.680"></a> |
| <span class="sourceLineNo">681</span> .withBytesPerCheckSum(HStore.getBytesPerChecksum(conf)).withBlockSize(blocksize)<a name="line.681"></a> |
| <span class="sourceLineNo">682</span> .withDataBlockEncoding(familyDescriptor.getDataBlockEncoding()).withIncludesTags(true)<a name="line.682"></a> |
| <span class="sourceLineNo">683</span> .build();<a name="line.683"></a> |
| <span class="sourceLineNo">684</span> halfWriter = new StoreFileWriter.Builder(conf, cacheConf, fs).withFilePath(outFile)<a name="line.684"></a> |
| <span class="sourceLineNo">685</span> .withBloomType(bloomFilterType).withFileContext(hFileContext).build();<a name="line.685"></a> |
| <span class="sourceLineNo">686</span> HFileScanner scanner = halfReader.getScanner(false, false, false);<a name="line.686"></a> |
| <span class="sourceLineNo">687</span> scanner.seekTo();<a name="line.687"></a> |
| <span class="sourceLineNo">688</span> do {<a name="line.688"></a> |
| <span class="sourceLineNo">689</span> halfWriter.append(scanner.getCell());<a name="line.689"></a> |
| <span class="sourceLineNo">690</span> } while (scanner.next());<a name="line.690"></a> |
| <span class="sourceLineNo">691</span><a name="line.691"></a> |
| <span class="sourceLineNo">692</span> for (Map.Entry<byte[], byte[]> entry : fileInfo.entrySet()) {<a name="line.692"></a> |
| <span class="sourceLineNo">693</span> if (shouldCopyHFileMetaKey(entry.getKey())) {<a name="line.693"></a> |
| <span class="sourceLineNo">694</span> halfWriter.appendFileInfo(entry.getKey(), entry.getValue());<a name="line.694"></a> |
| <span class="sourceLineNo">695</span> }<a name="line.695"></a> |
| <span class="sourceLineNo">696</span> }<a name="line.696"></a> |
| <span class="sourceLineNo">697</span> } finally {<a name="line.697"></a> |
| <span class="sourceLineNo">698</span> if (halfReader != null) {<a name="line.698"></a> |
| <span class="sourceLineNo">699</span> try {<a name="line.699"></a> |
| <span class="sourceLineNo">700</span> halfReader.close(cacheConf.shouldEvictOnClose());<a name="line.700"></a> |
| <span class="sourceLineNo">701</span> } catch (IOException e) {<a name="line.701"></a> |
| <span class="sourceLineNo">702</span> LOG.warn("failed to close hfile reader for " + inFile, e);<a name="line.702"></a> |
| <span class="sourceLineNo">703</span> }<a name="line.703"></a> |
| <span class="sourceLineNo">704</span> }<a name="line.704"></a> |
| <span class="sourceLineNo">705</span> if (halfWriter != null) {<a name="line.705"></a> |
| <span class="sourceLineNo">706</span> halfWriter.close();<a name="line.706"></a> |
| <span class="sourceLineNo">707</span> }<a name="line.707"></a> |
| <span class="sourceLineNo">708</span> }<a name="line.708"></a> |
| <span class="sourceLineNo">709</span> }<a name="line.709"></a> |
| <span class="sourceLineNo">710</span><a name="line.710"></a> |
| <span class="sourceLineNo">711</span> /**<a name="line.711"></a> |
| <span class="sourceLineNo">712</span> * Infers region boundaries for a new table.<a name="line.712"></a> |
| <span class="sourceLineNo">713</span> * <p/><a name="line.713"></a> |
| <span class="sourceLineNo">714</span> * Parameter: <br/><a name="line.714"></a> |
| <span class="sourceLineNo">715</span> * bdryMap is a map between keys to an integer belonging to {+1, -1}<a name="line.715"></a> |
| <span class="sourceLineNo">716</span> * <ul><a name="line.716"></a> |
| <span class="sourceLineNo">717</span> * <li>If a key is a start key of a file, then it maps to +1</li><a name="line.717"></a> |
| <span class="sourceLineNo">718</span> * <li>If a key is an end key of a file, then it maps to -1</li><a name="line.718"></a> |
| <span class="sourceLineNo">719</span> * </ul><a name="line.719"></a> |
| <span class="sourceLineNo">720</span> * <p><a name="line.720"></a> |
| <span class="sourceLineNo">721</span> * Algo:<br/><a name="line.721"></a> |
| <span class="sourceLineNo">722</span> * <ol><a name="line.722"></a> |
| <span class="sourceLineNo">723</span> * <li>Poll on the keys in order:<a name="line.723"></a> |
| <span class="sourceLineNo">724</span> * <ol type="a"><a name="line.724"></a> |
| <span class="sourceLineNo">725</span> * <li>Keep adding the mapped values to these keys (runningSum)</li><a name="line.725"></a> |
| <span class="sourceLineNo">726</span> * <li>Each time runningSum reaches 0, add the start Key from when the runningSum had started to a<a name="line.726"></a> |
| <span class="sourceLineNo">727</span> * boundary list.</li><a name="line.727"></a> |
| <span class="sourceLineNo">728</span> * </ol><a name="line.728"></a> |
| <span class="sourceLineNo">729</span> * </li><a name="line.729"></a> |
| <span class="sourceLineNo">730</span> * <li>Return the boundary list.</li><a name="line.730"></a> |
| <span class="sourceLineNo">731</span> * </ol><a name="line.731"></a> |
| <span class="sourceLineNo">732</span> */<a name="line.732"></a> |
| <span class="sourceLineNo">733</span> public static byte[][] inferBoundaries(SortedMap<byte[], Integer> bdryMap) {<a name="line.733"></a> |
| <span class="sourceLineNo">734</span> List<byte[]> keysArray = new ArrayList<>();<a name="line.734"></a> |
| <span class="sourceLineNo">735</span> int runningValue = 0;<a name="line.735"></a> |
| <span class="sourceLineNo">736</span> byte[] currStartKey = null;<a name="line.736"></a> |
| <span class="sourceLineNo">737</span> boolean firstBoundary = true;<a name="line.737"></a> |
| <span class="sourceLineNo">738</span><a name="line.738"></a> |
| <span class="sourceLineNo">739</span> for (Map.Entry<byte[], Integer> item : bdryMap.entrySet()) {<a name="line.739"></a> |
| <span class="sourceLineNo">740</span> if (runningValue == 0) {<a name="line.740"></a> |
| <span class="sourceLineNo">741</span> currStartKey = item.getKey();<a name="line.741"></a> |
| <span class="sourceLineNo">742</span> }<a name="line.742"></a> |
| <span class="sourceLineNo">743</span> runningValue += item.getValue();<a name="line.743"></a> |
| <span class="sourceLineNo">744</span> if (runningValue == 0) {<a name="line.744"></a> |
| <span class="sourceLineNo">745</span> if (!firstBoundary) {<a name="line.745"></a> |
| <span class="sourceLineNo">746</span> keysArray.add(currStartKey);<a name="line.746"></a> |
| <span class="sourceLineNo">747</span> }<a name="line.747"></a> |
| <span class="sourceLineNo">748</span> firstBoundary = false;<a name="line.748"></a> |
| <span class="sourceLineNo">749</span> }<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> return keysArray.toArray(new byte[0][]);<a name="line.752"></a> |
| <span class="sourceLineNo">753</span> }<a name="line.753"></a> |
| <span class="sourceLineNo">754</span><a name="line.754"></a> |
| <span class="sourceLineNo">755</span> /**<a name="line.755"></a> |
| <span class="sourceLineNo">756</span> * If the table is created for the first time, then "completebulkload" reads the files twice. More<a name="line.756"></a> |
| <span class="sourceLineNo">757</span> * modifications necessary if we want to avoid doing it.<a name="line.757"></a> |
| <span class="sourceLineNo">758</span> */<a name="line.758"></a> |
| <span class="sourceLineNo">759</span> private void createTable(TableName tableName, Path hfofDir, AsyncAdmin admin) throws IOException {<a name="line.759"></a> |
| <span class="sourceLineNo">760</span> final FileSystem fs = hfofDir.getFileSystem(getConf());<a name="line.760"></a> |
| <span class="sourceLineNo">761</span><a name="line.761"></a> |
| <span class="sourceLineNo">762</span> // Add column families<a name="line.762"></a> |
| <span class="sourceLineNo">763</span> // Build a set of keys<a name="line.763"></a> |
| <span class="sourceLineNo">764</span> List<ColumnFamilyDescriptorBuilder> familyBuilders = new ArrayList<>();<a name="line.764"></a> |
| <span class="sourceLineNo">765</span> SortedMap<byte[], Integer> map = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.765"></a> |
| <span class="sourceLineNo">766</span> visitBulkHFiles(fs, hfofDir, new BulkHFileVisitor<ColumnFamilyDescriptorBuilder>() {<a name="line.766"></a> |
| <span class="sourceLineNo">767</span> @Override<a name="line.767"></a> |
| <span class="sourceLineNo">768</span> public ColumnFamilyDescriptorBuilder bulkFamily(byte[] familyName) {<a name="line.768"></a> |
| <span class="sourceLineNo">769</span> ColumnFamilyDescriptorBuilder builder =<a name="line.769"></a> |
| <span class="sourceLineNo">770</span> ColumnFamilyDescriptorBuilder.newBuilder(familyName);<a name="line.770"></a> |
| <span class="sourceLineNo">771</span> familyBuilders.add(builder);<a name="line.771"></a> |
| <span class="sourceLineNo">772</span> return builder;<a name="line.772"></a> |
| <span class="sourceLineNo">773</span> }<a name="line.773"></a> |
| <span class="sourceLineNo">774</span><a name="line.774"></a> |
| <span class="sourceLineNo">775</span> @Override<a name="line.775"></a> |
| <span class="sourceLineNo">776</span> public void bulkHFile(ColumnFamilyDescriptorBuilder builder, FileStatus hfileStatus)<a name="line.776"></a> |
| <span class="sourceLineNo">777</span> throws IOException {<a name="line.777"></a> |
| <span class="sourceLineNo">778</span> Path hfile = hfileStatus.getPath();<a name="line.778"></a> |
| <span class="sourceLineNo">779</span> try (HFile.Reader reader =<a name="line.779"></a> |
| <span class="sourceLineNo">780</span> HFile.createReader(fs, hfile, CacheConfig.DISABLED, true, getConf())) {<a name="line.780"></a> |
| <span class="sourceLineNo">781</span> if (builder.getCompressionType() != reader.getFileContext().getCompression()) {<a name="line.781"></a> |
| <span class="sourceLineNo">782</span> builder.setCompressionType(reader.getFileContext().getCompression());<a name="line.782"></a> |
| <span class="sourceLineNo">783</span> LOG.info("Setting compression " + reader.getFileContext().getCompression().name() +<a name="line.783"></a> |
| <span class="sourceLineNo">784</span> " for family " + builder.getNameAsString());<a name="line.784"></a> |
| <span class="sourceLineNo">785</span> }<a name="line.785"></a> |
| <span class="sourceLineNo">786</span> reader.loadFileInfo();<a name="line.786"></a> |
| <span class="sourceLineNo">787</span> byte[] first = reader.getFirstRowKey().get();<a name="line.787"></a> |
| <span class="sourceLineNo">788</span> byte[] last = reader.getLastRowKey().get();<a name="line.788"></a> |
| <span class="sourceLineNo">789</span><a name="line.789"></a> |
| <span class="sourceLineNo">790</span> LOG.info("Trying to figure out region boundaries hfile=" + hfile + " first=" +<a name="line.790"></a> |
| <span class="sourceLineNo">791</span> Bytes.toStringBinary(first) + " last=" + Bytes.toStringBinary(last));<a name="line.791"></a> |
| <span class="sourceLineNo">792</span><a name="line.792"></a> |
| <span class="sourceLineNo">793</span> // To eventually infer start key-end key boundaries<a name="line.793"></a> |
| <span class="sourceLineNo">794</span> Integer value = map.containsKey(first) ? map.get(first) : 0;<a name="line.794"></a> |
| <span class="sourceLineNo">795</span> map.put(first, value + 1);<a name="line.795"></a> |
| <span class="sourceLineNo">796</span><a name="line.796"></a> |
| <span class="sourceLineNo">797</span> value = map.containsKey(last) ? map.get(last) : 0;<a name="line.797"></a> |
| <span class="sourceLineNo">798</span> map.put(last, value - 1);<a name="line.798"></a> |
| <span class="sourceLineNo">799</span> }<a name="line.799"></a> |
| <span class="sourceLineNo">800</span> }<a name="line.800"></a> |
| <span class="sourceLineNo">801</span> }, true);<a name="line.801"></a> |
| <span class="sourceLineNo">802</span><a name="line.802"></a> |
| <span class="sourceLineNo">803</span> byte[][] keys = inferBoundaries(map);<a name="line.803"></a> |
| <span class="sourceLineNo">804</span> TableDescriptorBuilder tdBuilder = TableDescriptorBuilder.newBuilder(tableName);<a name="line.804"></a> |
| <span class="sourceLineNo">805</span> familyBuilders.stream().map(ColumnFamilyDescriptorBuilder::build)<a name="line.805"></a> |
| <span class="sourceLineNo">806</span> .forEachOrdered(tdBuilder::setColumnFamily);<a name="line.806"></a> |
| <span class="sourceLineNo">807</span> FutureUtils.get(admin.createTable(tdBuilder.build(), keys));<a name="line.807"></a> |
| <span class="sourceLineNo">808</span><a name="line.808"></a> |
| <span class="sourceLineNo">809</span> LOG.info("Table " + tableName + " is available!!");<a name="line.809"></a> |
| <span class="sourceLineNo">810</span> }<a name="line.810"></a> |
| <span class="sourceLineNo">811</span><a name="line.811"></a> |
| <span class="sourceLineNo">812</span> private Map<LoadQueueItem, ByteBuffer> performBulkLoad(AsyncClusterConnection conn,<a name="line.812"></a> |
| <span class="sourceLineNo">813</span> TableName tableName, Deque<LoadQueueItem> queue, ExecutorService pool, boolean copyFile)<a name="line.813"></a> |
| <span class="sourceLineNo">814</span> throws IOException {<a name="line.814"></a> |
| <span class="sourceLineNo">815</span> int count = 0;<a name="line.815"></a> |
| <span class="sourceLineNo">816</span><a name="line.816"></a> |
| <span class="sourceLineNo">817</span> fsDelegationToken.acquireDelegationToken(queue.peek().getFilePath().getFileSystem(getConf()));<a name="line.817"></a> |
| <span class="sourceLineNo">818</span> bulkToken = FutureUtils.get(conn.prepareBulkLoad(tableName));<a name="line.818"></a> |
| <span class="sourceLineNo">819</span> Pair<Multimap<ByteBuffer, LoadQueueItem>, Set<String>> pair = null;<a name="line.819"></a> |
| <span class="sourceLineNo">820</span><a name="line.820"></a> |
| <span class="sourceLineNo">821</span> Map<LoadQueueItem, ByteBuffer> item2RegionMap = new HashMap<>();<a name="line.821"></a> |
| <span class="sourceLineNo">822</span> // Assumes that region splits can happen while this occurs.<a name="line.822"></a> |
| <span class="sourceLineNo">823</span> while (!queue.isEmpty()) {<a name="line.823"></a> |
| <span class="sourceLineNo">824</span> // need to reload split keys each iteration.<a name="line.824"></a> |
| <span class="sourceLineNo">825</span> final List<Pair<byte[], byte[]>> startEndKeys =<a name="line.825"></a> |
| <span class="sourceLineNo">826</span> FutureUtils.get(conn.getRegionLocator(tableName).getStartEndKeys());<a name="line.826"></a> |
| <span class="sourceLineNo">827</span> if (count != 0) {<a name="line.827"></a> |
| <span class="sourceLineNo">828</span> LOG.info("Split occurred while grouping HFiles, retry attempt " + count + " with " +<a name="line.828"></a> |
| <span class="sourceLineNo">829</span> queue.size() + " files remaining to group or split");<a name="line.829"></a> |
| <span class="sourceLineNo">830</span> }<a name="line.830"></a> |
| <span class="sourceLineNo">831</span><a name="line.831"></a> |
| <span class="sourceLineNo">832</span> int maxRetries = getConf().getInt(HConstants.BULKLOAD_MAX_RETRIES_NUMBER, 10);<a name="line.832"></a> |
| <span class="sourceLineNo">833</span> maxRetries = Math.max(maxRetries, startEndKeys.size() + 1);<a name="line.833"></a> |
| <span class="sourceLineNo">834</span> if (maxRetries != 0 && count >= maxRetries) {<a name="line.834"></a> |
| <span class="sourceLineNo">835</span> throw new IOException(<a name="line.835"></a> |
| <span class="sourceLineNo">836</span> "Retry attempted " + count + " times without completing, bailing out");<a name="line.836"></a> |
| <span class="sourceLineNo">837</span> }<a name="line.837"></a> |
| <span class="sourceLineNo">838</span> count++;<a name="line.838"></a> |
| <span class="sourceLineNo">839</span><a name="line.839"></a> |
| <span class="sourceLineNo">840</span> // Using ByteBuffer for byte[] equality semantics<a name="line.840"></a> |
| <span class="sourceLineNo">841</span> pair = groupOrSplitPhase(conn, tableName, pool, queue, startEndKeys);<a name="line.841"></a> |
| <span class="sourceLineNo">842</span> Multimap<ByteBuffer, LoadQueueItem> regionGroups = pair.getFirst();<a name="line.842"></a> |
| <span class="sourceLineNo">843</span><a name="line.843"></a> |
| <span class="sourceLineNo">844</span> if (!checkHFilesCountPerRegionPerFamily(regionGroups)) {<a name="line.844"></a> |
| <span class="sourceLineNo">845</span> // Error is logged inside checkHFilesCountPerRegionPerFamily.<a name="line.845"></a> |
| <span class="sourceLineNo">846</span> throw new IOException("Trying to load more than " + maxFilesPerRegionPerFamily +<a name="line.846"></a> |
| <span class="sourceLineNo">847</span> " hfiles to one family of one region");<a name="line.847"></a> |
| <span class="sourceLineNo">848</span> }<a name="line.848"></a> |
| <span class="sourceLineNo">849</span><a name="line.849"></a> |
| <span class="sourceLineNo">850</span> bulkLoadPhase(conn, tableName, queue, regionGroups, copyFile, item2RegionMap);<a name="line.850"></a> |
| <span class="sourceLineNo">851</span><a name="line.851"></a> |
| <span class="sourceLineNo">852</span> // NOTE: The next iteration's split / group could happen in parallel to<a name="line.852"></a> |
| <span class="sourceLineNo">853</span> // atomic bulkloads assuming that there are splits and no merges, and<a name="line.853"></a> |
| <span class="sourceLineNo">854</span> // that we can atomically pull out the groups we want to retry.<a name="line.854"></a> |
| <span class="sourceLineNo">855</span> }<a name="line.855"></a> |
| <span class="sourceLineNo">856</span><a name="line.856"></a> |
| <span class="sourceLineNo">857</span> if (!queue.isEmpty()) {<a name="line.857"></a> |
| <span class="sourceLineNo">858</span> throw new RuntimeException(<a name="line.858"></a> |
| <span class="sourceLineNo">859</span> "Bulk load aborted with some files not yet loaded." + "Please check log for more details.");<a name="line.859"></a> |
| <span class="sourceLineNo">860</span> }<a name="line.860"></a> |
| <span class="sourceLineNo">861</span> return item2RegionMap;<a name="line.861"></a> |
| <span class="sourceLineNo">862</span> }<a name="line.862"></a> |
| <span class="sourceLineNo">863</span><a name="line.863"></a> |
| <span class="sourceLineNo">864</span> private void cleanup(AsyncClusterConnection conn, TableName tableName, Deque<LoadQueueItem> queue,<a name="line.864"></a> |
| <span class="sourceLineNo">865</span> ExecutorService pool) throws IOException {<a name="line.865"></a> |
| <span class="sourceLineNo">866</span> fsDelegationToken.releaseDelegationToken();<a name="line.866"></a> |
| <span class="sourceLineNo">867</span> if (bulkToken != null) {<a name="line.867"></a> |
| <span class="sourceLineNo">868</span> conn.cleanupBulkLoad(tableName, bulkToken);<a name="line.868"></a> |
| <span class="sourceLineNo">869</span> }<a name="line.869"></a> |
| <span class="sourceLineNo">870</span> if (pool != null) {<a name="line.870"></a> |
| <span class="sourceLineNo">871</span> pool.shutdown();<a name="line.871"></a> |
| <span class="sourceLineNo">872</span> }<a name="line.872"></a> |
| <span class="sourceLineNo">873</span> if (!queue.isEmpty()) {<a name="line.873"></a> |
| <span class="sourceLineNo">874</span> StringBuilder err = new StringBuilder();<a name="line.874"></a> |
| <span class="sourceLineNo">875</span> err.append("-------------------------------------------------\n");<a name="line.875"></a> |
| <span class="sourceLineNo">876</span> err.append("Bulk load aborted with some files not yet loaded:\n");<a name="line.876"></a> |
| <span class="sourceLineNo">877</span> err.append("-------------------------------------------------\n");<a name="line.877"></a> |
| <span class="sourceLineNo">878</span> for (LoadQueueItem q : queue) {<a name="line.878"></a> |
| <span class="sourceLineNo">879</span> err.append(" ").append(q.getFilePath()).append('\n');<a name="line.879"></a> |
| <span class="sourceLineNo">880</span> }<a name="line.880"></a> |
| <span class="sourceLineNo">881</span> LOG.error(err.toString());<a name="line.881"></a> |
| <span class="sourceLineNo">882</span> }<a name="line.882"></a> |
| <span class="sourceLineNo">883</span> }<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> * Perform a bulk load of the given directory into the given pre-existing table. This method is<a name="line.886"></a> |
| <span class="sourceLineNo">887</span> * not threadsafe.<a name="line.887"></a> |
| <span class="sourceLineNo">888</span> * @param map map of family to List of hfiles<a name="line.888"></a> |
| <span class="sourceLineNo">889</span> * @param tableName table to load the hfiles<a name="line.889"></a> |
| <span class="sourceLineNo">890</span> * @param silence true to ignore unmatched column families<a name="line.890"></a> |
| <span class="sourceLineNo">891</span> * @param copyFile always copy hfiles if true<a name="line.891"></a> |
| <span class="sourceLineNo">892</span> * @throws TableNotFoundException if table does not yet exist<a name="line.892"></a> |
| <span class="sourceLineNo">893</span> */<a name="line.893"></a> |
| <span class="sourceLineNo">894</span> private Map<LoadQueueItem, ByteBuffer> doBulkLoad(AsyncClusterConnection conn,<a name="line.894"></a> |
| <span class="sourceLineNo">895</span> TableName tableName, Map<byte[], List<Path>> map, boolean silence, boolean copyFile)<a name="line.895"></a> |
| <span class="sourceLineNo">896</span> throws TableNotFoundException, IOException {<a name="line.896"></a> |
| <span class="sourceLineNo">897</span> if (!FutureUtils.get(conn.getAdmin().isTableAvailable(tableName))) {<a name="line.897"></a> |
| <span class="sourceLineNo">898</span> throw new TableNotFoundException("Table " + tableName + " is not currently available.");<a name="line.898"></a> |
| <span class="sourceLineNo">899</span> }<a name="line.899"></a> |
| <span class="sourceLineNo">900</span> // LQI queue does not need to be threadsafe -- all operations on this queue<a name="line.900"></a> |
| <span class="sourceLineNo">901</span> // happen in this thread<a name="line.901"></a> |
| <span class="sourceLineNo">902</span> Deque<LoadQueueItem> queue = new ArrayDeque<>();<a name="line.902"></a> |
| <span class="sourceLineNo">903</span> ExecutorService pool = null;<a name="line.903"></a> |
| <span class="sourceLineNo">904</span> try {<a name="line.904"></a> |
| <span class="sourceLineNo">905</span> prepareHFileQueue(conn, tableName, map, queue, silence);<a name="line.905"></a> |
| <span class="sourceLineNo">906</span> if (queue.isEmpty()) {<a name="line.906"></a> |
| <span class="sourceLineNo">907</span> LOG.warn("Bulk load operation did not get any files to load");<a name="line.907"></a> |
| <span class="sourceLineNo">908</span> return Collections.emptyMap();<a name="line.908"></a> |
| <span class="sourceLineNo">909</span> }<a name="line.909"></a> |
| <span class="sourceLineNo">910</span> pool = createExecutorService();<a name="line.910"></a> |
| <span class="sourceLineNo">911</span> return performBulkLoad(conn, tableName, queue, pool, copyFile);<a name="line.911"></a> |
| <span class="sourceLineNo">912</span> } finally {<a name="line.912"></a> |
| <span class="sourceLineNo">913</span> cleanup(conn, tableName, queue, pool);<a name="line.913"></a> |
| <span class="sourceLineNo">914</span> }<a name="line.914"></a> |
| <span class="sourceLineNo">915</span> }<a name="line.915"></a> |
| <span class="sourceLineNo">916</span><a name="line.916"></a> |
| <span class="sourceLineNo">917</span> /**<a name="line.917"></a> |
| <span class="sourceLineNo">918</span> * Perform a bulk load of the given directory into the given pre-existing table. This method is<a name="line.918"></a> |
| <span class="sourceLineNo">919</span> * not threadsafe.<a name="line.919"></a> |
| <span class="sourceLineNo">920</span> * @param tableName table to load the hfiles<a name="line.920"></a> |
| <span class="sourceLineNo">921</span> * @param hfofDir the directory that was provided as the output path of a job using<a name="line.921"></a> |
| <span class="sourceLineNo">922</span> * HFileOutputFormat<a name="line.922"></a> |
| <span class="sourceLineNo">923</span> * @param silence true to ignore unmatched column families<a name="line.923"></a> |
| <span class="sourceLineNo">924</span> * @param copyFile always copy hfiles if true<a name="line.924"></a> |
| <span class="sourceLineNo">925</span> * @throws TableNotFoundException if table does not yet exist<a name="line.925"></a> |
| <span class="sourceLineNo">926</span> */<a name="line.926"></a> |
| <span class="sourceLineNo">927</span> private Map<LoadQueueItem, ByteBuffer> doBulkLoad(AsyncClusterConnection conn,<a name="line.927"></a> |
| <span class="sourceLineNo">928</span> TableName tableName, Path hfofDir, boolean silence, boolean copyFile)<a name="line.928"></a> |
| <span class="sourceLineNo">929</span> throws TableNotFoundException, IOException {<a name="line.929"></a> |
| <span class="sourceLineNo">930</span> if (!FutureUtils.get(conn.getAdmin().isTableAvailable(tableName))) {<a name="line.930"></a> |
| <span class="sourceLineNo">931</span> throw new TableNotFoundException("Table " + tableName + " is not currently available.");<a name="line.931"></a> |
| <span class="sourceLineNo">932</span> }<a name="line.932"></a> |
| <span class="sourceLineNo">933</span><a name="line.933"></a> |
| <span class="sourceLineNo">934</span> /*<a name="line.934"></a> |
| <span class="sourceLineNo">935</span> * Checking hfile format is a time-consuming operation, we should have an option to skip this<a name="line.935"></a> |
| <span class="sourceLineNo">936</span> * step when bulkloading millions of HFiles. See HBASE-13985.<a name="line.936"></a> |
| <span class="sourceLineNo">937</span> */<a name="line.937"></a> |
| <span class="sourceLineNo">938</span> boolean validateHFile = getConf().getBoolean("hbase.loadincremental.validate.hfile", true);<a name="line.938"></a> |
| <span class="sourceLineNo">939</span> if (!validateHFile) {<a name="line.939"></a> |
| <span class="sourceLineNo">940</span> LOG.warn("You are skipping HFiles validation, it might cause some data loss if files " +<a name="line.940"></a> |
| <span class="sourceLineNo">941</span> "are not correct. If you fail to read data from your table after using this " +<a name="line.941"></a> |
| <span class="sourceLineNo">942</span> "option, consider removing the files and bulkload again without this option. " +<a name="line.942"></a> |
| <span class="sourceLineNo">943</span> "See HBASE-13985");<a name="line.943"></a> |
| <span class="sourceLineNo">944</span> }<a name="line.944"></a> |
| <span class="sourceLineNo">945</span> // LQI queue does not need to be threadsafe -- all operations on this queue<a name="line.945"></a> |
| <span class="sourceLineNo">946</span> // happen in this thread<a name="line.946"></a> |
| <span class="sourceLineNo">947</span> Deque<LoadQueueItem> queue = new ArrayDeque<>();<a name="line.947"></a> |
| <span class="sourceLineNo">948</span> ExecutorService pool = null;<a name="line.948"></a> |
| <span class="sourceLineNo">949</span> try {<a name="line.949"></a> |
| <span class="sourceLineNo">950</span> prepareHFileQueue(getConf(), conn, tableName, hfofDir, queue, validateHFile, silence);<a name="line.950"></a> |
| <span class="sourceLineNo">951</span><a name="line.951"></a> |
| <span class="sourceLineNo">952</span> if (queue.isEmpty()) {<a name="line.952"></a> |
| <span class="sourceLineNo">953</span> LOG.warn(<a name="line.953"></a> |
| <span class="sourceLineNo">954</span> "Bulk load operation did not find any files to load in directory {}. " +<a name="line.954"></a> |
| <span class="sourceLineNo">955</span> "Does it contain files in subdirectories that correspond to column family names?",<a name="line.955"></a> |
| <span class="sourceLineNo">956</span> (hfofDir != null ? hfofDir.toUri().toString() : ""));<a name="line.956"></a> |
| <span class="sourceLineNo">957</span> return Collections.emptyMap();<a name="line.957"></a> |
| <span class="sourceLineNo">958</span> }<a name="line.958"></a> |
| <span class="sourceLineNo">959</span> pool = createExecutorService();<a name="line.959"></a> |
| <span class="sourceLineNo">960</span> return performBulkLoad(conn, tableName, queue, pool, copyFile);<a name="line.960"></a> |
| <span class="sourceLineNo">961</span> } finally {<a name="line.961"></a> |
| <span class="sourceLineNo">962</span> cleanup(conn, tableName, queue, pool);<a name="line.962"></a> |
| <span class="sourceLineNo">963</span> }<a name="line.963"></a> |
| <span class="sourceLineNo">964</span> }<a name="line.964"></a> |
| <span class="sourceLineNo">965</span><a name="line.965"></a> |
| <span class="sourceLineNo">966</span> @Override<a name="line.966"></a> |
| <span class="sourceLineNo">967</span> public Map<LoadQueueItem, ByteBuffer> bulkLoad(TableName tableName,<a name="line.967"></a> |
| <span class="sourceLineNo">968</span> Map<byte[], List<Path>> family2Files) throws TableNotFoundException, IOException {<a name="line.968"></a> |
| <span class="sourceLineNo">969</span> try (AsyncClusterConnection conn = ClusterConnectionFactory<a name="line.969"></a> |
| <span class="sourceLineNo">970</span> .createAsyncClusterConnection(getConf(), null, userProvider.getCurrent())) {<a name="line.970"></a> |
| <span class="sourceLineNo">971</span> if (!FutureUtils.get(conn.getAdmin().tableExists(tableName))) {<a name="line.971"></a> |
| <span class="sourceLineNo">972</span> String errorMsg = format("Table '%s' does not exist.", tableName);<a name="line.972"></a> |
| <span class="sourceLineNo">973</span> LOG.error(errorMsg);<a name="line.973"></a> |
| <span class="sourceLineNo">974</span> throw new TableNotFoundException(errorMsg);<a name="line.974"></a> |
| <span class="sourceLineNo">975</span> }<a name="line.975"></a> |
| <span class="sourceLineNo">976</span> return doBulkLoad(conn, tableName, family2Files, isSilence(), isAlwaysCopyFiles());<a name="line.976"></a> |
| <span class="sourceLineNo">977</span> }<a name="line.977"></a> |
| <span class="sourceLineNo">978</span> }<a name="line.978"></a> |
| <span class="sourceLineNo">979</span><a name="line.979"></a> |
| <span class="sourceLineNo">980</span> @Override<a name="line.980"></a> |
| <span class="sourceLineNo">981</span> public Map<LoadQueueItem, ByteBuffer> bulkLoad(TableName tableName, Path dir)<a name="line.981"></a> |
| <span class="sourceLineNo">982</span> throws TableNotFoundException, IOException {<a name="line.982"></a> |
| <span class="sourceLineNo">983</span> try (AsyncClusterConnection conn = ClusterConnectionFactory<a name="line.983"></a> |
| <span class="sourceLineNo">984</span> .createAsyncClusterConnection(getConf(), null, userProvider.getCurrent())) {<a name="line.984"></a> |
| <span class="sourceLineNo">985</span> AsyncAdmin admin = conn.getAdmin();<a name="line.985"></a> |
| <span class="sourceLineNo">986</span> if (!FutureUtils.get(admin.tableExists(tableName))) {<a name="line.986"></a> |
| <span class="sourceLineNo">987</span> if (isCreateTable()) {<a name="line.987"></a> |
| <span class="sourceLineNo">988</span> createTable(tableName, dir, admin);<a name="line.988"></a> |
| <span class="sourceLineNo">989</span> } else {<a name="line.989"></a> |
| <span class="sourceLineNo">990</span> String errorMsg = format("Table '%s' does not exist.", tableName);<a name="line.990"></a> |
| <span class="sourceLineNo">991</span> LOG.error(errorMsg);<a name="line.991"></a> |
| <span class="sourceLineNo">992</span> throw new TableNotFoundException(errorMsg);<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> return doBulkLoad(conn, tableName, dir, isSilence(), isAlwaysCopyFiles());<a name="line.995"></a> |
| <span class="sourceLineNo">996</span> }<a name="line.996"></a> |
| <span class="sourceLineNo">997</span> }<a name="line.997"></a> |
| <span class="sourceLineNo">998</span><a name="line.998"></a> |
| <span class="sourceLineNo">999</span> public void setBulkToken(String bulkToken) {<a name="line.999"></a> |
| <span class="sourceLineNo">1000</span> this.bulkToken = bulkToken;<a name="line.1000"></a> |
| <span class="sourceLineNo">1001</span> }<a name="line.1001"></a> |
| <span class="sourceLineNo">1002</span><a name="line.1002"></a> |
| <span class="sourceLineNo">1003</span> private void usage() {<a name="line.1003"></a> |
| <span class="sourceLineNo">1004</span> System.err.println("usage: " + "bin/hbase completebulkload <-Dargs> "<a name="line.1004"></a> |
| <span class="sourceLineNo">1005</span> + "</path/to/hfileoutputformat-output> <tablename>\n"<a name="line.1005"></a> |
| <span class="sourceLineNo">1006</span> + "\t-D" + CREATE_TABLE_CONF_KEY + "=no can be used to avoid creation "<a name="line.1006"></a> |
| <span class="sourceLineNo">1007</span> + "of a table by this tool.\n"<a name="line.1007"></a> |
| <span class="sourceLineNo">1008</span> + "\t Note: if you set this to 'no', then target table must already exist.\n"<a name="line.1008"></a> |
| <span class="sourceLineNo">1009</span> + "\t-D" + IGNORE_UNMATCHED_CF_CONF_KEY + "=yes can be used to ignore "<a name="line.1009"></a> |
| <span class="sourceLineNo">1010</span> + "unmatched column families.\n"<a name="line.1010"></a> |
| <span class="sourceLineNo">1011</span> + "\t-loadTable switch implies your baseDirectory to store file has a "<a name="line.1011"></a> |
| <span class="sourceLineNo">1012</span> + "depth of 3, table must exist\n"<a name="line.1012"></a> |
| <span class="sourceLineNo">1013</span> + "\t and -loadTable switch is the last option on the command line.\n\n");<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 int run(String[] args) throws Exception {<a name="line.1017"></a> |
| <span class="sourceLineNo">1018</span> if (args.length != 2 && args.length != 3) {<a name="line.1018"></a> |
| <span class="sourceLineNo">1019</span> usage();<a name="line.1019"></a> |
| <span class="sourceLineNo">1020</span> return -1;<a name="line.1020"></a> |
| <span class="sourceLineNo">1021</span> }<a name="line.1021"></a> |
| <span class="sourceLineNo">1022</span> Path dirPath = new Path(args[0]);<a name="line.1022"></a> |
| <span class="sourceLineNo">1023</span> TableName tableName = TableName.valueOf(args[1]);<a name="line.1023"></a> |
| <span class="sourceLineNo">1024</span><a name="line.1024"></a> |
| <span class="sourceLineNo">1025</span> if (args.length == 2) {<a name="line.1025"></a> |
| <span class="sourceLineNo">1026</span> return !bulkLoad(tableName, dirPath).isEmpty() ? 0 : -1;<a name="line.1026"></a> |
| <span class="sourceLineNo">1027</span> } else {<a name="line.1027"></a> |
| <span class="sourceLineNo">1028</span> Map<byte[], List<Path>> family2Files = Maps.newHashMap();<a name="line.1028"></a> |
| <span class="sourceLineNo">1029</span> FileSystem fs = FileSystem.get(getConf());<a name="line.1029"></a> |
| <span class="sourceLineNo">1030</span> for (FileStatus regionDir : fs.listStatus(dirPath)) {<a name="line.1030"></a> |
| <span class="sourceLineNo">1031</span> FSVisitor.visitRegionStoreFiles(fs, regionDir.getPath(), (region, family, hfileName) -> {<a name="line.1031"></a> |
| <span class="sourceLineNo">1032</span> Path path = new Path(regionDir.getPath(), new Path(family, hfileName));<a name="line.1032"></a> |
| <span class="sourceLineNo">1033</span> byte[] familyName = Bytes.toBytes(family);<a name="line.1033"></a> |
| <span class="sourceLineNo">1034</span> if (family2Files.containsKey(familyName)) {<a name="line.1034"></a> |
| <span class="sourceLineNo">1035</span> family2Files.get(familyName).add(path);<a name="line.1035"></a> |
| <span class="sourceLineNo">1036</span> } else {<a name="line.1036"></a> |
| <span class="sourceLineNo">1037</span> family2Files.put(familyName, Lists.newArrayList(path));<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> }<a name="line.1040"></a> |
| <span class="sourceLineNo">1041</span> return !bulkLoad(tableName, family2Files).isEmpty() ? 0 : -1;<a name="line.1041"></a> |
| <span class="sourceLineNo">1042</span> }<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> public static void main(String[] args) throws Exception {<a name="line.1045"></a> |
| <span class="sourceLineNo">1046</span> Configuration conf = HBaseConfiguration.create();<a name="line.1046"></a> |
| <span class="sourceLineNo">1047</span> int ret = ToolRunner.run(conf, new BulkLoadHFilesTool(conf), args);<a name="line.1047"></a> |
| <span class="sourceLineNo">1048</span> System.exit(ret);<a name="line.1048"></a> |
| <span class="sourceLineNo">1049</span> }<a name="line.1049"></a> |
| <span class="sourceLineNo">1050</span><a name="line.1050"></a> |
| <span class="sourceLineNo">1051</span>}<a name="line.1051"></a> |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| </pre> |
| </div> |
| </body> |
| </html> |